Expression of the aspartate/glutamate mitochondrial carriers aralar1 and citrin during development and in adult rat tissues Araceli del Arco 1,3 , Julia ´ n Morcillo 2 , Juan Ramon Martı ´ nez-Morales 2 , Carmen Galia ´ n 1 , Vera Martos 1 , Paola Bovolenta 2 and Jorgina Satru ´ stegui 1 1 Departamento de Biologı ´ a Molecular, Centro de Biologı ´ a Molecular Severo Ochoa, Universidad Auto ´ noma de Madrid Spain; 2 Departamento de Neurobiologı ´ a del Desarrollo, Instituto Cajal, Consejo Superior de Investigaciones Cientı ´ ficas, Madrid, Spain; 3 Facultad de Ciencias del Medio Ambiente, Universidad de Castilla La Mancha, Toledo, Spain Aralar1 and citrin are members of the subfamily of calcium- binding mitochondrial carriers and correspond to two iso- forms of the mitochondrial aspartate/glutamate carrier (AGC). These proteins are activated by Ca 2+ acting on the external side of the inner mitochondrial membrane. Although it is known that aralar1 is expressed mainly in skeletal muscle, heart and brain, whereas citrin is present in liver, kidney and heart, the precise tissue distribution of the two proteins in embryonic and adult tissues is largely unknown. We investigated the pattern of expression of aralar1 and citrin in murine embryonic and adult tissues at the mRNA and protein levels. Insituhybridization analysis indicates that both isoforms are expressed strongly in the branchial arches, dermomyotome, limb and tail buds at early embryonic stages. However, citrin was more abundant in the ectodermal components of these structures whereas aralarl had a predominantly mesenchymal localization. The strong expression of citrin in the liver was acquired postnatally, whereas the characteristic expression of aralar1 in skeletal muscle was detected at E18 and that in the heart began early in development (E11) and was preferentially localized to auricular myocardium in late embryonic stages. Aralar1 was also expressed in bone marrow, T-lymphocytes and macrophages, including Kupffer cells in the liver, indicating that this is the major AGC isoform present in the hemato- poietic system. Both aralar1 and citrin were expressed in fetal gut and adult stomach, ovary, testis, and pancreas, but only aralar1 is enriched in lung and insulin-secreting b cells. These results show that aralar1 is expressed in many more tissues than originally believed and is absent from hepatocytes, where citrin is the only AGC isoform present. This explains why citrin deficiency in humans (type II citrullinemia) only affects the liver and suggests that aralar1 may compensate for the lack of citrin in other tissues. Keywords: aspartate/glutamate carrier; calcium; citrulline- mia; development; mitochondria. Metabolites are transported through the inner mitochondrial membrane by proteins belonging to the mitochondrialcarrier (MC) superfamily [1]. The structure of these carriers (molecular mass  30 kDa) consists of a threefold repetition of a sequence of about 100aminoacids [2,3] with two putative transmembrane domains. In the last few years, a number of new MCs have been identified [3–5], including a subfamily of Ca 2+ -binding mitochondrial carriers (CaMCs) with new structural characteristics [6–8]. The CaMC subfamily mem- bers have a bipartite structure. Their C-terminal domains have the features of the MC superfamily and their N-terminal extensions harbor EF-hand Ca 2+ -binding motifs [6]. Aralar1 and citrin, two members of the CaMC subfamily, are nuclear-encoded proteins, with genes in human chromosome 2 (SLC25A12 [9,10]) and 7 (SLC25A13 [8,11]), respectively. As recently demonstrated, aralar1 and citrin are isoforms of the mitochondrial aspartate/glutamate carrier (AGC) [12] which catalyzes a 1 : 1 exchange of aspartate for glutamate and plays an important role in the malate/aspartate shuttle, urea synthesis and gluconeogenesis from lactate [13–15]. These two AGC isoforms are activated by Ca 2+ on the external face of the inner mitochondrial membrane [12]. Mutations in the human gene coding for citrin are responsible for adult-onset type II citrullinemia (CTLN2: 603471) [8,16], an autosomal recessive disease caused by a liver-specific deficiency in argininosuccinate synthetase (ASS). In the liver, the AGC plays an important role in the urea cycle by providing aspartate for incorporation into argininosuccinate [17]. The mutations in the citrin gene in patients affected by CTLN2 cause either truncation of the protein or deletion of a loop between the transmembrane spans [8,16], impairing the function of citrin as an AGC in mitochondria. This impairment would presumably lead to a failure in the supply of aspartate from mitochondria for argininosuccinate synthesis, with consequent alterations in the stability/activity of liver ASS, one of the symptoms of CTLN2. Citrin is strongly expressed in both liver and kidney [6–8,18]. However, CTLN2 is a liver-specific metabolic Correspondence to J. Satru´ stegui, Departamento de Biologı ´ a Molecular, Centro de Biologı ´ a Molecular Severo Ochoa, Universidad Auto ´ noma de Madrid, 28049-Madrid, Spain. Fax: 00 34 91 3974799, Tel.: 00 34 91 3974872, E-mail: jsatrustegui@cbm.uam.es Abbreviations: AGC, aspartate/glutamate carrier; MC, mitochondrial carrier; CaMC, calcium-binding mitochondrial carrier; CTLN2, adult-onset type II citrullinemia; ASS, argininosuccinate synthetase. (Received 14 February 2002, revised 19 April 2002, accepted 23 May 2002) Eur. J. Biochem. 269, 3313–3320 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03018.x deficiency, and ASS levels are normal in other tissues such as kidney [8,19]. This difference can be explained by the observation that aralar1, the second human AGC isoform, is also expressed in kidney [6,18] and human kidney cell lines [12,18], therefore it may compensate for the loss of citrin in the kidney of patients with CTLN2. This raises a general question of whether the two isoforms are expressed in the same tissues and cell types and whether these isoforms play thesamerole. To address the first of these questions, we have now studied the expression of aralar1 and citrin throughout mouse development and in tissues of the adult rat with the use of isoform-specific probes and antibodies. Our results indicate that the two isoforms are widely expressed throughout embryogenesis with a dynamic expression pattern. The characteristic expression of aralar1 in skeletal muscle and citrin in liver is only manifested at E18 or after birth, respectively. Aralar1, but not citrin, is expressed early (E11) in heart and it is preferentially localized to auricular myocardium in late embryonic stages. Whereas citrin is preferentially expressed in liver and kidney, the classical gluconeogenic organs, a number of adult tissues and cell types were found to express aralar1 preferentially over citrin, including the adult lung and hematopoietic cells. MATERIALS AND METHODS Animals and tissues mRNA expression was examined in Balb/c mice and Wistar rats. Animals were kept in climate-controlled quarters under a 12-h light cycle with free access to water and standard chow diet. The animal facilities fulfilled the requirements of the European laws, and the highest standards of animal care were met in all experimental protocols. Mouse embryos were collected from timed pregnant Balb/c mice. The day of vaginal plug appearance was considered embryonic day 0.5 (E0.5). The following tissues and organs were dissected from 3-month-old rats: liver, forebrain, cerebellum, heart, small intestine, stomach, lung, kidney, testis, ovary, white adipose tissue, pancreas, bone marrow, spleen and muscle. Bone marrow was obtained from the tibia bone. Rat pancreatic b islets were isolated by collagenase digestion and standard procedures [20]. Brown adipose tissue was collected from 1-day-old pups. Rat fetuses staged at embryonic day 18 (E18) delivered by cesarian section and newborn pups (1–6 h after sponta- neous delivery) were used to study postnatal development of the liver. Cell lines HEK-293T cells were cultured in Dulbecco’s modified Eagle’s medium containing 5% fetal bovine serum (Gibco- BRL) at 37 °Cina7%CO 2 atmosphere. RAW 264 and Jurkat cells were grown in RPMI 1640 medium with 5% fetal bovine serum under identical conditions. Probes The mouse aralar1 probe used was a 381-bp PstIfragment obtained from the mouse EST clone W82002 (ATCC). A probe specific for mouse citrin was generated by RT-PCR using 2 lg total RNA obtained from adult mouse liver as template. The oligonucleotides used, ara2-rat5 (5¢-AT CTGTCCTGTGTGCTCCGG-3¢) and ara2-mouse3 (5¢-TCCATGGGTGTAACCTGACC-3¢), were designed from mouse citrin cDNA sequence [11]. The amplified fragment was subcloned into the blunted pSTBlue-1 (Novagen) and verified by sequencing. In situ hybridization The 381-bp and 557-bp fragments of aralar1 and citrin cDNA were transcribed to generate digoxigenin-labeled antisense and sense cRNA probes. Whole-mount in situ hybridizations were performed as described [21]. Briefly, hybridizations were carried out at 65 °Cin50%formamide. Post-hybridization washes were performed at the same temperature and in the same buffer. Embryos staged at embryonic day 11 (E11) were hybridized in toto.After hybridization, embryos were photographed, dehydrated, embedded in Paraplast, and sectioned with a microtome at 18 lm. For older embryos, E18.5, hybridizations were carried out on tissue sections. Embryos were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.3, at 4 °C overnight and then cryoprotected by immersion in 30% sucrose solution in phosphate buffer. Cryostat sections 16–20 lm thick were mounted on 2% 3-aminopropyltri- ethoxysilane-coated slides, air-dried, and permeabilized with proteinase K (10 lgÆmL )1 in NaCl/P i containing 0.1% Tween) for 5–10 min at room temperature. Sections were then postfixed in 4% paraformaldehyde in phosphate buffer, prehybridized for 1 h at 65 °C in 50% formamide, and incubated with probes for 16 h at 65 °C. All the staining patterns described below were obtained only with antisense riboprobes and not with control sense riboprobes. RNA analysis Total RNA was extracted from rat tissues using the guanidine isothiocyanate method. Northern blot analysis was carried out using 20 lg total RNA from different rat tissues as previously described [7]. As human and rat nucleotide sequences are highly homologous ( 90% identity), we used fragments of human citrin and aralar1 cDNAsasprobes. 1 The blot was stripped on 0.1% SDS at 100 °C for 30 min, and reprobed under identical conditions. Antibodies An antibody to the N-terminal half of aralar1 (amino acids 12–343) was described previously [6]. A citrin-specific antibody was generated to amino acids 9–278 of the N-terminal half of citrin expressed in bacteria. The construct for bacterial expression has been previously described [7]. In addition, selected regions of human citrin (amino acids 305– 319) and human aralar1 (amino acids 507–520), with Jameson and Wolf antigenic indexes of  1.7, as predicted by the peptidestructure program from the CGC (Genetic Computer Group, Madison, WI, USA) package, were used to generate epitope-specific antibodies. These regions of human aralar1 and citrin are conserved in the mouse proteins. The citrin 305–319 peptide was conjugated with mcKLH (mariculture keyhole limpet hemocyanin) using an Imject Immunogen EDC conjugation kit (Pierce). The 3314 A. del Arco et al.(Eur. J. Biochem. 269) Ó FEBS 2002 aralar1 507–520 peptide was conjugated with maleimide- activated mcKLH (Pierce) through a cysteine added at the N-terminus of the peptide, as recommended by the supplier. The purified citrin protein (amino acids 9–278) and mcKLH-conjugated peptides were injected into rabbits using standard immunization procedures. Westerns blots Rat tissues were homogenized in 250 m M sucrose/10 m M Tris/HCl (pH 7.4)/protease inhibitors (1 m M iodoacetamide and 1 m M phenylmethanesulfonyl fluoride) and centrifuged at 750 g (10 min). The supernatant was then centrifuged at 10 000 g (15 min), and the pellets were collected to obtain the crude mitochondrial fractions. Cells were scraped into 250 m M sucrose/20 m M Hepes/10 m M KCl/1.5 m M MgCl 2 / 1m M EDTA/1 m M EGTA/1 m M dithiothreitol/protease inhibitors, pH 7.4, homogenized and subjected to differen- tial centrifugations as described above. Mitochondrial fractions were analysed by Western blot- ting using an Enhanced Chemiluminiscence (ECL) kit (Amersham). Antibody to the N-terminus of aralar1 was used at a dilution of 1 : 5000, and antibodies to the N-terminus of citrin, citrin 305–319 and aralar1 507–520 were used at a dilution of 1 : 2000. To control for the amount of mitochondrial protein loaded, blots were stripped and incubated with an antibody to the mitochondrial protein b-F 1 ATPase (a gift from J. M. Cuezva, Centro de Biologia Molecular devero Ochoa, JAM, Spain) 2 at a dilution of 1 : 5000. The densities of the bands were evaluated with a Bio-Rad GS-710 calibrated imaging densitometer. Immunocytochemistry The animals were anesthetized with sodium pentobarbital, and perfused through the cardiac ventricle, first with 50 mL 0.9% NaCl followed by 250 mL fixative solution containing 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4, at room temperature. The tissues were removed, postfixed at 4 °C for 24 h, and cryoprotected by immersion in 30% sucrose. Free-floating cryostat 40-lm-thick sections were first quenched with 3% H 2 O 2 in 10% methanol for 20 min in potassium phosphate-buffered saline (NaCl/P i ). After this treatment, the sections were preincubated for 2–3 h in NaCl/ P i containing 5% horse serum and 0.25% Triton X-100 and incubated overnight with antiaralar1 antibody at a dilution of 1 : 100 in 1% horse serum and 0.25% Triton X-100 in NaCl/P i . Secondary biotinylated antibody (goat anti-rabbit; Vector; 1 : 150 dilution) was then incubated for 1–2 h, followed by a 1-h reaction with avidin–biotin–peroxidase complexes (regular ABC kit Vectastain; Vector). Sections were developed using 0.05% 3,3-diaminobenzidine (Sigma) in the presence of 0.03% H 2 O 2 in NaCl/P i for 1–2 min. Sections were mounted on to polylysine-coated slides, dehydrated, delipidated, and mounted in DPX (BDH). RESULTS Expression of aralar1 and citrin during embryonic mouse development The expression of aralar1 and citrin was studied by using in situ hybridizations in toto or on cryostat tissue sections, depending on the stage of the embryos. The data obtained partially confirmed and further extended those reported by Sinasac et al. [11] on embryonic expression of citrin in mouse. At E11, the earliest stage analyzed, both aralar1 and citrin were expressed throughout the developing embryo, with stronger expression in the branchial arches, the developing dermomyotome, the limb and the tail buds (Fig. 1A–E,a–e). In spite of the apparent similarities in distribution, citrin expression was predominantly, although not uniquely, associated with the ectodermal, whereas aralar1 expression was more abundant in the mesenchymal components of these structures (Fig. 1B–E,b–e). In particular, citrin but not aralar1 was strongly expressed in the apical ectodermal ridge of the limb and on the tip of the tail bud (Fig. 1A,B). As an additional difference between the two genes, aralar1 but not citrin transcripts were found in the heart (Fig. 1C,c). At later stages of development (E13–E15), the mRNAs of the two genes were also detected in neural tissue. A few days later (E18), the expression of aralar1 and, to a lesser extent, citrin became clearly localized to selected brain regions such as the cortex and hippocampus, the ventromedial thalamus, the mitral cell layer of the olfactory bulb (Fig. 1L–M,l–m), and the developing striatum (not shown). In the peripheral nervous system, aralar1 and citrin mRNAs were detected, at similar levels, in the trigeminal ganglia (Fig. 1L,l). At E18, when organogenesis becomes a predominant event in embryonic development, the mRNAs of the two genes became differentially localized to particular organs and tissues. Skeletal muscle showed high levels of aralar1 expression, whereas the detection of citrin was negligible in this tissue (compare Fig. 1K with 1k). Similarly, aralar1 but not citrin transcripts were present in the heart, mainly confined to the auricular myocardium (compare Fig. 1I with 1i). The gut endothelium expressed both citrin and aralar1 (Fig. 1f–g,F–G), aralar1 transcripts being localized to the basolateral region of the enterocytes (Fig. 1g). Gut endothelium is a site where arginine biosynthesis occurs in the suckling rat [22] and where the AGCs probably function to provide aspartate for argininosuccinate synthesis. Both genes were also expressed in the kidney but with a differential distribution (Fig. 1H,h). In particular, only high levels of citrin transcripts were found in the epithelium of the tubules, whereas the expression of aralar1 was associated with mesenchymal components (Fig. 1H,h). In summary, the two AGC isoforms have a partially overlapping expression pattern at early stages of embryo- genesis. At later stages, the expression domain of the two genes diverges, and aralar1 distribution becomes predom- inant in brain, heart and skeletal muscle, whereas citrin expression only predominates in kidney. Distribution of AGC isoforms in tissues from the adult rat In adult rat tissues, the distribution of aralar1 and citrin transcripts was analysed by Northern blot. Two aralar1 transcripts of  2.7 and 3.8 kb were detected in all positive tissues (Fig. 2A) as in humans [6]. The hybridization signal was higher for the 2.7-kb than the 3.8-kb mRNA. Expres- sion was stronger in heart and skeletal muscle, followed by brain, and lower in kidney. No aralar1 mRNA was detected in liver. On the other hand, the rat citrin gene presented a Ó FEBS 2002 Development of aspartate/glutamate mitochondrial carriers (Eur. J. Biochem. 269) 3315 single transcript of about 3 kb, consistent with the size of the mouse citrin cDNA and with the data reported for human [7,11,18]. Citrin mRNA was abundant in liver, kidney and heart but was notably absent from brain or skeletal muscle. Therefore, the expression pattern of both citrin and aralar1 in rat is consistent with that described for human and mouse tissues [6,8,18]. The data on the mRNA distribution of the AGC isoforms were complemented by the analysis of the content of the respective proteins in mitochondria-enriched extracts using Western blots with isoform-specific antibodies. As shown in Fig. 2B, aralar1 levels were highest in heart, forebrain, cerebellum and skeletal muscle, in agreement with its mRNA distribution. Mitochondrial extracts from two types of skeletal muscles, the fast-twitch glycolytic extensor digitorum longus and the slow-twitch oxidative soleus, showed similar levels of aralar1 protein. In heart, aralar1 levels were higher in auricular than ventricular Fig. 1. Comparison of expression pattern of citrin and aralar1 during murine embryonic development. Whole embryos from embryonic day (E) 11 (A–E, a–e), in toto E18 isolated organs (H–I, h–i), or transverse E18 cryostat tissue sections (F, G, L, M; f, g, l, m) were hybridized with digoxigenin- labeled probes specific for the citrin or aralar1 genes. Images in (A–E) and (a–e) show E11 embryos, in toto (A, a) and in transverse paraffin sections (B–E, b–e) taken from the embryos in (A and a) at the axial levels indicated by the dotted lines. Note the strong expression in the limb (B) and tail buds (arrowhead in A), in the branchial arches (D) and dermomyotome (E), more strongly localized in the ectodermal component for citrin (A–E), while to the mesenchyme for aralar1 (a–e). At E11, aralar1 (c) but not citrin (C) was expressed in the heart. Images in (F–M; f–m) illustrate the comparative expression of citrin and aralar1 in the gut (F–G; f–g), kidney (H, h), heart (I, i), skeletal muscle (K, k), cortex (L, l), olfactory bulbs (M, m) from E18 embryos. Note the stronger expression of aralar1 in skeletal muscle, heart and neural tissue. Note also the basolateral localization of the in situ hybridization signal in the cells of the gut. Abbreviations: aer, apical ectodermal ridge; am, auricular myocardium; ba, branchial arch; cx, cortex; dm, dermomyotome; h, heart, hc, hippocampus; lb, limb bud; mcl, mitral cell layer; smf, skeletal muscle fiber; tg, trigeminal ganglia; vmt; ventromedial thalamus. Scale bars ¼ 500 lm (B, b; D, d; F, f; H, h; I, i; L, i); 250 lm (C, c; E, e; M, m); 100 lm(K,k);50lm(G,g). 3316 A. del Arco et al.(Eur. J. Biochem. 269) Ó FEBS 2002 myocardium (Fig. 2C). Lower levels of aralar1 were also present in a wide range of tissues, including lung, kidney, ovary, spleen, pancreas (particularly in b islets) and stomach and to an even lower extent in mitochondrial extracts of testis, intestine and liver. Citrin was absent from the central nervous system, skeletal muscle and lung but abundant in liver, kidney and heart, in agreement with its mRNA distribution. Weaker expression was also observed in ovary, testis, spleen, stomach and pancreas, but not in b islets (Fig. 2B). Citrin and aralar1 were hardly observed at all in the adult intestine (Fig. 2B), where their expression was instead high at embryonic day 18–19 (Fig. 1F–G,f–g and [18]). Neither aralar1 nor citrin were detectable in either brown or white adipose tissue (Fig. 2B). It is interesting to note that aralar1 protein is present in lung, where aralar1 mRNA was not detectable [6,8,18]. Similarly, citrin and aralar1 proteins, but not their corresponding mRNAs [18], were detected in spleen and testis. Altogether, these results show that mRNA levels are poor indicators of the levels of the AGC proteins. To assess the relative expression of aralar1 and citrin, mitochondrial fractions from a few representative tissues were processed in parallel with either known amounts of recombinant citrin [7] and aralar1 [6] or mitochondrial fractions from HEK-293T cells overexpressing either aralar1 or citrin, or control HEK-293T cells, which have citrin/aralar1 ratios of about 0.5, 12, and 2.4, respectively [12]. Serial dilution of the recombinant proteins (Fig. 2D) or mitochondrial extracts from HEK-293T cells overexpress- ing aralar1 or citrin (not shown) revealed a linear relation between the amounts of the protein and the densities of the immunoreactive bands. The analysis indicated that spleen, heart (particularly the ventricle), ovary, and stomach have similar levels of citrin and aralar1. Liver, kidney, whole pancreas and testis clearly have higher levels of citrin than aralar1. In contrast, central nervous system tissue, skeletal muscle, lung and possibly auricular myocardium (Fig. 2C) predominantly have aralar1. Expression of aralar1 in cells from the hematopoietic system Citrin is expressed at high levels in the liver. This molecule is the liver-specific AGC isoform, as citrin deficiency causes Fig. 2. Pattern of expression of aralar1 and citrin in adult rat tissues. (A) Northern blot analysis of tissue-specific expression patterns of AGC isoforms. Northern blots with 20 lg total RNA from adult rat heart, kidney, brain, liver and skeletal muscle were hybridized with a 32 P-labeled DNA probe of human aralar1 under high-stringency conditions. The blot was subsequently stripped and reprobed under identical conditions with a human citrin probe The size of the specific transcripts is indicated. Staining with ethidium bromide was carried out to verify the amount of RNA loaded (lower panel). (B) Distribution of AGC isoforms in rat tissues; 20–30 lg mitochondrial protein was used for all tissues except for b islets where 20 lg total protein extract obtained from 500 pancreatic islets was used. Blots for citrin and aralar1 were performed in parallel and reincubated with anti-(b-F 1 ATPase). Aralar1 was detected with an antibody directed against its N-terminus, or against Aralar1 amino acids 507– 520 (results not shown). Citrin antibodies were against its N-terminus (upper panels) or against citrin amino acids 305–319 (lower panels), both at 1 : 2000 dilution. Bands correspond to 70 kDa (aralar1 and citrin) and 52 kDa for b-F 1 ATPase. (C) Western blot analysis of aralar1 and citrin in auricular and ventricular myocardium. 20 lg of atria (A) and ventricle (V) mitochondrial extracts from two different animals were analysed. The blot was incubated with anti-(aralar N-terminus) (1 : 5000) and anti-(b-F 1 ATPase) (1 : 5000), stripped and probed again with anti-(citrin N-terminus) (1 : 2000). The amount of aralar1 (standardized to that of b-F 1 ATPase) was 1.5 and 0.9 in atria and ventricles, respectively. No significant changes between atria and ventricle are observed for citrin levels (0.86 and 0.64 standardized values, in atria and ventricles, respectively). (D) Immunoblotting of increasing amounts of recombinant citrin and aralar1. Known amounts (as indicated) of bacterially expressed aralar1 and citrin N-terminal regions were loaded on to gels and blotted and processed in parallel with an identical dilution (1 : 2000) of their respective specific antiserum antibodies. Note the different amounts of recombinant aralar1 and citrin used. Ó FEBS 2002 Development of aspartate/glutamate mitochondrial carriers (Eur. J. Biochem. 269) 3317 CTLN2 [8,23], indicating that aralar1 does not compensate for the loss of citrin in liver. Surprisingly, however, we detected a low but unequivocal presence of aralar1 protein in the adult liver (Fig. 2B, see also Fig. 3C). Furthermore, even though aralar1 mRNA was not detected by Northern blots of whole liver (Fig. 2A), similarly to that reported by Iijima et al. [18] for different liver cell types (hepatocytes, stellate, endothelial and Kupffer cells), rat aralar1 cDNA was readily amplified by RT-PCR from liver mRNA (A. del Arco et al., unpublished data), indicating that aralar1 transcripts are present in this tissue, albeit at very low levels. To determine the cellular source of aralar1, sections from adult rat liver were immunostained with specific antibodies. As observed in Fig. 3A, aralar1 was not localized to the parenchymal hepatocytes, but to sparse spindle-shaped cells which, by morphological criteria and position, probably correspond to Kupffer cells, the liver resident macrophages [24]. The fact that Iijima et al. [18] did not detect aralar1 mRNA in isolated Kupffer cells in Northern blots probably reflects either changes in aralar1 mRNA during isolation and plating of the cells or the lack of correspondence between the AGC mRNA and protein levels, a situation also found for both aralar1 and citrin in lung, spleen and testis, as mentioned above. The presence of aralar1 in Kupffer cells is further supported by the observation that aralar1 is expressed by other cells of the hematopoietic system. Thus, it is present in mitochondrial extracts obtained from a murine macrophage-like cell line, the RAW 264 cells (Fig. 3B). Aralar1 mRNA and protein were also detected in Jurkat cells and human T-lymphocytes, respectively (Fig. 3B and data not shown) and in bone marrow (Fig. 3B). In contrast, citrin was not detected in RAW 264 cells (Fig. 3B), and citrin mRNA and protein were absent from human T-lymphocytes and Jurkat cells [7] (Fig. 3B). Fetal liver together with the yolk sac are the hemato- poietic organs in prenatal mammalian development [25]. We detected by Western blotting aralar1 and citrin in liver mitochondrial extracts from rat embryos (E18), neonates (1–6 h after birth, P0) and adult animals (3 months old, A). Figure 3C shows that aralar1 levels decreased dramatically after birth, from about 3.9 at E18 to 2.3 in P0 and 0.9 in adults (numbers correspond to the aralar1 signal standard- ized to that of b-F 1 ATPase; mean of two experiments). This decrease matches the gradual loss of liver hematopoiesis at the end of fetal life, when spleen and bone marrow become the major hematopoietic organs [24]. In contrast, citrin levels in liver mitochondrial extracts increased markedly during postnatal development (Fig. 3C). Indeed, Iijima et al. [18] found that citrin expression increases in liver just before birth, in parallel to that of ASS and carbamoyl phosphate synthetase, to provide full development of the urea cycle early in postnatal life [26]. DISCUSSION This study shows the mitochondrial expression levels of aralar1 and citrin proteins in a large number of tissues and organs. The distribution of aralar1 and citrin mRNAs has been previously compared in different tissues and during postnatal development [8,18]. Although the protein levels in some tissues are consistent with their mRNA data, we have obtained new information on the tissue distribution of the AGCs which is of great interest in the search for specific functions for these proteins. In particular, aralar1 is present in many more tissues than suggested by its mRNA distribution [6–8,18]. Thus, although aralar1 mRNA is highly represented in brain, skeletal muscle and heart, aralar1 protein is not restricted to these excitable tissues but it is also found in lung, stomach, pancreas (particularly b cells), kidney, and ovary, and it is the main isoform present in hematopoietic tissues. On the other hand, citrin was expressed not only in kidney and liver, the classic gluconeogenic organs, but was present at significant levels in heart, stomach, pancreas and testis. The absence of detectable aralar1 mRNA in tissues where aralar1 protein is readily observed suggests that its expression may be regulated at post-transcriptional levels, as is known for other mitochondrial proteins involved in bioenergetic func- tions [27]. The distribution of citrin mRNA in mouse embryos has been studied previously [11]. However, this is the first time that aralar1 expression has been studied in mouse embryos by in situ hybridization and compared with that of citrin.In contrast with the situation in the adult animal, this study shows that there is a wide overlap in the expression of the Fig. 3. Expression of aralar1 in cells of the immune system. (A) Immunohistochemical detection of aralar1 in liver rat sections. Aralar1 positive cells are indicated by arrows. No signal is observed in hepatocytes. Those sections in which either the primary or secondary antibodies or the ABC reagent were omitted were negative. Scale bar ¼ 75 lm. (B) Western blot analysis of aralar1 and citrin in cells from the immune system and hematopoietic tissues. Mitochondrial extracts (20 lg) obtained from human Jurkat cells (a T-cell leukemia cell line) and the mouse macrophage cell line RAW 264, as well as the hematopoietic tissues, bone marrow and spleen, were loaded on to gels. Mitochondrial extracts from HEK-293T cells with a known aralar1/citrin ratio were included as an internal control [12]. The membranes were processed as described in Fig. 2B 4 with antibodies to aralar1 N-terminus (1 : 5000) and citrin N-terminus (1 : 2000). (C) Western blot analysis of AGC isoforms during rat liver development. The mitochondrial extracts (20 lg per lane) were obtained from fetuses at embryonic day 18 (E18), from pups 1–6 h after birth (P0), and from 3-month- old rats (A). The blots were processed as described in the legend to Fig. 2B 5 . 3318 A. del Arco et al.(Eur. J. Biochem. 269) Ó FEBS 2002 two isoforms during early embryogenesis. At early stages of embryonic development, the mRNAs of both isoforms were localized particularly in actively growing structures (limb and tail buds, apical ectodermal ridge, etc.) but with different tissue distributions. Later, the two isoforms show a widespread and dynamic expression pattern that does not always reflect their final distribution in adult tissues. For example, both aralar1 and citrin mRNAs are expressed in the dermamyotome, from which skeletal muscle will originate, but there is progressive loss of citrin expression throughout embryogenesis (compare Fig. 1K with 1k), and it is finally absent from adult skeletal muscle. Similarly, both aralar1 and citrin are expressed in the central and peripheral nervous system at E18, but only aralar1 is observed in the adult brain. In contrast, only aralar1 was expressed in the developing heart, but both AGC isoforms are present in the adult rat tissue (Fig. 2). Interestingly, however, the two isoforms are differentially distributed, and aralar1 is more abundant in the atrial myocardium. Overall, the substantial overlap in the distribution of the two AGC isoforms in early embryos suggests a redundancy of function and may explain why CTLN2 patients with non- functional citrin protein do not suffer from major develop- mental symptoms. The finding that most adult tissues express aralar1, with the notable exception of liver hepato- cytes, contrasts with previous indications that aralar1 distribution was restricted to excitable tissues [6–8] and may explain why citrin deficiency only affects the liver. Indeed, CTLN2 patients have normal levels of ASS in tissues other than the liver [8,18,19,28], suggesting that the function provided by citrin, i.e, the efflux of aspartate from mitochondria as substrate of ASS, can also be accomplished by aralar1, a protein more widely expressed than previously believed. This argues against major func- tional differences between the two isoforms, and is consis- tent with the results obtained with the recombinant proteins reconstituted in proteoliposomes, and expressed in human cells [12]. On the other hand, the presence of a single major AGC isoform, aralar1, in skeletal muscle, central nervous system, and cells from the hematopoietic system suggests that mutations in aralar1 would have a preferential impact in these tissues. ACKNOWLEDGEMENTS This work was supported by grants from the Spanish Direccion General de Investigacio ´ nCientı ´ ficayTe ´ cnica, Comunidad Auto ´ noma de Madrid, Quı ´ mica Farmace ´ utica Bayer, S.A., and by an institutional grant from the Fundacio ´ nRamo ´ n Areces to the Centro de Biologı ´ a Molecular ÔSevero OchoaÕ. We thank Dr Isabel Valverde for providing the extracts of b islets and Professor J. M. Cuezva for the gift of antibodies to b-F 1 ATPase. We also thank Dr Alberto Martı ´ nez- Serrano for critical reading of the manuscript. REFERENCES 1. Walker, J.E. & Runswick, M.J. (1993) The mitochondrial trans- port protein superfamily. J. Bioenerg. Biomembr. 25, 435–446. 2. Indiveri, C., Iacobazzi, V., Giangregorio, N. & Palmieri, F. (1997) The purified and reconstituted ornithine/citrulline carrier from rat liver mitochondria: electrical nature and coupling of the exchange reaction with H + translocation. Biochem. J. 321, 713–719. 3. Fiermonte, G., Dolce, V., Arrigoni, R., Runswick, M.J. & Walker, J.E. (1999) Organization and sequence of the gene for the human mitochondrial dicarboxylate carrier: evolution of the carrier family. Biochem. J. 344, 953–960. 4. Fiermonte, G., Dolce, V., Palmieri, L., Ventura, M., Runswick, M.J., Palmieri, F. & Walker, J.E. (2001) Identification of the human mitochondrial oxodicarboxylate carrier: bacterial expres- sion, reconstitution, functional characterization, tissue distribu- tion and chromosomal location. J. Biol. Chem. 276, 8225–8230. 5. Dolce, V., Fiermonte, G., Runswick, M.J., Palmieri, F. & Walker, J.E. (2001) The human mitochondrial deoxynucleotide carrier and its role in toxicity of nucleoside antivirals. Proc. Natl Acad. Sci. USA 98, 2284–2288. 6. del Arco, A. & Satru´ stegui, J. (1998) Molecular cloning of Aralar, a new member of the mitochondrial carrier superfamily that binds calcium and is present in human muscle and brain. J. Biol. Chem. 273, 23327–23334. 7. delArco,A.,Agudo,M.&Satru´ stegui, J. (2000) Characterization of a second member of the subfamily of calcium-binding mitochondrial carriers expressed in human non-excitable tissues. Biochem. J. 345, 725–732. 8. Kobayashi, K., Sinasac, D.S., Iijima, M., Boright, A.P., Begum, L.,Lee,J.R.,Yasuda,T.,Ikeda,S.,Hirano,R.,Terazono,H., Crackower, M.A., Kondo, I., Tsui, L C., Scherer, S.W. & Saheki, T. (1999) The gene mutated in adult-onset type II citrullinaemia encodes a putative mitochondrial carrier protein. Nat. Genet. 22, 159–163. 9. Sanz, R., del Arco, A., Ayuso, C., Ramos, C. & Satru´ stegui, J. (2000) Assignment of the calcium-binding mitochondrial carrier gene ARALAR1, to human chromosome band 2q31 by in situ hybridization. Cytogenet. Cell. Genet. 89, 143–144. 10. Crackower, M.A., Sinasac, D.S., Lee, J.R., Herbrick, J A., Tsui, L C. & Scherer, S.W. (1999) Assignment of the SLC25A12 gene coding for the human calcium-binding mitochondrial solute car- rier protein aralar to human chromosome 2q24. Cytogenet. Cell. Genet. 87, 197–198. 11. Sinasac, D.S., Crackower, M.A., Lee, J.R., Kobayashi, K., Saheki, T., Scherer, S.W. & Tsui, L C. (1999) Genomic structure of the adult-onset type II citrullinemia gene, SLC25A13, and cloning and expression of its mouse homologue. Genomics 62, 289–292. 12. Palmieri, L., Pardo, B., Lasorsa, F.M., del Arco, A., Kobayashi, K., Iijima, M., Runswick, M.J., Walker, J.E., Saheki, T., Satru´ stegui,J.&Palmieri,F.(2001)Citrinandaralar1areCa 2+ stimulated aspartate/glutamate transporters in mitochondria. EMBO J. 20, 5060–5069. 13. Dierks, T., Riemer, E. & Kramer, R. (1988) Reaction mechanism of the reconstituted aspartate/glutamate carrier from bovine heart mitochondria. Biochim. Biophys. Acta 943, 231–244. 14. Williamson, J.R., Meijer, A.J. & Ohkawa, K. (1974) Interrelations between anion transport, ureogenesis and gluconeogenesis in isolated rat liver cells. In Regulation of Hepatic Metabolism (Lundquist, F. & Tygstrup, N., eds), pp. 457–479. Munksgaard, Copenhagen. 15. LaNoue, K.F. & Schoolwerth, A.C. (1979) Metabolite transport in mitochondria. Annu. Rev. Biochem. 48, 871–922. 16. Yasuda, T., Yamaguchi, N., Kobayashi, K., Nishi, I., Horinouchi, H., Jalil, M.A., Li, M.X., Ushikai, M., Iijima, M., Kondo, I. & Saheki, T. (2000) Identification of two novel mutations in the SLC25A13 gene and detection of seven mutations in 102 patients with adult-onset type II citrullinemia. Hum. Genet. 107, 537–545. 17. Meijer, A.J., Gimpel, J.A., Deleeuw, G., Tischler, M.E., Tager, M.E. & Williamson, J.R. (1978) Interrelationships between glu- coneogenesis and ureogenesis in isolated hepatocytes. J. Biol. Chem. 253, 2308–2320. 18. Iijima, M., Jalil, A., Begum, L., Yasuda, T., Yamaguchi, N., Xian Li, M., Kawada, N., Endou, H., Kobayashi, K. & Saheki, T. (2001) Pathogenesis of adult-onset type II citrullinemia caused by deficiency of citrin, a mitochondrial solute carrier protein: tissue Ó FEBS 2002 Development of aspartate/glutamate mitochondrial carriers (Eur. J. Biochem. 269) 3319 and subcellular localization of citrin. Adv. Enzyme Regul. 41, 325–342. 19. Saheki, T., Tsuda, M., Takada, S., Kusumi, K. & Katsunuma, T. (1980) Role of argininosuccinate synthetase in the regulation of urea synthesis in the rat and argininosuccinate synthetase-asso- ciated metabolic disorders in man. Adv. Enzyme Regul. 18, 221–238. 20. Villanueva-Penacarrillo, M.L., Cancelas, J., de Miguel, F., Redondo, A., Valin, A., Valverde, I. & Esbrit, P. (1999) Para- thyroid hormone-related peptide stimulates DNA synthesis and insulin secretion in pancreatic islets. J. Endocrinol. 163, 403–408. 21. Bovolenta, P., Mallamaci, A., Puelles, L. & Boncinelli, E. (1998) Expression pattern of cSix3, a member of the Six/sine oculis family of transcription factors. Mech. Dev. 70, 201–203. 22. De Jonge, W.J., Dingemanse, M.A., de Boer, P.A., Lamers, W.H. & Moorman, A.F. (1998) Arginine-metabolizing enzymes in the developing rat small intestine. Pediatr. Res. 43, 442–451. 23. Kobayashi, K., Iijima, M., Yasuda, T., Sinasac, D.S., Yamaguchi, N., Tsui, L C., Scherer, S.W. & Saheki, T. (2000) Type II citrullinemia (citrin deficiency): a mysterious disease caused by a defect of calcium-binding mitochondrial carrier protein. In Calcium: the Molecular Basis of Calcium Action in Biology and Medicine (Pochet, R., Donato, R., Haiech, J., Heiz- mann, C. & Gerke, V., eds), pp. 557–579. Kluwer Academic Publishers, Dordrecht. 24. Morris, L., Graham, C.F. & Gordon, S. (1991) Macrophages in haemopoietic and other tissues of rthe developing mouse detected by the monoclonal antibody F4/80. Development 112, 517–526. 25. Gilbert, S.F. (1991) Developmental Biology. Sinauer Associates, Inc., Sunderland, MA, USA. 26. Morris,S.M.Jr,Kepka,D.M.,Sweeney,W.E.Jr&Avner,E.D. (1989) Abundance of mRNAs encoding urea cycle enzymes in fetal and neonatal mouse liver. Arch. Biochem. Biophys. 269, 175–180. 27. Izquierdo, J.M. & Cuezva, J.M. (1997) Control of the transla- tional efficiency of beta-F1-ATPase mRNA depends on the reg- ulation of a protein that binds the 3¢ untranslated region of the mRNA. Mol. Cell. Biol. 17, 5255–5268. 28. Kobayashi, K., Shaheen, N., Kumashiro, R., Tanikawa, K., O’Brien, W.E., Beaudet, A.L. & Saheki, T. (1993) A search for the primary abnormality in adult-onset type II citrullinemia. Am. J. Hum. Genet. 53, 1024–1030. 3320 A. del Arco et al.(Eur. J. Biochem. 269) Ó FEBS 2002 . embryonic and adult tissues is largely unknown. We investigated the pattern of expression of aralar1 and citrin in murine embryonic and adult tissues at the. Role of argininosuccinate synthetase in the regulation of urea synthesis in the rat and argininosuccinate synthetase-asso- ciated metabolic disorders in