Cloning and Characterization of a Novel Human Phosphatidylinositol Transfer Protein, rdgBβ*

The various PITP, retinal degeneration B (rdgB), and amino-terminal domain interacting receptor (Nir) phosphatidylinositol transfer proteins can be divided into two structural families. The small, soluble PITP isoforms contain only a phosphatidylinositol transfer domain and have been implicated in phosphoinositide signaling and vesicle trafficking. In contrast, the rdgB proteins, which include Nir2 and Nir3, contain an amino-terminal PITP-like domain, an acidic, Ca2+-binding domain, six putative transmembrane domains, and a conserved carboxyl-terminal domain. However, the biological function of rdgB proteins is unclear. Here, we report the isolation of a cDNA encoding a novel rdgB protein, mammalian rdgBβ (MrdgBβ). The 38-kDa MrdgBβ protein contains an amino-terminal PITP-like domain and a short carboxyl-terminal domain. In contrast to other rdgB-like proteins, MrdgBβ contains no transmembrane motifs or the conserved carboxyl-terminal domain. Using Northern and reverse transcription-polymerase chain reaction analysis, we demonstrate that MrdgBβ mRNA is ubiquitously expressed. Immunofluorescence analysis of ectopic MrdgBβ showed cytoplasmic staining, and the ability of recombinant MrdgBβ to transfer phosphatidylinositolin vitro was similar to other PITP-like domains. Although early reports found functional degeneracy in vitro, the identification of a fifth mammalian PITP-like protein with a unique domain organization and widespread expression supports more recent results that suggest that different PITP-like domains have distinct functions in vivo.

Further support for an important in vivo role for specific PtdIns transfer proteins came from the characterization of Drosophila rdgB (DrdgB) mutations. The DrdgB protein has been reported to be a membrane-bound PtdIns transfer protein, which has been exclusively implicated in retinal and olfactory neurosensory signaling (13,14). DrdgB is a 160-kDa protein containing an acidic, Ca 2ϩ -binding domain, six putative membrane-spanning regions, and a carboxyl-terminal domain. The amino-terminal 281 amino acids of DrdgB share Ͼ40% identity with PITP␣. Although transfer activity still remains to be demonstrated for the full-length protein, the PITP-like domain of DrdgB does possess PtdIns and phosphatidylcholine transfer activity in vitro (15).
DrdgB mutations were originally identified by defects in the compound eye: null mutations cause light-induced retinal degeneration and abnormal termination of the light response (16 -18). A combination of genetic, biochemical, and electrophysiological evidence indicates that DrdgB plays a critical role in the phospholipase C-dependent phototransduction cascade in Drosophila, both downstream of phospholipase C and in the recovery phase of the light response (19 -25). Nevertheless, the exact biochemical role of DrdgB remains to be determined. Interestingly, although the expression of the PITP-like domain of DrdgB was sufficient for complete rescue of specific DrdgB mutants, PtdIns transfer activity alone appears not to be sufficient, as PITP␣ was unable to rescue the same mutants (26).
A mammalian homologue of DrdgB has recently been cloned and termed mammalian rdgB␣ (MrdgB␣) (27)(28)(29). Unlike PITP␣, expression of MrdgB␣ in DrdgB mutant flies was sufficient to completely restore the wild type phenotype, suggesting that a biochemical activity required for invertebrate phototransduction has been conserved by the Drosophila and mammalian proteins (27). Using a yeast two-hybrid approach to screen for proteins interacting with the protein tyrosine kinase PYK2, Lev et al. (30) identified MrdgB␣ and two novel human rdgB proteins. Because all three proteins bound the amino-terminal domain of PYK2, they have been designated PYK2 amino-terminal domain interacting receptor (Nir) proteins. According to this nomenclature, which we employ in this report, Nir2 corresponds to MrdgB␣.
Like DrdgB, the Nir proteins have a multiple domain structure, containing an acidic, Ca 2ϩ -binding domain, six putative * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  transmembrane domains, and a carboxyl-terminal domain. Both Nir2 and Nir3 contain an amino-terminal PITP-like domain; however, this domain is absent in Nir1. Although Nir1-3 mRNAs possess different tissue expression patterns, all three are abundantly expressed in the brain and retina. Lev et al. (30) also demonstrated that Nir proteins form a complex with PYK2 via their carboxyl-terminal domain, leading to their tyrosine phosphorylation in brain tissue and cultured cell lysates. The authors therefore postulated that Nir proteins function in concert with PYK2 in the regulation of Ca 2ϩ and phosphoinositide-dependent pathways.
Here, we report the identification of a cDNA encoding a novel human rdgB protein, which we have provisionally termed MrdgB␤. Interestingly, unlike DrdgB and the Nir proteins, the predicted amino acid sequence of MrdgB␤ contains no recognizable transmembrane motifs. Furthermore, the absence of the carboxyl-terminal domain, which is present in the DrdgB and Nir proteins, suggests that MrdgB␤ does not interact with PYK2. We show that MrdgB␤ is a ubiquitously expressed, cytoplasmic protein that possesses a similar ability to transfer PtdIns compared with other PITP-related proteins.

EXPERIMENTAL PROCEDURES
Isolation of the MrdgB␤ cDNA-MrdgB␤ was originally detected as a human brain expressed sequence tag (EST) sequence (GenBank TM accession number R24545) in BLAST searches against the DrdgB sequence. The clone containing R24545 was obtained (IMAGE Consortium, Livermore, CA) and sequenced. To isolate sequences encompassing the initiation codon, 5Ј-rapid amplification of cDNA ends (RACE) PCR was performed using a DR2 fetal brain cDNA library (approximately 10 7 plaqueforming units/reaction; CLONTECH, Cambridge, UK) using antisense primers complementary to the 5Ј-region of R24545 (initial primer, 5Ј-AAAACTCGAGTGGC TCGTTCAAATTCTCG-3Ј; nested primer, 5Ј-AAAACTCGAGCCTGTCTATGTCCAATCA GC-3Ј) and Taq DNA polymerase (Promega, South Hampton, UK). A 700-base pair RACE product was isolated, subcloned into pGEM-T Easy (Promega) and sequenced. In order to confirm the integrity of the complete cDNA, the full-length cDNA, termed PCR-1, was amplified from the DR2 library (forward primer, 5Ј-ATATGAATTCTAATGCTGCTGAAAGAGTACCG-3Ј; reverse primer, 5Ј-ATATCTCGAGCCTCAGATTTGGGCCGACATGG-3Ј) and subcloned into pGEM-T Easy using the XhoI and EcoRI restriction sites. Several independent clones were isolated and sequenced in full to confirm the full-length MrdgB␤ cDNA sequence. (The nucleotide sequence for MrdgB␤ has been deposited in GenBank TM under accession number AF171102.) Sequence Determination and Analysis-All DNA sequencing was carried out using the dideoxy-chain termination reaction (PRISM Dye Deoxy Terminator Cycle Kit; Perkin-Elmer Biosystems) and an automated DNA sequencer (PRISM 377; Perkin-Elmer Biosystems). Sequence comparisons and multiple sequence alignments were performed using the BESTFIT and the PILEUP programs, respectively (version 7, Genetics Computer Group).
Northern Blot Analysis-A human multiple-tissue Northern blot (CLONTECH) was probed under high stringency conditions in accordance with the manufacturer's instructions with a 32 P-labeled cDNA fragment of MrdgB␤ encoding nucleotides 1-600, which was generated by PCR.
First Strand cDNA Analysis-Total RNA was prepared from different murine tissues using Trizol (Life Technologies) in accordance with the manufacturer's instructions. Total RNA was resuspended in DEPCtreated, 0.5ϫ saline/sodium phosphate/EDTA and heated to 68°C. Chilled samples were incubated with oligo(dT) cellulose, previously equilibrated with 0.5ϫ SSPE (3 M NaCl, 0.2 M NaH 2 PO 4 , 0.2 M EDTA), for 20 min at room temperature. Slurries were loaded onto Wizard columns (Promega) and washed four times with 0.5ϫ SSPE. The poly(A) ϩ fraction was eluted with DEPC-treated H 2 O at 70°C and precipitated using ethanol and 2 g of glycogen carrier. 1 g of poly(A) ϩ RNA was reverse transcribed using a Superscript reverse transcription (RT) PCR kit (Life Technologies) in accordance with the manufacturer's instructions. The quality of the cDNA was tested by the amplification of residues 106 -332 of glyceraldehyde-3-phosphate dehydrogenase. 5% of each first-strand cDNA synthesis reaction was used as a PCR template using primers (forward primer, 5Ј-ATGCTGCTCAAGGAGTAC-CGAATC-3Ј; reverse primer, 5Ј-GATGGTGTATGGGTAA TAATTCC-3Ј) derived from a murine EST (GenBank TM accession number AI019450). The predicted amino acid sequence of this EST is identical to residues 1-91 of MrdgB␤. The PCR products were subcloned into pGEM-T Easy and verified by sequence analysis.
Cell Culture and Transfection-Human embryo kidney HEK293 cells were cultured in Dulbecco's modified Eagle's medium containing Glutamax (Life Technologies) and 10% fetal calf serum, 50 IU/ml penicillin and 50 g/ml streptomycin. Cells were transfected with expression plasmids in 100-mm dishes containing glass coverslips using Superfect (Qiagen) according to the manufacturer's instructions. Cells were harvested 48 h after the addition of DNA.
Immunofluorescence Analysis-Transfected HEK293 cells grown on glass coverslips were fixed with 4% paraformaldehyde in phosphatebuffered saline for 20 min at room temperature. Fixed cells were permeabilized in phosphate-buffered saline containing 0.2% Triton X-100 for 5 min. Cells were incubated for 1 h at room temperature with M2 anti-FLAG monoclonal antibody (Sigma) diluted 1:360 in phosphate-buffered saline containing 0.1% bovine serum albumin. Secondary staining was performed using FITC-labeled anti-mouse antibodies (Jackson Immunoresearch Laboratories, West Grove, PA) diluted 1:200 in phosphatebuffered saline. Actin filaments were detected by incubation with 0.8 nM TRITC-labeled phalloidin (Sigma) in order to identify untransfected cells. Stained samples were mounted in Mowiol (Calbiochem) and analyzed by confocal laser scanning microscopy (Zeiss LSM 510).
Expression and Purification of Recombinant Protein-A PCR fragment from pGEX-KG (32) encoding glutathione S-transferase (GST) and containing the NdeI site at the beginning of the open reading frame, was subcloned into the NdeI and XhoI sites in pET21b (Novagen, Cambridge, UK) and designated pET21-GST. The PCR-1 cDNA fragment was subcloned into the EcoRI and XhoI sites of pET21-GST and sequenced. Recombinant protein expression was induced in Escherichia coli strain BL21(DE3)pLysS (Novagen) using 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h at room temperature. Bacteria expressing the GST-MrdgB␤ fusion protein were sonicated in lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM KCl, 0.5% Triton X-100, 1 mM EDTA, 1 mM EGTA, 0.1% ␤-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 1 mM leupeptin, 1 mM aprotinin) and centrifuged at 20,000 ϫ g for 20 min. Supernatants were incubated for 30 min at 4°C with glutathione-Sepharose (Amersham Pharmacia Biotech), washed extensively with lysis buffer without protease inhibitors, and resuspended in 80 mM Tris-HCl, pH 7.9, 2 mM NaCl. Bovine thrombin was added at a concentration of 25units/mg of fusion protein, and cleaved protein was collected following centrifugation. In order to obtain a similar buffer composition to His 6 -tagged protein preparations, EDTA was added to a final concentration of 400 mM. Recombinant GST was purified essentially as above, with the exception that GST was eluted from the Sepharose using reduced glutathione.
PITP␣ and PITP␤ isoforms were cloned and expressed as His 6 fusion proteins in bacteria as described previously (10). The PITP-like domain of human Nir2 was obtained using PCR with Vent polymerase (New England Biolabs, Hitchin, UK) from the pDR2 human infant brain cDNA library (forward primer, 5Ј-AAAACATATGCTCATCAAGGAAT-ACCAC-3Ј; reverse primer, 5Ј-AAAACTCGAGCTCGGTGCTCGGTT-TCCC-3Ј). Products were treated with Taq polymerase and subcloned into pGEM-T Easy. Several independent clones were sequenced to check for mutations before insertion into pET21a (Novagen) using NdeI and XhoI restriction sites.
Transfer Assays-PtdIns transfer activity was assayed using rat liver microsomes and [ 3 H]PtdIns as described previously (3) and either equal concentrations of each protein (24 g/ml) or corresponding buffer to assess background counts.

RESULTS
Cloning of MrdgB␤-In order to isolate novel human homologues of DrdgB, EST data bases were screened for human sequences showing similarity to DrdgB. A human brain EST fragment (R24545) was found, the complete sequence of which encoded amino acid residues 210 -333 of the final open reading frame, as well as 1782 base pairs of the 3Ј-untranslated region. Using PCR primers designed to anneal either side of the stop site in R24545, a fragment of the expected size was amplified from an infant brain cDNA library. The sequence of this fragment confirmed the position of the stop site in R24545. Furthermore, several EST sequences (GenBank TM accession numbers AA035468, H17821, AA588404, and AA807607) identical to the 3Ј-region of R24545 were identified by performing additional BLAST searches. The position of the stop site in these ESTs was identical to that observed in R24545. In order to obtain the sequence encoding amino acid residues 4 -209, 5Ј-RACE PCR was used to screen the infant brain cDNA library. The remaining cDNA sequence encoding residues 1-3 was derived from the EST data base by screening for sequences homologous to the cDNA sequence encoding residues 4 -8. Three partial cDNA sequences (GenBank TM accession numbers AA021507, H86340, and AA808293) were identified from human tissue. The complete open reading frame was verified by PCR using the infant brain cDNA library. The closer similarity of the predicted amino acid sequence to rdgB compared with PITP isoforms (see below) suggested that this sequence defined a novel human rdgB protein, which we have therefore termed mammalian rdgB␤ (MrdgB␤).
MrdgB␤ has an open reading frame of 999 base pairs, which encodes a 333-amino acid polypeptide of molecular mass 38.2 kDa (Fig. 1). MrdgB␤ contains a PITP-like amino-terminal domain and a small carboxyl-terminal domain that exhibits no sequence homology to the Ca 2ϩ -binding and the conserved carboxyl-terminal domains of the Nir1-3 ( Fig. 2A). Interestingly, unlike the DrdgB and Nir proteins, the predicted protein sequence of MrdgB␤ contains no recognizable transmembrane regions.
MrdgB␤ exhibits 58 and 40% identity with two Drosophila ESTs (GenBank TM accession numbers AA439582 and AA698247), both of which are distinct from DrdgB. Consequently, from here onward, we refer to DrdgB as DrdgB␣. The predicted amino acid sequence of the first EST shows a greater level of identity with MrdgB␤ than with DrdgB␣, Nir1, or Nir2, thereby indicating the existence of DrdgB␤. We therefore suggest that rdgB␤ proteins occur in mammals and insects, al- though further analysis is required to establish the full-length DrdgB␤ sequence. The second Drosophila EST sequence is more similar to mammalian PITPs than to any protein in the rdgB family, and is therefore termed Drosophila PITP (PITP-DM in Fig. 2B). The genome of Caenorhabditis elegans appears to contain only one PITP-like (Wormpep accession number Y71G12A_205.C (produced by the C. elegans Sequencing Group at the Sanger Center)) and one rdgB-like gene (GenBank TM accession number Z77131). The probable functions of the Drosophila and C. elegans PITPs are unclear from sequence comparisons, as they are less similar to either mammalian PITP isoform than the latter are to each other (Fig. 2B). The pre-dicted C. elegans rdgB protein (CrdgB) has a similar domain organization to DrdgB␣, Nir2, and Nir3 and may therefore bind a PYK2-like protein.
Analysis of the deduced MrdgB␤ protein sequence using the MOTIFS algorithm (GCG) revealed several potential protein kinase A and protein kinase C phosphorylation sites, suggesting possible mechanisms of functional regulation. Furthermore, in common with all proteins containing a PITP-like domain, rdgB␤ contains a threonine residue corresponding to residue 59 of PITP␣, which has been suggested to allow protein kinase C to regulate the PtdIns transfer activity of PITP␣ (33).
Tissue Distribution of MrdgB␤-Northern analysis of multi-

FIG. 2. Comparisons of amino acid sequences of PITP-related proteins.
Amino acid sequences of the PITP-like domains of the five human and three Drosophila (DrdgB␣, DrdgB␤, and PITP-Dm) PITP-related proteins are compared using multiple sequence alignment (A) and a dendrogram (B). *, DrdgB␤ and PITP-Dm are incomplete sequences derived from EST data bases (see text for details); consequently, complete sequence information on these two proteins is required to confirm this analysis. Positions containing identities between Ͼ4 sequences are shown in black boxes, and remaining positions containing similarities between Ͼ3 sequences are shown in gray boxes. The consensus sequence is derived from identities in all 8 sequences (uppercase) or similarity in Ͼ3 sequences (lowercase). ple human tissues indicated that MrdgB␤ is ubiquitously expressed (Fig. 3A). The 2.0-kilobase MrdgB␤ transcript was expressed strongly in heart, muscle, kidney, liver, and peripheral blood leukocytes and weakly expressed in all other tissues. In comparison, DrdgB␣ shows multiple transcripts ranging from 3.9 to 9.5 kilobases with expression limited to the brain and retina, and Nir2 is ubiquitously expressed as a transcript of 4.5 kilobases, whereas Nir1 and Nir3 exhibit more limited expression patterns with transcripts of around 7.5 kilobases (15,30). The size of the MrdgB␤ transcript is consistent with the absence of sequence encoding the transmembrane domain, which is present in the DrdgB␣, CrdgB, and Nir proteins. RT-PCR analysis of cDNA from various mouse tissues confirmed the ubiquitous expression of MrdgB␤ (Fig. 3B). The detection of MrdgB␤ transcripts in all tissues analyzed is consistent with the presence of MrdgB␤ ESTs derived from a variety of human tissues (data not shown).
Subcellular Localization of MrdgB␤-In order to define the intracellular localization of MrdgB␤, transient expression of FLAG-tagged MrdgB␤ in HEK293 cells was assessed using immunoprecipitation with an anti-FLAG monoclonal antibody followed by Western blot analysis of the precipitates using the same anti-FLAG antibody. Two proteins of similar size (approximately 40 and 48 kDa) were detected, possibly due to degradation and/or posttranslational modification (Fig. 4A). Chang et al. (27) demonstrated that a Nir2-specific antibody also recognized two proteins of similar size using Western blot analysis of retinal samples. Transfected cells were stained with anti-FLAG antibody and analyzed by confocal immunofluorescence microscopy. Immunoreactive cells revealed that the FLAG-tagged protein was diffusely present throughout the cytoplasm (Fig. 4B). No detectable staining of nontransfected cells was observed (Fig. 4C).
Transfer Activity-The presence of a conserved PITP-like domain suggested that MrdgB␤ may possess PtdIns transfer activity. To assess the PtdIns transfer activity of MrdgB␤, we ex-pressed the full-length protein and the PITP-like domain in bacteria (Fig. 5A). The ability of the recombinant proteins to transfer rat liver microsomal [ 3 H]PtdIns to liposomes in vitro was compared with PITP␣, the PITP-like domain of Nir2, and GST (Fig. 5B). RdgB␤ exhibited PtdIns transfer activity that was comparable to all of the other PITP-related proteins. Both fulllength MrdgB␤ and the PITP-like domain mediated a robust transfer of PtdIns between the bilayers, with 19.4 Ϯ 3 and 16.6 Ϯ 2%, respectively, of the total counts transferred. Likewise, the PITP domain of Nir2, PITP␣, and PITP␤ transferred between 16.7 and 25% Ϯ 2% of the total [ 3 H]PtdIns. In contrast, no transfer activity was detected using the GST protein alone. DISCUSSION Although the first rdgB protein to be identified was an invertebrate phototransduction protein, the conservation of amino acid sequence and domain topology between DrdgB␣, CrdgB, Nir2, and Nir3 suggests that the functions of the rdgB family, like the PITP family, have been conserved during metazoan evolution. These functions have so far been only partially characterized and are discussed later.
We have isolated the cDNA of a novel PtdIns transfer pro- tein. Sequence alignment of its conserved PITP-like domain revealed that, although it is a member of the rdgB family, it contains no apparent membrane-spanning domains, nor the domain required for interaction with PYK2. Accordingly, we have provisionally adopted the name rdgB␤. The existence of five mammalian proteins with PITP-like domains suggests that there are differences in their cellular functions. Furthermore, as the MrdgB␤ protein defines a novel structural form, it may have a novel biological function. The differential subcellular distribution of PITP isoforms (11,34), the properties of the mouse vibrator mutant (12), DrdgB␣ rescue studies (26), and the co-expression of Nir and PITP isoforms within mammalian tissues (1,30) suggest that different PITP domains are not functionally degenerate in vivo, although recombinant PITP proteins can behave similarly in vitro (8,10,35).
In addition to tissue distribution, the roles of different rdgB isoforms in vivo are likely to be defined by their intracellular localization and cognate binding partners. Although the intracellular localization of the Nir1 and 3 proteins has yet to be addressed, Nir2 has been found in Golgi and endoplasmic reticulum membranes (36). DrdgB␣ is also localized to the retinal endoplasmic reticulum (subrhabdomeric cisternae) (15,37). We show that ectopically expressed MrdgB␤ is present throughout the cytoplasm, although it is possible that ectopic expression influences the intracellular localization of a protein, for example, if a binding partner such as Nir1 is required for appropriate localization. In this case, binding may be saturated by overexpression or masked by the carboxyl-terminal epitope tag. Thus, the intracellular localization of endogenous MrdgB␤ will be addressed in future studies. An association between Nir1, which lacks a PITP-like domain, and MrdgB␤ would form a heterodimer containing all of the domains present in other rdgB proteins. However, the restricted tissue distribution of Nir1 suggests that if such an association occurs in vivo, it would not be ubiquitous.
The demonstration that the conserved carboxyl-terminal domain of each Nir protein forms a complex with PYK2 and is tyrosine-phosphorylated in response to PYK2 activation led to the suggestion that PYK2 is an upstream regulator of Nir proteins (30). The absence of this conserved carboxyl-terminal domain in MrdgB␤ suggests that this isoform of rdgB does not interact directly with PYK2. Indeed, we have been unable to detect PKY2 in immunoprecipitates of FLAG-tagged MrdgB␤ from transfected HEK293 cells. 2 Studies of dominant negative DrdgB␣ mutants in Drosophila suggest that DrdgB␣ associates with at least one other protein in vivo (26). Whether or not any binding partners in addition to PYK2 exist remains to be demonstrated. The future identification of proteins that interact with MrdgB␤ not only will assist the characterization of rdgB function but also may provide insight into regulatory mechanisms.
As recombinant MrdgB␤ exhibits PtdIns transfer activity in vitro, its function may be to mobilize PtdIns. On the other hand, as MrdgB␤ and PITP␣ are both expressed ubiquitously within the cytoplasm, their lipid binding and transfer specificity may be expected to differ. Consequently, MrdgB␤ may also bind and transfer alternative phospholipids. Indeed, it remains possible that phospholipid binding, rather than transfer activity, is critical to MrdgB␤ function. In this regard, mutation studies suggest that PtdIns transfer is not the only essential activity of DrdgB␣ (26).
In addition to the PITP-like domain, MrdgB␤ contains a small carboxyl-terminal domain, which has no apparent affect on the ability of MrdgB␤ to transfer PtdIns. Although this domain contains acidic residues, it exhibits no significant homology to the acidic Ca 2ϩ -binding domains of the Nir proteins. Because the sequence of MrdgB␤ diverges from other PITPrelated sequences at the start of this domain, the function of this unique carboxyl-terminal sequence is of particular interest to future studies.
Although it has been proposed that DrdgB␣ may perform functions specific to rapid neurosensory transmission in invertebrates (15,26), the occurrence and expression of Nir proteins throughout the central nervous system and other tissues suggests that additional roles exist for these proteins in vivo. We have demonstrated here that a novel member of the rdgB family is expressed ubiquitously, an observation that extends the role of this family toward more fundamental cellular processes. Future work is aimed at identifying specific in vivo roles for MrdgB␤.