Characterization of Human RhCG and Mouse Rhcg as Novel Nonerythroid Rh Glycoprotein Homologues Predominantly Expressed in Kidney and Testis*

In mammals, the Rh family includes the variable Rh polypeptides and invariant RhAG glycoprotein. These polytopic proteins are confined to the erythroid lineage and are assembled into a multisubunit complex essential for Rh antigen expression and plasma membrane integrity. Here, we report the characterization of RhCG and Rhcg, a pair of novel Rh homologues present in human and mouse nonerythroid tissues. Despite sharing a notable similarity to the erythroid forms, including the 12-transmembrane topological fold, theRHCG/Rhcg pair is distinct in chromosome location, genomic organization, promoter structure, and tissue-specific expression.RHCG and Rhcg map at 15q25 of human chromosome 15 and the long arm of mouse chromosome 7, respectively, each having 11 exons and a CpG-rich promoter. Northern blots detected kidney and testis as the major organs of RHCG or Rhcgexpression. In situ hybridization revealed strong expression of Rhcg in the kidney collecting tubules and testis seminiferous tubules. Confocal imaging of transiently expressed green fluorescence protein fusion proteins localized RhCG exclusively to the plasma membrane, a distribution confirmed by cellular fractionation and Western blot analysis. In vitrotranslation and ex vivo expression showed that RhCG carries a complex N-glycan, probably at the48NLS50 sequon of exoloop 1. These results pinpoint RhCG and Rhcg as novel polytopic membrane glycoproteins that may function as epithelial transporters maintaining normal homeostatic conditions in kidney and testis.

NLS 50 sequon of exoloop 1. These results pinpoint RhCG and Rhcg as novel polytopic membrane glycoproteins that may function as epithelial transporters maintaining normal homeostatic conditions in kidney and testis.
The Rh antigens, originally identified in human red blood cells (RBCs), 1 are potent immunogens and, when incompatible, cause hemolytic disease of the newborn and blood transfusion reaction (1). They are defined by two erythroid-specific transmembrane (TM) proteins, the Rh polypeptides and the Rhassociated glycoprotein (RhAG), which form a multisubunit complex and display exodomains as D or CcEe antigens (2). The Rh polypeptides are highly polymorphic and are distinguished from RhAG by two biochemical features, i.e. palmitoylation but no glycosylation. In contrast, RhAG is largely invariant at the population level and is thought to modulate the assembly of the Rh complex and its surface expression. Human Rh polypeptides and RhAG are encoded by RHCED and RHAG loci that reside on chromosomes 1 and 6, respectively (3,4). Despite this difference, these RH genes have originated from a common ancestor during evolution in that they are homologous in coding sequence and are grossly similar in exon/intron organization (5,6).
Although their exact functions remain to be identified, Rh proteins (and their complex) are necessary for the maintenance of RBC morphology and plasma membrane integrity. The Rh deficiency syndrome, a rare inherited form of hemolytic anemia, is caused by mutations at the RHAG or RHCED locus (2). In this disorder, RBCs deficient in all Rh antigens exhibit spherostomatocytosis and multiple membrane abnormalities (7), implying that Rh proteins have some functional roles in membrane physiological processes. At the level of secondary structure, RhAG and RhCE/D bear a 12-TM topology resembling many transporters (8) and thus are likely to possess a transporter activity across the lipid bilayer. Ammonium or amino group-containing compounds have been suggested as candidate ligands for erythroid Rh proteins, owing to their connective sequence similarity with the NH 4 ϩ transporters from bacteria, yeast, and plants (2,9).
The Rh family of genes and proteins is rooted deeply in evolution, and the erythroid homologues of human RHAG and RHCED genes are coexpressed in RBC of all mammalian species. The closest RHAG and RHCED relatives are those from nonhuman primates, as they share not only high sequence identity but also some antigen reactivity (10,11). We previously showed that the mouse erythroid Rhag and Rhced are identical to their human counterparts in the arrangement of exon/intron junctions, in the conservation of chromosome synteny, and in the pattern of coexpression (12). Furthermore, Rh homologues have been identified in primitive organisms, e.g. the unicellular slime mold Dictyostelium discoideum (2), marine sponge Geodia cydonium (13), and earthworm Caenorhabditis elegans (14,15). Structurally, these primitive Rh forms are more similar to RhAG than they are to RhCE/D, providing some important insights into the origin of erythroid members from nonerythroid ancestors. However, it is not known whether these Rh homologues in primitive organisms had branched out, giving rise to new family members expressed in nonerythroid tissues or organs of vertebrate species.
To understand the structure-function relationships of Rh proteins, we have been using a homology approach to clone and identify new Rh genes from lower organisms and from nonerythroid tissues. We describe here the characterization of Rh type C glycoproteins, human RhCG and mouse Rhcg, two orthologous members of the Rh protein family expressed in nonerythroid tissues. We show that RhCG and Rhcg share with their erythroid counterparts the conserved 12-TM topology and cognate signature sequences. However, as the first vertebrate members recognized in the nonerythroid Rh subfamily, RhCG and Rhcg are distinct in primary structure, genomic organization, chromosomal location, tissue-specific expression, and biochemical properties. The studies detailed herein pinpoint RhCG and Rhcg as novel polytopic membrane glycoproteins that may function as epithelial transporters maintaining normal homeostatic conditions in the kidney and testis.
Comparison of RhCG, Rhcg, and Other Homologues-The amino acid sequences of RhCG, Rhcg and other Rh homologues were aligned by means of Clustal W (19) (MegAlign software, DNASTAR). The hydropathy plots were obtained using the Kyte-Doolittle method (20).
Production of Polyclonal Antisera for RhCG C-tail-The RhCG C-tail was expressed in E. coli BL21 (0.3 mM isopropylthio-␤-galactoside at 30°C for 4 h) as a glutathione S-transferase fusion protein or a His 6tagged peptide. After sonication, glutathione S-transferase-RhCG was purified on glutathione Sepharose 4B, and ϳ300 g was emulsified with adjuvant and injected into rabbits five times (21). Antisera were affinity-purified by passing through a glutathione S-transferase column (Pierce) and then a Ni-NTA column (Qiagen) bound with His-tagged RhCG tail. After washing, the antibodies were eluted with 4 M MgCl 2 and dialyzed in 1ϫ phosphate-buffered saline at 4°C. The IgG fractions showed binding to recombinant RhCG C-tails only.
Screening of Genomic Clones and Definition of Exon/Intron Boundaries-Bacterial artificial chromosome (BAC) clones of RHCG or Rhcg were screened by PCR. The GSPs for RHCG (E6s, CGTGGGTACCGCT-GCTGAGAT; E7a, TATGATGCCAGGAATGCCATGCAG) gave a 360-bp band. The GSPs for Rhcg (E6s, GCTCTCATCGTGGGGTTCT-TCTGC; E7a, CAGGTTGTGRATGCCACATGTGTC) gave a 310-bp band. BAC DNA was fingerprinted by exon PCR and mapped by Southern blotting. Exon/intron boundaries were amplified in two steps from genomic libraries (5, 12) and sequenced. Exon/intron junctions were defined by alignment with cDNA sequences. Introns were amplified from BAC DNA, and their sizes were estimated by gel electrophoresis.
5Ј-RACE and 5Ј Genomic Walking-5Ј-RACE was performed as mentioned above. The 5Ј region of RHCG or Rhcg gene was amplified from the human or mouse genomic libraries (5, 12) using adaptor primers and exon 1 GSPs. The PCR products were purified and sequenced.
Northern Blot Analysis and RNA in Situ Hybridization-Two sets of Northern blots (CLONTECH) were hybridized to RHCG or Rhcg cDNA. The RHCG (codon 250 -437) and Rhcg (codon 194 -377) probes were 563 and 552 bp in size, respectively. As a control, the ␤-actin probe was used.
RNA in situ hybridization to mouse embryos and tissues was carried out using standard protocols, as described (24). A 384-bp sequence covering the 3Ј region of Rhcg cDNA (nucleotides 1110 -1494) was subcloned in pCRScript Sk(ϩ) vector (Stratagene) and used to prepare 33 P-labeled probes. The antisense and sense probes were generated by in vitro transcription of the BamHI-and NotI-linearized Rhcg plasmids by T7 and T3 RNA polymerases, respectively.
Transfection of RHCG cDNA and Confocal Imaging of RhCG Protein-HEK293 and HeLa cells (ATCC) were grown at 37°C in Dulbecco's modified Eagle's medium containing 10% fetal calf serum (Life Technologies, Inc.) under 5% CO 2 . After 24 h, cells were transferred to either a six-well plate or cover glass and transfected by LipofectAMINE (Life Technologies, Inc.). For transient expression, 1 g of RHCG cDNA was used to transfect 3 ϫ 10 5 HEK293 or HeLa cells.
For stable selection, RHCG.myc/HisA plasmid (1 g, PvuI cut) was used to transfect 3 ϫ 10 5 HEK293 cells/well in a six-well plate. Stable cell lines were selected in Dulbecco's modified Eagle's medium (G418, 800 g/ml) and clonally isolated by standard procedures (18). For confocal imaging, 1 g of RhCG-GFP or GFP-RhCG plasmid was transfected in 3 ϫ 10 5 HEK293 or HeLa cells plated onto a 35-mm coverglass (MatTek) and cultured for 24 h. GFP was excited at 488 nm with an argon laser, and the light emitted between 506 and 538 nm was recorded for fluorescein isothiocyanate filter. Images were collected using a Bio-Rad MRC 600 confocal scan head on a Nikon Eclipse 200 microscope with a 60XN.A.1.4 planapo infinity corrected objective. The captured three-dimensional images were processed with Adobe Photoshop (version 4.0).
In Vitro Translation and N-Glycosylation-To define the biochemical features of RhCG, in vitro translation and posttranslational processing were carried out. As templates, RhCG plasmids of pYES or pcDNA3.1/MycHisA were used in a transcription-coupled translation system (Promega) with [ 35 S]methionine (15 mCi/ml, Amersham Pharmacia Biotech). Labeled RhCG was analyzed by 12% SDS-PAGE (25) after incubation with or without canine pancreatic microsomal membranes (CPMM) (Promega).

RESULTS
Nucleotide Sequences of RHCG and Rhcg cDNAs-Sequencing revealed that RHCG and Rhcg cDNAs, 1952 and 2097 bp long, respectively, share an overall 68.8% identity at the nucleotide level, forming an orthologous pair. The variation in the sequence is mainly in the 5Ј-UTR and 3Ј-end of open reading frames (Fig. 1). RHCG and Rhcg have a consensus polyadenylation signal but lack the typical Kozak sequence (28) and stop codon preceding the first in-frame AUG codon, two features common to the erythroid Rh genes (5, 12). However, RHCG and Rhcg are G/C-rich (RHCG, 57.5%; Rhcg, 55.4%; versus RHAG, 42.6%; Rhag, 41.8%) and have completely different 5Ј-and 3Ј-UTR sequences.
The Primary Structure and Predicted Topology of RhCG and Rhcg Proteins-The open reading frame of human RHCG encodes a 53-kDa, 479-amino acid polypeptide, whereas that of mouse Rhcg encodes a 55-kDa, 498-amino acid polypeptide (Fig. 1). The RhCG and Rhcg proteins are highly conserved, sharing 77.2% identity and 90.4% similarity with regard to their primary structures. RhCG differs from Rhcg by three small deletions: one serine at position 51 and two amino acid stretches in the extreme C-terminal region are absent in RhCG (Fig. 1). The majority of variations are conserved amino acid changes, with ϳ50% of them being clustered at the C-terminal positions of 360/361.
The RhCG (pI ϭ 6.2) and Rhcg (pI ϭ 6.5) proteins are negatively charged at physiologic pH and are composed mainly of hydrophobic (44%) and polar (25%) amino acids. Hydropathy analysis (20) showed that they share identical topology and charge distribution, each spanning the lipid bilayer 12 times, with both N and C termini facing the cytoplasm (Fig. 2). The 12-TM fold and its cognate signatures, including the charged residues (Asp 129 and Glu 166 ), are conserved and reminiscent of other homologs, i.e. the RhAG/Rhag (4, 12) and RhBG/Rhbg pairs (Fig. 3). 2 However, RhCG and Rhcg are distinct in that they have a much elongated C-terminal segment and unique N-terminal and exoloop sequences. There are also three Nglycosylation NX(S/T) motifs in RhCG and two in Rhcg (Fig. 1 suggesting that these proteins are expressed as polytopic glycoproteins. Nevertheless, except for 48 NLS 50 (RhCG) (Fig. 2) or 48 NIS 50 (Rhcg), other NX(S/T) motifs were predicted to reside in the TM or in the cytoplasmic domain and thus are unlikely to become glycosylated in vivo.
Relationship of RhCG/Rhcg with Other Rh Protein Homologs-Along with the identification of RhCG and Rhcg, we also isolated cDNAs encoding new Rh homologues from model organisms and nonerythroid tissues. 2 Multiple sequence alignment yielded a dendrogram in which the known Rh homologues are clustered in three distinct groups: primitive, erythroid, and nonerythroid (Fig. 3). RhCG and Rhcg were placed as a pair in the nonerythroid subgroup that also includes Hs.RhBG, Mm.Rhbg, Dr.Rhg, and Gc.Rhg. All these proteins are polytopic glycoproteins that are likely to contain a single N-glycan on their exoloop 1.
As shown by pairwise sequence comparison, RhCG and Rhcg are most closely related to the nonerythroid RhBG/Rhbg pair (51.7-54.5% identity) 3 and erythroid RhAG/Rhag pair (45.4 -50.9% identity) (Fig. 3). This degree of overall identity is largely attributable to conserved TM segments (particularly TM2-11) and their immediately adjacent amino acids, suggesting that these domains define an important function. Significantly, both RhCG and Rhcg are much less similar to antigen carriers, RhCE/D, or Rhced (24.4 -28.5% identity). Considering their relationship with the primitive subgroup, RhCG and Rhcg are more closely related to the fruit fly homologue Dm.Rhp, with which they share a 37% overall sequence identity (Fig. 3). These results suggest that the RHCG/Rhcg pair originated from early primitive gene precursors and was subjected to an independent evolutionary pathway following its duplication and separation from erythroid members.
Chromosomal Location of RHCG and Rhcg Genes-The chromosomal location of the human RHCG gene was mapped by fluorescence in situ hybridization using the two BAC genomic clones as probes. The result showed that RHCG is located at 15q25 of chromosome 15 (Fig. 4A). Using RHCG as a query for BLAST search, we detected an exactly matched expressed sequence tag (U75833) from the human CEPH YAC clone that had been mapped at 15q26 (29). These data establish a separate location of RHCG from the erythroid RHCED and RHAG loci that are mapped at chromosomes 1p34 -36 and 6p11-21.1, respectively (3,4).
The mouse Rhcg gene was mapped by linkage analysis (Fig.  4B). Rhcg is nonrecombinant to the D7Xrf229 locus, with an LOD score of 28.3. This linkage placed Rhcg distal to Gnb2-rs1 but proximal to Pcsk3 on the long arm of chromosome 7, a region showing conserved synteny with human 15q25 containing RHCG gene (Fig. 4A). Notably, this locus map is also separate from the erythroid Rhced and Rhag loci that are localized to chromosomes 4 and 17, respectively (12).
Genomic Organization and Exon/Intron Structures-BAC clones retaining human RHCG or mouse Rhcg gene were isolated and characterized in order to delineate their structural 3  organization. As shown in Fig. 5, the two genes share a nearly identical organization, each having 11 exons that range from 74 to 187 bp and 10 introns that range from 0.2 to 8.8 kb. In both RHCG and Rhcg genes, exon 11 occurs as a noncoding 3Ј-UTR segment, a unique feature that distinguishes from all other Rh homologs. Their internal exons 2-9 are identical in size, and their exons 2-6 are arranged in the same fashion as exons 2-6 of RHAG and Rhag genes (Fig. 5). Because exons 2-6 encode TM2-9 domains in both erythroid and nonerythroid homologs, their conservation in sequence and size implies strict control of the length of the corresponding TM domains and flanking loops. Fig. 6 shows the sequences of RHCG and Rhcg genes containing splice sites and exon/intron junctions. All 5Ј donor and 3Ј acceptor sites conform to the "GT-AG" rule and show consensus to the primate and rodent gene splicing signals (30). Of the 10 coding exons assigned, all but exons 6 and 10 are asymmetrical containing one (exons 1, 3, 4, 5, 7, and 9) or two (exons 2 and 8) split intercodons at the 5Ј and/or 3Ј exon/intron junction. The amino acids encoded by exon/exon boundaries are the same between RhCG and Rhcg, except for four conserved changes (Fig. 6). Such a division of coding exons is observed in mammals (10 -12) but not in more primitive organisms (data not shown).
Proximal Promoter Sequence and Transcription Start Site-Genomic walking revealed that the 5Ј region upstream of the first ATG codon of RHCG or Rhcg contains multiple cis-acting elements known to bind various transcription factors (Fig. 7). This proximal sequence is highly enriched in CpG dinucleotides (39 CpGs out of 443 bp for RHCG and 25 CpGs out of 279 bp for Rhcg), a hallmark of CpG islands that often overlap with generic promoters (31). The sequence is also highly asymmetrical in strand composition alternating with pyrimidine and purine stretches. Notably, these features and many cis-acting motifs are absent from erythroid RH promoters (5, 6, 12, 32-34).
Sequencing of the 5Ј-RACE products identified -24A and -126A as the major transcription start sites in the RHCG and Rhcg genes, respectively (Fig. 7). The sites were determined by using total RNA isolated from the human kidney and mouse testis tissues, and this mapping result was consistent with the beginning of 5Ј-UTR sequence observed in the full-length cDNAs (Fig. 1). The fact that no in-frame ATG occurs in the genomic sequence further upstream of the transcription start site (Fig. 7) also supports the predicted AUG codon as the translation initiation signal (Fig. 1).
Tissue Expression of RHCG and Rhcg-The expression of RHCG and Rhcg was assessed by Northern blot analysis (Fig.  8) and by in situ hybridization (Fig. 9). In human adult tissues, a 2.0-kb RHCG transcript was found abundantly expressed in the kidneys and also in the brain, testis, placenta, pancreas, and prostate (Fig. 8A). (Its apparent lower expression in the testis was probably caused by an uneven sampling of testicular tissues for poly(A)ϩ RNA preparation.) In human fetal tissues, the kidney was the only organ found to express RHCG transcripts. In mouse adult tissues, Rhcg was shown to be also highly expressed in kidney and testis but not in other tissues, e.g. brain (Fig. 8B). Furthermore, no Rhcg transcript was detected in embryos of 7-19 days of gestation nor in erythroid tissues (data not shown). Taken together, the results indicate that RHCG or Rhcg is largely expressed at very late stages of development and with limited tissue specificities.
The results from RNA in situ hybridization further helped refining of the sites of Rhcg expression in mouse tissues (Fig.  9). In mouse testis, Rhcg was abundantly expressed in the seminiferous tubules (Fig. 9B), the main components of the adult testicle that produce spermatozoa (35). Higher magnification revealed that the Rhcg signals decorated the complex stratified epithelium (Fig. 9C), a lining of the seminiferous tubules known to contain spermatogenic cells and supporting Sertoli cells (35). In mouse kidney, Rhcg was broadly and abundantly expressed in both the renal cortex and the medulla (Fig. 9D). The signals exhibited a radially distributed pattern along their length that corresponded to the parallel alignment of the myriad minute uriniferous tubules that form the functional renal units (35). At higher magnification, the Rhcg signal was primarily confined to the epithelial linings of the collecting tubules (Fig. 9, E and F). These results point toward specific expression of Rhcg in the tubular structures of the testis and the kidney. Cellular Localization of RhCG by Confocal Imaging Analysis-Confocal microscopy (Fig. 10) revealed that the fluorescence signal was confined to the plasma membrane of cells transfected with RhCG-GFP constructs (B, C, E, and F), whereas it was evenly distributed in the cytoplasm of cells transfected with GFP vector alone (A and D). The membrane localization was not dependent on the orientation of GFP in the fusion constructs or on the type of cells used in the transfection assay. These results imply that the RhCG protein is normally destined to the plasma membrane.
In Vitro Translation and N-Glycosylation of RhCG Pro-tein-To assess ER translocation and early processing events during biosynthesis, RhCG was translated in vitro, in either the absence or presence of CPMM, and analyzed by SDS-PAGE. In the absence of CPMM, in vitro translated RhCG, with or without a c-Myc tag, migrated as a single band with a molecular mass less than the predicted molecular mass of 53 kDa (Fig. 11A). This mobility anomaly is also seen in erythroid homologues and is most likely the result of the highly hydrophobic nature of these proteins (36). In the presence of CPMM, the RhCG proteins exhibited a slower migration pattern suggesting that they may be differentially glycosylated (Fig. 11B). The observed size differences between the CPMM-treated and untreated protein products can be accounted by the size of a single complex N-glycan. These in vitro studies indicated that RhCG could be glycosylated. Western Blot and N-Glycanase Analysis of RhCG Protein-To establish the in vivo expression of RhCG as a glycoprotein, membrane proteins were isolated from HEK293 cells stably transfected with the RHCG.myc/His construct. Western blots probed with the RhCG C-tail polyclonal antiserum showed a major protein band with a molecular mass of ϳ58 kDa (Fig. 12A, lanes 3 and 4); this size was slightly larger than that of in vitro glycosylated RhCG (Fig. 11B, lane 4). After N-glycanase treatment, the size of membrane-bound RhCG was decreased to 46 kDa (Fig. 12A, lanes 5 and 6), which was equivalent to that of in vitro translated but unglycosylated RhCG (Fig. 11A, lane 6). This result indicated that the same AUG initiation codon could be used for RhCG synthesis in both in vivo and in vitro. Because the stably expressed RhCG pro- tein was expected to carry a Myc/His tag at the C-terminal end, anti-Myc monoclonal antibody was used to probe the same Western blots. This analysis revealed the same banding pattern (Fig. 12B), confirming that RhCG exists as a membrane glycoprotein of which the N-glycan is most probably attached to the 48 NLS 50 sequon on exoloop 1 (Fig. 2). DISCUSSION In the present study, we have isolated several new Rh genes and performed detailed characterization of the human RHCG and mouse Rhcg genes encoding a novel pair of polytopic plasma membrane glycoproteins. We showed that the RhCG and Rhcg cDNAs are highly conserved with regard to their nucleotide and derived protein sequences and membrane topologic organization. These results firmly establish the genuine orthologous relationship between the two homologues, suggesting that they possess the same functional role(s) in both human and mouse nonerythroid tissues. Clustal analyses of known Rh homologues define RhCG and Rhcg as the first new members of the nonerythroid Rh subfamily in vertebrate species, including all mammals. The detailed studies concerning the genetic and biochemical aspects of RhCG/Rhcg revealed common features as well as distinct differences between the erythroid and nonerythroid Rh homologues. Collectively these results provide novel insights into the molecular evolution, structural conservation, tissue-specific expression, and possible biological function of the Rh family of genes and proteins.
The many homologous sequences assembled suggests the existence of a Rh superfamily, which to date consists of five and four discrete gene members in humans and mice, respectively. Based on their genomic synteny, sequence identity, and tissue specificity, the human and mouse homologues can be divided into four orthologous groups, i.e. two erythroid gene pairs (RHAG/Rhag and RHCED/Rhced) and two nonerythroid gene pairs (RHCG/Rhcg and RHBG/Rhbg). Notably, all the RH loci harbor single copy genes, except for human RHCED (37), and all have a different map location, although each orthologous group shows comparable chromosomal synteny. By contrast, there is only one Rh gene in the unicellular organism slime mold (2), two tightly linked copies in C. elegans (15), and three discrete members in the zebrafish. 2 These observations suggest that both the erythroid and nonerythroid branches came into existence before mammalian radiation and that their multiplicity occurred mainly via intergenic translocation events. This pattern of RH duplication is likely to have produced functional novelty in a different temporal and spatial context within the organism, given changes in both coding sequence and regulatory elements of the gene (5-7, 10 -12, 32-34).
Compared with erythroid homologues, both primitive and nonerythroid members, including RhCG and Rhcg, are structurally much closer to RhAG/Rhag than they are to RhCED/ Rhced. By excluding the more distant RhCED/Rhced group, a membrane fold has been identified, which is characterized by shared signatures conserved among human/mouse RhCG/ Rhcg, RhBG/Rhbg, and RhAG/Rhag pairs. This relationship conforms to the approximate equidistance between the erythroid and nonerythroid branches and extends the observation that the RHCED series originated from an Rhag ancestor gene but evolved at a much increased rate (11,38). The slower divergence of RHAG and the faster evolution of RHCED imply an adapted functional specification that pertains to Rh complex assembly in the RBC membrane through heteromeric interactions. These interactions are likely mediated by both N-and C-terminal domains and involve hydrophobic contacts between RhAG and RhCED (39 -41). Without RhAG expression, as observed in the regulator type of Rh null disease (40,41), the RhCE/D proteins and their associated antigens are not dispositioned at the cell surface. In sharp contrast, such heteromeric interaction is not a prerequisite for surface expression of RhCG, as RhCG could by itself reach the plasma membrane, whether transfected into homologous or heterologous cells. Most likely, nonerythroid Rh homologues possess intrinsic topogenic signal sequences that can direct their own intracellular transport and plasma membrane destination. Further studies are required to determine whether RhCG/Rhcg is on apical or basal membrane in the cells.
Besides their capability of being expressed in heterologous cells, RhCG and Rhcg differ from erythroid members in the following features. 1) They each have a larger exoloop 6 and much longer cytoplasmic C-terminal sequence that is rich in Pro and Ser residues. This novel C-terminal domain may be involved in functional modulation of RhCG or Rhcg through interactions with some yet-to-be recognized intracellular proteins. 2) RHCG/Rhcg is unique, having a CpG island-rich promoter and an extra 3Ј-UTR exon, suggesting a different mode in controlling tissue-specific expression. 3) The human RHCG gene is mapped at 15q25 and close to the type I tyrosinemia disease locus (42), although its phenotypic relationship with the latter is unknown. 4) The onset of Rhcg expression occurs very late in development, contrary to the early coexpression of Rhag and Rhced that parallels erythroid differentiation at embryonic stages (12). 5) Compared with the erythroid restriction of Rhag/Rhced (12), RHCG or Rhcg is abundantly and broadly expressed in the kidney and in the testis. 6) Although residing in the plasma membrane, as erythroid Rh proteins, Rhcg is concentrated in the epithelial linings of the tubular structures, the kidney collecting tubules, and the testis seminiferous tubules. These features suggest new testable hypotheses as to the function of RhCG/Rhcg, possibly acting as a previously unrec- ognized transporter across the plasma membrane of epithelial cells.
Rh proteins have long been thought to possess a transport activity due to their characteristic TM fold and associated morphological and physiological changes in Rh null disease (2,7). However, their ligand(s) has not been identified, and their functional dependence on the assembly of a heteromeric complex between RhAG and RhCED remains to be established. Prior studies suggested that Rh might function as an ATP-dependent phosphatidylserine transporter (phosphatidylserine flipase) of the RBC membrane (43, 44). Nonetheless, two lines of evidence disfavor this proposal: 1) Rh proteins lack the ATP-binding cassette (ABC) (2, 7); and 2) Rh null cells have a normal phosphatidylserine transport activity (45).
Recently, a relationship between erythroid Rh proteins and NH 4 ϩ transporters of the Mep/Amt family has been suggested based on marginal sequence identity (11.1-17.7%) (9). RhCG/ Rhcg and other nonerythroid homologues lie in the same range of identity with these transporters. Although NH 4 ϩ transporters are found in all three domains of life (9), Rh family members occur only in Eucarya (2). In Eucarya, Mep/Amt members occur in yeast (46), plants (47), and nematodes (14), but not other animals; in contrast, Rh homologues are present in animals (10) and slime mold (2) but absent in yeast and plants, except algae. 2 The sequence connection, along with the tissue specificity, raises the possibility that RhCG and Rhcg occur as potential NH 4 ϩ transporters in kidney and testis. However, it should be noted that the homology between RhCG/Rhcg and Mep/Amt proteins is quite dispersed and only distantly related. This evolutionary distance could be large enough to endow RhCG/Rhcg with a pivotal structural difference in functional specification. Furthermore, evolution economy also argues against the coexistence of two Rh homologues and three Mep/ Amt proteins as the same functional duplicates that all transport NH 4 ϩ in C. elegans (14). Finally, the topology of RhCG/ Rhcg resembles numerous 12-TM transporters lacking an ABC domain (8) but differs from that of Mep/Amt members, which generally contain 10 TM helical domains (46 -48). If the spatially conserved amino acids are to form a binding site for the amino group only, RhCG or Rhcg may serve to transport some novel amines rather than ammonium itself. The conserved structure and abundant expression suggest that RhCG and Rhcg play an essential role in maintaining normal homeostatic conditions in kidney and testis. In future work, it will thus be important to examine the above hypotheses using models amenable to biochemical and physiological manipulations. To this end, the transient and stable expression systems described here should serve as useful tools for the functional identification of both erythroid and nonerythroid Rh family proteins.