Cloning and Characterization of a Novel Apolipoprotein A-I Binding Protein, AI-BP, Secreted by Cells...

Articledoi:10.1006/geno.2002.6761, available online at http://www.idealibrary.com on IDEAL

Cloning and Characterization of a Novel Apolipoprotein A-IBinding Protein, AI-BP, Secreted by Cells of the Kidney

Proximal Tubules in Response to HDL or ApoA-IMirko Ritter,1 Christa Buechler,1 Alfred Boettcher,1 Stefan Barlage,1 Anna Schmitz-Madry,1Evelyn Orsó,1 Salim Maa Bared,1 Gerno Schmiedeknecht,1 Carsten H. Baehr,2 Gert Fricker,2

and Gerd Schmitz1,*

1Institute for Clinical Chemistry and Laboratory Medicine, University of Regensburg, D-93042 Regensburg, Germany2Institute of Pharmaceutical Technology and Biopharmacy, University of Heidelberg, D-69120 Heidelberg, Germany

*To whom correspondence and reprint requests should be addressed. Fax: (49)-941-944-6202. E-mail: [email protected].

Apolipoprotein A-I (apoA-I) is the major apolipoprotein of high-density lipoproteins (HDL)and has an important role in the regulation of the stability, lipid transport, and metabolismof HDL particles. To identify novel proteins that are involved in HDL metabolism, we usedmature apoA-I (amino acids 25–267) as a bait for the screening of a human liver two-hybridcDNA library. Among the identified genes, several encoded known proteins, including serumamyloid A2a (SAA2a), apoC-I, and phosphodiesterase HCAM1 (PDE1A), found to interact withapoA-I. In addition, we have cloned a novel 29 kDa apoA-I interacting protein, which wenamed AI-BP (apoA-I binding protein). The AI-BP encoding gene, APOA1BP, which is locatedon chromosome 1q21, is composed of six exons and five introns and spans 2.5 kb. Northernblot analysis demonstrated ubiquitous expression of the APOA1BP mRNA with the highestexpression in kidney, heart, liver, thyroid gland, adrenal gland, and testis. AI-BP protein isnot detectable in serum of healthy probands, but serum samples of patients with septic syn-dromes may contain elevated levels of AI-BP. Significant amounts of AI-BP protein are foundin cerebrospinal fluid and urine of healthy probands. The stimulation of cells derived fromthe kidney proximal tubules with apoA-I or HDL induces a concentration-dependent secre-tion of AI-BP indicating an important role for AI-BP, in the renal tubular degradation orresorption of apoA-I.

Key Words: HDL, apolipoprotein A-I, AI-BP, kidney proximal tubules, urine, CSF, sepsis

INTRODUCTION

High density lipoproteins (HDL) have a major role inreverse cholesterol transport to remove excess cholesterolfrom extrahepatic cells, providing a protection against thedevelopment of atherosclerosis in the arterial wall. HDL par-ticles containing apoA-I are the most efficient acceptors ofcellular cholesterol [1]. The reverse cholesterol transportpathway includes the ABCA1 and lecithin:cholesterol acyl-transferase (LCAT) catalyzed maturation of nascent pre�-HDL to spherical, circulating �-HDL particles [2]. In accor-dance with this, LCAT deficiency results in an accumulationof pre�-HDL and a decrease of �-HDL particles [3]. ABCA1deficiency is associated with a severe reduction of plasma

GENOMICS Vol. 79, Number 5, May 2002Copyright © 2002 Elsevier Science (USA). All rights reserved.0888-7543/02 $35.00

HDL levels and an increased catabolism of apoA-I contain-ing HDL particles [4].

Circulating �-HDL bind to cell surface receptors such asthe scavenger receptor SR-BI and cholesterol is taken up bythe liver and excreted into bile [5]. In addition, the kidneymay be involved in the clearance of pre�-HDL particles viathe megalin/cubilin receptor system expressed in the proxi-mal tubules. The megalin/cubilin complex has been shownto function as an endocytic receptor for HDL [6,7] and it isassumed that apoA-I containing HDL or the HDL-associatedapolipoproteins can be targeted by two different ways in thecells of the proximal tubules. One way results in the lysoso-mal degradation of HDL and its apolipoproteins, another waymay be similar to a transcytosis or retroendocytosis pathway,

693

Article doi:10.1006/geno.2002.6761, available online at http://www.idealibrary.com on IDEAL

l domains of

in which endocytosis of HDL particles is fol-lowed by deloading of cholesterol and resecre-tion of lipid-depleted particles. In addition toSR-BI and megalin/cubilin, other receptors likeHBP (HDL-binding protein) [8] or HB2 (HDL-binding protein 2) [9] are involved in the metab-olism of apoA-I containing HDL particles.�-HDL also provides cholesterol to the placentaand to steroidogenic tissues including the adre-nal glands, ovaries, and testis for steroid biosyn-thesis. ApoA-I containing lipoproteins are alsofound in the central nervous system (CNS) [10].It is assumed that these particles are derivedfrom small HDL-like particles of the plasmacompartment, which enter the CNSparenchyma via the blood–brain barrier and areinvolved in cholesterol efflux from neuronalcells.

ApoA-I as the major apolipoprotein of HDLhas an important role in the regulation of thestructure, stability, lipid transport, and metabo-lism of HDL particles [11]. ApoA-I is synthe-sized as a pre-proprotein of 267 amino acids.After cleavage of the 18-amino-acid signal pep-tide, the proprotein is secreted into the plasma,where the 6-amino-acid propetide is removedby cleavage through a metalloprotease formingthe 243-amino-acid mature apoA-I [12]. Severa
apoA-I were identified, which mediate different functions ofapoA-I. Thus, the amino-terminal helix comprising the aminoacid residues 44–65 [13] and the carboxy-terminal helix ofamino acids 210–241 [14] are recognized to be responsible forthe initial interaction of apoA-I with lipids. The N-terminaldomain of the mature apoA-I contains a thyroid hormonebinding motif (amino acids 18–22) [15]. The LCAT activationproperties of apoA-I depend on a central domain including theamino acids 144–165 [16]. As the central region of apoA-I mayalso be involved in the binding of cholesterol to HDL, thisdomain may serve to present cholesterol to LCAT. Several nat-ural variants have been described that display reduced LCATactivation properties and most of them correspond to muta-tions in the central domain, amino acids 144–186 (R151C,V156E, L159R, P165R, R173C) [16]. The C-terminal domain ofapoA-I is also involved in the maintenance of HDL in the cir-culation and in the promotion of lipid efflux. The deletion ofC-terminal residues 201–243, 217–243, or 226–243 increasedthe rates of apoA-I catabolism [17]. An apoA-I mutant(Glu235→0) is associated with low HDL-cholesterol levels dueto an increased apoA-I catabolism [18]. Considering this struc-ture–function relationship of apoA-I, it is reasonable to assumethat mutations within these domains might affect HDL metab-olism and HDL subclass distribution.
To identify proteins that interact with apoA-I and areinvolved in the regulation of apoA-I metabolism, we used themature form of apoA-I (amino acids 25–267) as a bait in a yeasttwo-hybrid screening. Among the identified candidates,

694

TABLE 1: Analysis of yeast two-hybrid interactions

GAL4 DNA binding GAL4 activation His3 activation lacZ activationdomain hybrid domain hybrid

apoA-I apoC-I ++ ++

lamin C apoC-I _ _

BD apoC-I _ _

apoA-I SAA2� +++ +++

lamin C SAA2� _ _

BD SAA2� _ _

apoA-I Hcam1 ++ +++

lamin C Hcam1 _ _

BD Hcam1 _ _

apoA-I AI-BP + ++

lamin C AI-BP _ _

BD AI-BP _ _

apoA-I AD _ _Yeast Y190 cells were co-transformed with expression vectors encoding the indicated hybrid proteins.Induction of reporter genes by interacting hybrid proteins was analyzed by the extent of growth on histidinedeficient medium (His3 activation) and by a �-galactosidase assay (lacZ activation)._ , no induction, + , low induction; ++, medium induction; +++, strong induction of the reporter gene. BD, GAL4 DNA binding domain (without an additional fusion protein); AD, GAL4 activation domain (without an additional fusion protein).

several known proteins including serum amyloid A2�, apoC-I, and phosphodiesterase HCAM1 (PDE1A) have been foundto interact with apoA-I. In addition, we cloned a novel apoA-I interacting protein, which we named AI-BP (apoA-I bindingprotein). The cloning and characterization of this novel protein are the main focus of this report.

RESULTS AND DISCUSSION

Identification of ApoA-I Interacting Proteins Using theYeast Two-Hybrid SystemTo identify proteins that interact with apoA-I we used theyeast two-hybrid system. We generated a fusion construct ofthe GAL4 DNA-binding domain with the mature apoA-I pro-tein sequence (amino acids 25–267). After sequencing, thisconstruct was used as a bait for screening of a human livertwo-hybrid cDNA library. We isolated 30 clones that dis-played a Ade+His+LacZ+ phenotype, but only 4 of these clonesturned out to be true positives. The interaction of apoA-I withthese proteins was specific, as no interaction was observedbetween the identified proteins and the GAL4 DNA-bindingdomain (BD) alone or in fusion with an unrelated protein(lamin C; Table 1).

One positive yeast clone contained the complete cDNA ofhuman apolipoprotein C-I (apoC-I). ApoC-I is a constituentof triglyceride-rich chylomicrons and very low densitylipoproteins (VLDL), as well as high-density lipoproteins

GENOMICS Vol. 79, Number 5, May 2002Copyright © 2002 Elsevier Science (USA). All rights reserved.


(HDL). Another positive yeast clone encoded the completecDNA of serum amyloid A2� (SAA2�). Acute phase SAA isfound strongly associated with HDL particles and it mayaccount for 17–87% of total apolipoproteins present in acutephase HDL [19]. The fact that we identified two proteinswhich are known to be associated with HDL particles con-taining apoA-I is a good evidence that the two-hybrid systemis a suitable tool to analyze apolipoprotein interactions thatmay occur upon cellular processing of apoA-I.

A

FIG. 1. Analysis of AI-BP proteinsequence. (A) cDNA and protein sequenceof AI-BP. A putative signal peptide fortranslocation into the endoplasmic reticu-lum is underlined. Putative phosphoryla-tion sites are shown in bold and are under-lined. Potential O-glycosylation sites areindicated as shaded letters. (B) Alignmentof the human, murine, bovine, and porcineAI-BP protein sequence. The completesequence of the human and murine AI-BPprotein and the fragments of the bovineand porcine homologs are shown.Identical amino residues are shown in cap-ital letters.

B


In addition we identified one yeast clone that encoded aC-terminal fragment of human phosphodiesterase HCAM1(amino acids 212–535), which comprises most of the catalyticdomain of HCAM1. HCAM1 is a type I phosphodiesterasethat is activated by calmodulin in the presence of Ca2+ andcatalyzes the hydrolysis of cGMP and cAMP [20]. The inter-action of HCAM1 and apoA-I observed in the two-hybridassay was verified using recombinant HCAM1 and purifiedapoA-I by pull-down assays (data not shown). HCAM1 is anintracellular protein indicating a possible function of the inter-action of HCAM1 and apoA-I in the retroendocytosis pathwayof the HDL particle [21], a process that has been shown to bedependent on cAMP and Ca2+ [22]. As the interaction betweenapoA-I and HCAM1 might influence the enzymatic activity ofthe phosphodiesterase, it might modulate the concentration ofthe second messenger cAMP and thereby regulate the endo-cytosis and resecretion of apoA-I. Further experiments willreveal the effect of apoA-I on the enzymatic activity ofHCAM1 and the role of HCAM1 in the regulation of HDLmetabolism.

AI-BP: cDNA, Protein and Gene StructureOne positive yeast clone contained an unknown, 1120-bpcDNA (Fig. 1A). There is a poly(A) tail at the 3�-site of theidentified sequence indicating that the 3�-end of the mRNAis complete. Its open reading frame encodes a 288-amino-acidunknown protein with a theoretical molecular weight of 31.6kDa, which was named AI-BP (apoA-I binding protein; Fig.1A). BLAST searches of databases identified no significanthomology matches against proteins with known function. Noextension of the 5�-cDNA was obtained by 5�-RACE-PCR(data not shown), indicating that the first methionine residueof the sequence represents the start codon. In accordance withthis, the computer-based program SignalP predicted a signalpeptide for translocation into the endoplasmic reticulum withthe most probable cleavage site located between amino acids24 and 25 (IKS-QT), giving rise to a mature protein of 29.1 kDaand an isoelectric point (pI) of 5.48.

Database searches detected several homologous ESTsfrom mouse (EMBL acc. nos. BG864411 and BG298999), cat-tle (EMBL acc. nos. BF039553 and BF044968), and pig (EMBL

695


A

B

APOA1BP m
acc. nos. BG732731 and BE235291). These fractional ESTs gaverise to cDNAs encoding complete proteins or protein frag-ments which display a high homology to the human AI-BPprotein (Fig. 1B). Thus, human and murine AI-BP display88.2% identity. The murine AI-BP protein lacks amino acids44–49 of the human homolog. Bovine and porcine AI-BP pro-tein fragments display 89.0% or 89.5% identity to human AI-BP, respectively. These data demonstrate that AI-BP is ahighly conserved protein and indicate an important physio-logical function for the AI-BP protein.
BLAST searches of the APOA1BP cDNA sequence againstthe human genomic NCBI database revealed a genomic frag-ment on chromosome 1q21.2–q22 (EMBL acc. no. AL139412)containing the complete coding region of APOA1BP. No sig-nificant hits against other genomic fragments of other chro-mosomal regions were found. A familial combined hyper-lipidemia (FCHL) locus was mapped within the chromosomalregion 1q21 with a lod score of 3.50 in a two-point analysisand 5.93 with a multipoint analysis [23]. Current studies willreveal whether mutations or polymorphism within APOA1BPare associated with increased triglyceride levels. Analysis of

696

the exon–intron junctions indicated that theAPOA1BP gene is composed of six exons andfive introns and spans 2.5 kb. The exon–intronstructure of APOA1BP and the adjacentnucleotide sequences are shown in Fig. 2A. Allexon–intron junctions are consistent with theAG-GT rule. Exon 1 encodes the start codon and27 bp of 5�-UTR. Exon 6 contains the stop codonand a polyadenylation signal. The 5�-flankingregion of APOA1BP is shown in Fig. 2B. Thetranscription start site determined by RACE-PCR is indicated. The putative promoter regioncontains no TATA box, but a GC-rich regionproximal to the transcription start site. The 5�-flanking region of APOA1BP contains severalputative binding sites for transcription factorsincluding Sp1, AP-1, and E-box motifs that arealso found in promoters of proteins involved inlipid metabolism like LDL-receptor [24], SR-BI[25], and CD36 [26].

Tissue Expression of APOA1BP mRNADot-blot analysis revealed an abundantRNA expression in all tissues investigated, indi-

FIG. 2. Structure of APOA1BP. (A) Exon–intron structure ofhuman APOA1BP. The gene is composed of six exons andfive introns and spans about 2.5 kb. The nucleotide sequenceat the exon–intron junctions and the size of exons and intronsare shown below. Exon sequences are shown in capital letters,intron sequences are shown in lowercase letters. (B) Sequenceof the 5�-region of APOA1BP. The transcription start sitemapped by 5�-RACE-PCR is indicated with +1. The startcodon is shown in bold.

cating an ubiquitous expression of APOA1BP at a basal level(Fig. 3). Among the analyzed tissues, the highest mRNAexpression was found in kidney, apex of the heart, liver, thy-roid gland, adrenal gland, and testis. The high expression ofAPOA1BP mRNA in kidney and liver is noteworthy, as theseorgans are the major sites of apoA-I catabolism [27,28]. Thethyroid hormones produced by the thyroid gland are at leastin part associated with apoA-I containing HDL [15] and aremajor regulators of HDL metabolism [29]. Adrenal gland andtestis are both steroidogenic organs displaying a high HDLand cholesterol turnover. The heart muscle, which displays ahigh APOA1BP mRNA expression, is an organ sensitive tothyroid hormones. Because both HDL and thyroid hormoneshave been shown to influence the heart rate [30], there mightbe a functional relationship between HDL metabolism, thy-roid hormone action, and APOA1BP expression in heart mus-cle. Regarding the APOA1BP expression in cell lines, the high-est expression was detected in HeLa S3 cells, which arederived from an epitheloid cervix carcinoma, and in epithe-lial A549 cells, which are derived from a lung carcinoma.



detectable in
Interaction of AI-BP and ApoA-ITo verify the interaction of AI-BP with apoA-I observed in thetwo-hybrid system, ELISA plates were coated with apoA-I orcontrol proteins and incubated with purified recombinant AI-BP protein expressed with a glutathione S-transferase tag(GST–AI-BP; Fig. 4A). Binding of GST–AI-BP was determinedusing an anti-GST antibody. By this approach we demon-strated a specific interaction of AI-BP with apoA-I, apoA-II,and HDL3, whereas no interaction of AI-BP with albumin(BSA) or �1-antichymotrypsin (ACT) was detected. No bind-ing was detectable when GST alone was used, indicating thatthe observed interactions of GST–AI-BP were not due to theGST tag.
For further studies, we produced a polyclonal antibodyagainst AI-BP by immunization of a rabbit with a peptide com-


prising amino acids 263–282 of the AI-BP pro-tein. Subsequent western blot analysis demon-strated that this antibody strongly detects a 29-kDa protein in the lysate of HepG2 cells and toa lower extent a 29-kDa band in the supernatantof these cells (Fig. 4C), indicating that AI-BP ismainly localized intracellularly but is alsosecreted to a detectable level by the cells. Toanalyze whether AI-BP is associated withapoA-I, the cells were adapted to serum-freemedium supplemented with BSA and after 24hours of incubation the supernatant was har-vested and passed over an anti-apoA-I column(Fig. 4B). Bound proteins were extensivelywashed with PBS, subsequently eluted in 0.1 Mglycine buffer, pH 2.5, and analyzed by west-ern blot. Significant amounts of the 29 kDa pro-tein corresponding to the mature AI-BP proteinco-purify with apoA-I indicating a direct inter-action of both proteins (Fig. 4B). No signal wasobtained in the control experiment using theisotype control antibody (IgG).

Analysis of AI-BP in Serum, CerebrospinalFluid, and UrineTo further study the secretion of AI-BP, weanalyzed several serum samples for the pres-ence of AI-BP protein by western blot (Fig. 5A).As inflammation induces severe changes inapoA-I metabolism [31], we also analyzedserum samples from patients suffering fromsystemic inflammation (SIRS) and sepsis. Theresults demonstrate that AI-BP protein is not

serum of healthy donors. One of the three

FIG. 3. Tissue expression of APOA1BP mRNA. A normalizeddot blot containing mRNAs of 76 different human tissues washybridized with a radioactively labeled APOA1BP cDNAprobe and signals were analyzed using a phosphoimager. Therelative tissue expression of APOA1BP mRNA is shown. Blackbars indicate tissues with the highest expression.

patients with septic syndromes serum contained detectableamounts of the AI-BP protein (Fig. 5A). Recent data demon-strated that there are profound changes affecting the apoA-Imetabolism caused by inflammation leading to changed dis-tribution and composition of lipoprotein subclasses. Thus,there is a decrease of HDL in inflammation that is mainly dueto a decrease in LCAT activity and an increase of apoE-con-taining particles and PLTP activity [31]. Whether the increaseof AI-BP observed in sepsis serum contributes to thesechanges of the lipoprotein pattern or whether the AI-BPincrease is a consequence of the changes of the lipoproteinmetabolism in sepsis has to be investigated. As significantamounts of AI-BP were found in the supernatant of liverHepG2 cells, the absence of detectable AI-BP levels from

697


serum samples of healthy probands might indicate a rapidcatabolism of AI-BP after secretion by the liver, which mightbe inhibited under inflammatory conditions like sepsis.

The presence of apolipoproteins including apoA-I andapoE in human cerebrospinal fluid (CSF) was first describedby Roheim et al. [32]. To investigate whether AI-BP protein isalso detectable in CSF, we analyzed several CSF samples bywestern blot. Significant amounts of AI-BP are present in allanalyzed CSF samples (Fig. 5B). However, there is no corre-lation of AI-BP and apoA-I amounts in the analyzed samples(data not shown).

ApoA-I is among those proteins that are filtered by theglomeruli of the kidney and is at least in part reabsorbed bytubular epithelial cells. As AI-BP is also highly expressed inthe kidney tubular cells, we investigated whether AI-BP pro-tein is found in urine (Fig. 5C). Therefore urinary proteins ofhealthy donors (lanes 3–7) and patients with an impaired renal

FIG. 4. Interaction of AI-BP with apoA-I. (A) ELISA plates were coated with10 �g/ml of the indicated proteins. Coated plates were incubated with recom-binant expressed AI-BP protein labeled with a glutathione S-transferase (GST)tag. Binding of GST–AI-BP was determined using an anti-GST antibody-per-oxidase conjugate. BSA, bovine serum albumin; ACT, �1-antichymotrypsin. (B)AI-BP is associated with apoA-I in supernatants of cultured HepG2 cells.HepG2 cells were adapted to serum-free medium supplemented by 0.2% BSA.After 24 hours of incubation the supernatant was harvested and passed overan anti-apoA-I column or a column coated with an isotype control antibody(IgG). Bound proteins were extensively washed and subsequently eluted in 0.1M glycine buffer, pH 2.5. The proteins were analyzed by western blot usingan anti-AI-BP antibody (top) and an anti-apoA-I antibody (bottom). (C) AI-BPwestern blot analysis of HepG2 cells (left) and HepG2 supernatant (right).

FIG. 5. AI-BP is found in serum, CSF, and urine. (A) Three different serum samples of sepsis patients (lanes 1–3) and four control sera from healthy probands(lanes 4–7) were analyzed by SDS-PAGE and western blot using an anti-AI-BP antibody. (B) AI-BP western blot analysis of CSF samples obtained from sevendifferent probands. (C) AI-BP and apoA-I western blot analysis of urine samples (pooled over a period of 24 hours) from healthy probands (lanes 3–7) and patientswith an impaired renal function (lanes 1, 2). (D) AI-BP and apoA-I western blot analysis of urine samples from megalin-deficient mice and wild-type mice (Wt).Mice were given 10% sucrose in drinking water and urine was collected for 16 hours.

A B

C D

698

function (lanes 1 and 2) were concentrated by ultrafiltrationand subsequently analyzed by western blot. Indeed, signifi-cant amounts of AI-BP were detected in the urine samples ofhealthy donors (Fig. 5C). Probands with an impaired renalfunction and an increased urinary excretion of serum proteins(lane 1 and 2) showed an increased amount of apoA-I and a

A

B C



FIG. 6. Native 2D gel electrophoresis and analysis of AI-BP migration pattern. Serum of a sepsis patient (A), serum of a healthy proband (B), CSF (C), and aurine sample (D) were separated by 2D gel electrophoresis under native conditions. The migration pattern of AI-BP (top) and apoA-I (bottom) were analyzedby western blot. The migration pattern of apoA-I is shown as a projection (top). The AI-BP containing particle that is recognized specifically by the anti-AI-BPantibody is marked by an arrow. The additional immunostaining observed in a molecular range of 180 to 600 kDa in (A) and (B) is due to an unspecific cross-reaction of the antibodies used for western blot analysis. The unspecific staining of these spots is often observed in native gel electrophoresis of serum and ismost probably caused by a cross-reaction with the immunoglobulins or immunoglobulin binding proteins in the serum samples. It should be noted that themigration of the proteins does not reflect the true molecular weight, because the standards do not provide adequate information on particle size as electrophoresiswas not run to completion.

A B C D

reduced amount of AI-BP in urine. Because AI-BP is notdetectable in serum of healthy probands, the presence of AI-BP in the urine indicates that AI-BP is released from the tubu-lar cells of the kidney into the urine. The multiligand receptorcomplex of the HDL receptor cubilin/megalin is involved inthe tubular resorption of filtered plasma proteins [33] andmegalin mutant mice exhibit a resorption deficiency andexcrete low molecular weight plasma proteins [34]. Analysisof AI-BP levels in urine of megalin-mutant mice indicated thatthese mice display an increased excretion of AI-BP (Fig. 5D).No significant difference in the urinary creatinine concentra-tion was observed between wild-type and megalin mutantmice. This observation indicates that AI-BP, like apoA-I, maybe a ligand for the cubilin/megalin receptor complex.However, further studies are necessary to reveal interindi-vidual differences and to analyze a possible correlation ofapoA-I and AI-BP excretion. No significant differences in uri-nary AI-BP excretion was observed in LCAT-deficient micecompared with wild-type mice or in ABCA1-deficient Tangierpatients compared with healthy probands (data not shown).

The migration pattern of AI-BP in sepsis serum, normalCSF, and urine was also analyzed by native two-dimensionalgel electrophoresis (Fig. 6). In sepsis serum and in CSF, thenative AI-BP protein migrates as a complex with a mobilitysimilar to pre�-HDL and an apparent molecular weight ofapproximately 80 kDa (Figs. 6A and 6C). A 29-kDa


monomeric form of AI-BP could not be detected. No AI-BPcontaining protein complex was detectable in the serum ofhealthy probands (Fig. 6B). The urinary AI-BP complexmigrates with pre�-mobility but has a slightly reduced molec-ular weight compared with the particle found in serum andCSF (Fig. 6D). The reduced molecular weight of this complexmay be due to the partial dissociation or degradation of com-ponents of the particle in the urine. Under the conditions ofthe electrophoresis there is no comigration of AI-BP withapoA-I containing lipoproteins in serum, CSF, or urine.Similarly, no co-migration of apoA-I and AI-BP was observedin HepG2 supernatants, which contained the same pre�-migrating 80-kDa AI-BP particle found in sepsis serum andCSF (data not shown) indicating that the conditions of elec-trophoresis cause the dissociation of the apoA-I/AI-BP com-plex found in HepG2 supernatant (Fig. 4). In contrast to theresults obtained by immunopurification of apoA-I fromHepG2 cell culture supernatants (Fig. 4), we detected no asso-ciation of AI-BP and apoA-I by immunopurification of theproteins in sepsis serum. However, it should be consideredthat apoA-I containing lipoprotein particles secreted byHepG2 cells differ from apoA-I lipoproteins found in serumand that additional changes of HDL metabolism are associ-ated with inflammatory diseases like sepsis [31]. Therefore weassume that the association of apoA-I with AI-BP depends onthe composition of the apoA-I lipoprotein particles.

699


FIG. 7. Secretion of AI-BP by HK-2 cells is induced by apoA-I and HDL. (A) HK-2 cells derived from proximal tubules of the kidney were stimulated with apoA-I, apoA-II, HDL, LDL, or BSA for 18 hours. The control cells were not stimulated with protein. The supernatants of the cells were analyzed by western blot usingthe AI-BP antibody. (B) HK-2 cells were incubated with different concentrations of apoA-I or BSA. After 18 hours of incubation the supernatant was analyzedfor AI-BP protein by western blot. (C) Northern blot analysis of APOA1BP mRNA in HK-2 cells. HK-2 cells were incubated for 18 hours with BSA, apoA-I orHDL3. APOA1BP mRNA was detected by northern blot analysis using an APOA1BP cDNA probe (top). The blot was stripped and hybridized with a �-actinprobe to demonstrate equal RNA loading (bottom). Intestine Caco-2 (D) or liver HepG2 (E) cells were adapted to serum-free medium and subsequently stimu-lated with the indicated proteins for 18 hours. The supernatant of the cells was analyzed for AI-BP protein by western blot.

A

D

B

E

C

AI-BP Secretion Is Induced in Kidney Proximal TubuleCells by ApoA-ITo further investigate whether cells of the proximal tubules ofthe kidney are involved in the secretion of AI-BP protein, westimulated HK-2 cells, a proximal tubular cell line, with apoA-Iand HDL. Incubation of the cells with apoA-I resulted in a strongincrease of secreted AI-BP (Fig. 7A). Similarly, HDL2 and HDL3induced a higher secretion of AI-BP compared with the unstim-ulated control cells. In contrast, stimulation of the cells withapoA-II, BSA, or LDL had no effect on the AI-BP release indi-cating that the effect is specific for apoA-I. Incubation of the HK-2 cells with increasing concentrations of apoA-I (5–75 �g/ml)revealed that the secretion of AI-BP correlates with the apoA-Iamount used for stimulation (Fig. 7B). The intracellular level ofAI-BP of the HK-2 cells was not significantly altered followingstimulation with apoA-I (data not shown). Therefore, apoA-Iand HDL induce an upregulation of total AI-BP protein that issubsequently secreted. As northern blot analysis demonstratedthat APOA1BP mRNA is not upregulated following stimulationof HK-2 cells with apoA-I or HDL3 (Fig. 7C) the increase in AI-BP protein is probably due to a post-transcriptional or post-trans-lational mechanism. In contrast to the kidney HK-2 cells, apoA-I producing intestine Caco-2 cells (Fig. 7D) or liver HepG2 cells(Fig. 7E) show a constitutive secretion of AI-BP, which was notincreased by apoA-I or HDL.

To investigate the origin of AI-BP in CSF, we analyzedwhether apoA-I induces AI-BP expression and secretion by

700

porcine epithelial cells of the choroid plexus which regulatethe entry of molecules into the CSF. Although the cells expressAI-BP protein, there was no constitutive or apoA-I inducedAI-BP secretion obeserved in these cells (data not shown).Further experiments will reveal whether other factors induceAI-BP secretion from the cells of the choroid plexus orwhether other regions of the brain are responsible for therelease of AI-BP into the CSF.

No constitutive or apoA-I induced AI-BP secretion couldbe observed in other cells including epithelial cells from adre-nal glands or thyroid glands, two tissues displaying a highAPOA1BP mRNA expression (Fig. 3). These results indicatethat the apoA-I induced AI-BP secretion is specific for theproximal tubule cell line HK-2. Considering the specificapoA-I induced secretion of AI-BP in these cells and the pres-ence of AI-BP and apoA-I in the urine, it might be speculatedthat AI-BP is involved in the transcytosis or renal degradationof apoA-I in the kidney proximal tubules. Analysis of themegalin and cubilin expression of HK-2 cells by RT-PCR indi-cates that the HK-2 cells display no detectable expression ofmegalin and cubilin, but demonstrate a high expression ofthe scavenger receptor SR-BI (data not shown). Therefore, weassume that the apoA-I- and HDL-induced secretion of AI-BPis initially mediated by an interaction with SR-BI. Furtherexperiments are necessary to reveal the mechanism of apoA-I induced AI-BP secretion and to define the precise physio-logical function of AI-BP.



MATERIALS AND METHODS

Materials. ApoA-I and apoA-II were obtained from Calbiochem (San Diego,CA). Lipid-free BSA was obtained form Sigma (Deisenhofen, Germany). LDL(d = 1.006–1.063 g/ml) and HDL3 (d = 1.125–1.21 g/ml) were prepared fromhuman plasma of healthy donors using standard methods. The antibodyagainst AI-BP was produced by immunization of a rabbit with a peptide cor-responding to amino acids 263–282 of the AI-BP protein (VPPALEKKYQLNLP-PYPDTE). The antibody was purified by affinity chromatography (PinedaAntikörper Service, Berlin, Germany). The rabbit anti-apoA-I antibody waspurchased from Calbiochem, the rabbit isotype control antibody was fromR&D Systems (Wiesbaden, Germany). Urine samples of megalin deficient micewere a kind gift of T. Willnow (Max-Delbrueck-Center for Molecular Medicine,Berlin, Germany)

Yeast two-hybrid library screening. The yeast two-hybrid screening was per-formed with the Matchmaker system 2 offered by Clontech (Palo Alto, CA) fol-lowing the instructions of the manufacturer. Briefly, the cDNA encoding themature apoA-I (amino acids 25–267) was amplified from total RNA preparedfrom human liver by RT-PCR using the oligonucleotides A-Iuni 5�-GTGAATTCGATGAACCCCCCCAGAGCCCCTG-3� and A-Irev 5�-CTCT-GCAGCTGGGTGTTGAGCTTCTTAGTGTACTCC-3�. The cDNA was clonedinto the EcoRI/PstI restriction site of the vector pAS-2.1 encoding a fusion pro-tein of the GAL4 DNA-binding domain and apoA-I. This bait construct wastransformed together with a human liver two-hybrid cDNA library into theyeast strain Y190. Positive clones were isolated by growth-selection on mediumlacking leucine, tryptophan and histidine and tested for the activity of the LacZ-reporter gene by filter lift assays. To eliminate false-positive clones, severalcontrol transformations were performed as recommended by the manufac-turer.

DNA sequencing and sequence analysis. Plasmid DNA was sequenced usingthe Cy5-AutoRead kit and an ALFexpress sequencer (Amersham Pharmacia,Freiburg, Germany). Searches of nucleotide databases were performed usingBLAST and FASTA programs accessed via the EMBL or NCBI homepage.Protein sequence alignments and sequence analysis were performed using theClustalW and SignalP program.

Expression of recombinant proteins. The cDNA encoding the mature AI-BPprotein (lacking the signal peptide) was amplified by RT-PCR using humanmacrophage total RNA as template and the forward primer exp-uni 5�-GGGGAATTCCTGCCGCGGATCAAAAGCCAGACCATC-3� together with thereverse primer exp-rev 5�-GGGCTCGAGTCACTGCAGACGATAGACA-CACTCAGTG-3�. For expression of full-length HCAM1, the cDNA was ampli-fied from human kidney mRNA by RT-PCR using the forward primer Hcam-uni, 5�-GGGGGGATCCACATGGGGTCTAGTGCCACAGAGATTG-3�, and thereverse primer Hcam-rev, 5�-GGGGAATTCTTACTGATGAATAAACTCA-CACTTC-3�. The cDNAs were cloned into the EcoRI/XhoI (APOA1BP cDNA) orBamHI/EcoRI (Hcam1 cDNA) restriction site of the vector pGEX-5X-I (AmershamPharmacia) in-frame with the glutathione S-transferase (GST). For expression ofrecombinant fusion proteins, Escherichia coli Bl21 was grown to mid-log phaseand expression was induced by the addition of IPTG to a final concentration of1 mM. Cells were incubated for an additional 4 hours and subsequently lysed inPBS containing 1% Triton X-100 and a protease inhibitor cocktail (Sigma) bysonification. GST fusion proteins were purified with glutathione sepharose 4Bbeads (Amersham Pharmacia) in a batch procedure following the manufacturer’sinstructions. The amount of recombinant protein was estimated by Coomassieblue staining of SDS-PAGE gels.

RACE PCR. To determine the 5� end of APOA1BP mRNA, a 5�-RACE PCR wasperformed using a liver Marathon cDNA library (Clontech) as template. PCR wascarried out according to the instructions of the manufacturer using the providedadapter primer together with the gene-specific primer AI-BP-R300 (5�-CAGCTCAGCCCGGCCAGTTCCATAAGTTGG-3�). An aliquot of the initialPCR reaction served as template in a secondary (nested) PCR reaction with anested anchor primer and the nested primers AI-BPR240 (5�-CTCCTGGTC-CACGGCCTGGGCCTCCTCCTGG-3�) or AI-BPR180 (5�-CTCTGAGTC-CCAGCGGCCACCCGAGTTCAG-3�).

Binding assays. Maxisorp plates (Nunc, Wiesbaden, Germany) were coatedwith the indicated proteins at a concentration of 10 �g/ml in PBS by overnight


incubation at 4�C. Plates were washed and blocked with 1% BSA in PBS. Afterwashing, the purified recombinant proteins were added at a concentration of 1�g/ml in PBS. After 2 hours incubation at room temperature the plates werewashed with PBS containing 0.05% Tween-20. Binding of proteins was analyzedusing an anti-GST-antibody-peroxidase conjugate (Sigma) at a concentration of1:10,000 in PBS and a colorimetric substrate.

For co-purification of apoA-I and AI-BP, a rabbit anti-apoA-I antibody(Calbiochem) was covalently coupled to agarose beads using the CarboLink kitfrom Pierce (Rockford, IL). In control experiments rabbit IgG coupled beads wereused. ApoA-I was purified from the supernatant of HepG2 cells in a batch pro-cedure using the anti-apoA-I beads. The supernatant was incubated with thebeads at 4°C overnight followed by extensive washing of the beads with PBS.Proteins were eluted in 0.1 M glycine buffer, pH 2.5. The samples were neutral-ized and subjected to western blot analysis.

The interaction of HCAM1 and apoA-I was demonstrated using recombi-nant GST–HCAM1 and biotinylated apoA-I. For biotinylation of apoA-I, EZ-linkNHS-SS-biotin (Pierce) was used following the instructions of the manufacturer.We incubated 0.5 �g biotinylated apoA-I and purified GST–HCAM1 or GST in100 �l PBS for 2 hours at room temperature. Biotinylated apoA-I was isolatedwith avidin beads (Pierce), washed with PBS/0.05% Tween-20, and eluted inLaemmli buffer. Bound GST proteins were detected by SDS-PAGE and westernblot analysis using an anti-GST antibody.

SDS-PAGE and two-dimensional non-denaturing gradient gel electrophoresis.SDS-polyacrylamide gel electrophoresis was performed according to standardprocedures. CSF (20 �l) and serum (2 �l) were used for analysis. The urine sam-ples were concentrated from 200 �l urine by ultrafiltration. For urine collectionfrom mice, the animals were placed in metabolic cages for 16 hours and given10% sucrose in drinking water and urine was collected on ice.

After electrophoresis, proteins were transferred to PVDF membranes (Pall,Dreieich, Germany) as reported. Transfer was confirmed by Ponceau S stainingof the membrane. The antibody for AI-BP was used at a 1:1000 dilution in 5%nonfat dry milk in PBS/0.05% Tween-20. The secondary peroxidase conjugatedanti-rabbit antibody (Pharmingen) was diluted 1000-fold. Detection of theimmune complexes was performed with the ECL Western blot detection system(Amersham Pharmacia).

Two-dimensional non-denaturing gradient gel electrophoresis was per-formed as described [31]. For the first dimension, 5 ml of plasma, 20 �l of CSF,or 20 �l of urine (concentrated by ultrafiltration) was separated in 0.7% (w/v)agarose gels in 50 mM barbital buffer (pH 8.6). Samples were electrophoresed at10�C at constant 100 V for about 1 hour. For the second dimension, the agarosegel strips from the first dimension were transferred to a 4–15% polyacrylamidegradient gel (Bio-Rad, Munich, Germany). Separation in the second dimensionwas performed at 20 mA per gel for 4 hours at 4�C. After electrophoresis, pro-teins were electrophoretically transferred to Fluorotrans transfer membranes(Pall). Transfer was carried out for 18 hours in 20 mM Tris and 150 mM glycinebuffer, in a Trans-Blot cell (Bio-Rad) at a constant 30 V and 10�C.

Northern blot analysis. Total cellular RNA was isolated by the guanidine isoth-iocyanate-cesium chloride technique. Total RNA (10 �g) was separated througha 1.2% agarose gel containing 6% formaldehyde and blotted onto nylon mem-branes. After crosslinking with UV irradiation, the membranes were hybridizedwith a cDNA probe spanning nucleotides 102–893 of the APOA1BP cDNA,stripped, and subsequently hybridized with a human �-actin probe (Clontech).The probes were radiolabeled with [�-32P] dCTP using the Prime It Oligolabelingkit from Stratagene (La Jolla, CA). Hybridization and washing conditions wereperformed as recommended by the manufacturer of the membrane. For analy-sis of the APOA1BP tissue mRNA expression a multiple tissue expression arrayfrom Clontech was hybridized following the instructions of the manufacturer.

Cell culture. HepG2, Caco-2, 850C, and HK-2 cells were obtained from ATCCand cultivated according to the instructions of ATCC. At 24 hours before stim-ulation, HepG2 and Caco-2 cells were adapted to serum-free medium. HK-2 cellswere continuously cultivated in serum-free medium supplemented with EGF andpituitary gland extract (GibcoBRL, Berlin, Germany) as recommended by ATCC.Porcine choroid plexus epithelial cells were isolated and cultivated as described[35] with minor modifications [36]. The cells were incubated with the(apo)lipoproteins at the indicated concentrations for 18 hours. Subsequently themedium was removed, centrifuged, and analyzed by SDS-PAGE and westernblot. The cells were washed with PBS, lysed in Laemmli buffer, and subjectedto western blot analysis.

701


ACKNOWLEDGMENTS

We thank Thomas E. Willnow (Max-Delbrueck-Center for Molecular Medicine, Berlin,Germany) for the urine samples of the megalin mutant mice; Charalampos Aslanidis(Institute for Clinical Chemistry and Laboratory Medicine, University of Regensburg,Germany) for critically reading of the manuscript; and Cornelia Hasenknopf and SylviaKirchner-Luft (Institute for Clinical Chemistry and Laboratory Medicine, Universityof Regensburg, Germany) for technical assistance.

RECEIVED FOR PUBLICATION DECEMBER 27, 2001; ACCEPTED MARCH 5, 2002.

REFERENCES1. Fielding, C. J., and Fielding, P. E. (1995). Molecular physiology of reverse cholesterol

transport. J. Lipid Res. 36: 211–228.2. Kwiterovich, P. O., Jr. (2000). The metabolic pathways of high-density lipoprotein,

low-density lipoprotein, and triglycerides: a current review. Am. J. Cardiol. 86: 5L–10L.3. Sakai, N., et al. (1997). Targeted disruption of the mouse lecithin:cholesterol acyltrans-

ferase (LCAT) gene. Generation of a new animal model for human LCAT deficiency. J. Biol. Chem. 272: 7506–7510.

4. Bojanovski, D., et al. (1987). In vivo metabolism of proapolipoprotein A-I in Tangier disease. J. Clin. Invest 80: 1742–1747.

5. Krieger, M. (1999). Charting the fate of the “good cholesterol”: identification and char-acterization of the high-density lipoprotein receptor SR-BI. Annu. Rev. Biochem.68: 523–558.

6. Hammad, S. M., et al. (1999). Cubilin, the endocytic receptor for intrinsic factor-vitaminB(12) complex, mediates high-density lipoprotein holoparticle endocytosis. Proc. Natl.Acad. Sci. USA 96: 10158–10163.

7. Kozyraki, R., et al. (1999). The intrinsic factor-vitamin B12 receptor, cubilin, is a high-affinity apolipoprotein A-I receptor facilitating endocytosis of high-density lipoprotein.Nat. Med. 5: 656–661.

8. Graham, D. L., and Oram, J. F. (1987). Identification and characterization of a high density lipoprotein-binding protein in cell membranes by ligand blotting. J. Biol. Chem.262: 7439–7442.

9. Tozuka, M., and Fidge, N. (1989). Purification and characterization of two high-density-lipoprotein-binding proteins from rat and human liver. Biochem. J. 261: 239–244.

10. Koch, S., et al. (2001). Characterization of four lipoprotein classes in human cerebrospinalfluid. J. Lipid Res. 42: 1143–1151.

11. Zannis, V. I., Kardassis, D., and Zanni, E. E. (1993). Genetic mutations affecting humanlipoproteins, their receptors, and their enzymes. Adv. Hum. Genet. 21: 145–319.

12. Bojanovski, D., et al. (1985). Human apolipoprotein A-I isoprotein metabolism: proapoA-I conversion to mature apoA-I. J. Lipid Res. 26: 185–193.

13. Palgunachari, M. N., et al. (1996). Only the two end helixes of eight tandem amphipathichelical domains of human apo A-I have significant lipid affinity. Implications for HDLassembly. Arterioscler. Thromb. Vasc. Biol. 16: 328–338.

14. Laccotripe, M., Makrides, S. C., Jonas, A., and Zannis, V. I. (1997). The carboxyl-termi-nal hydrophobic residues of apolipoprotein A-I affect its rate of phospholipid bindingand its association with high density lipoprotein. J. Biol. Chem. 272: 17511–17522.

15. Benvenga, S., Gregg, R. E., and Robbins, J. (1988). Binding of thyroid hormones to humanplasma lipoproteins. J. Clin. Endocrinol. Metab. 67: 6–16.

16. Frank, P. G., and Marcel, Y. L. (2000). Apolipoprotein A-I: structure-function relation-ships. J. Lipid Res. 41: 853–872.

702

Sequence data from this article have been deposited with the DDBJ/EMBL/GenBank DAPOA1BP cDNA, respectively).

17. Schmidt, H. H., et al. (1995). Carboxyl-terminal domain truncation alters apolipoproteinA-I in vivo catabolism. J. Biol. Chem. 270: 5469–5475.

18. Han, H., et al. (1999). A novel mutant, ApoA-I nichinan (Glu235—>0), is associated withlow HDL cholesterol levels and decreased cholesterol efflux from cells. Arterioscler.Thromb. Vasc. Biol. 19: 1447–1455.

19. Malle, E., and De Beer, F. C. (1996). Human serum amyloid A (SAA) protein: a promi-nent acute-phase reactant for clinical practice. Eur. J. Clin. Invest. 26: 427–435.

20. Loughney, K., et al. (1996). Isolation and characterization of cDNAs corresponding to twohuman calcium, calmodulin-regulated, 3�,5�-cyclic nucleotide phosphodiesterases. J. Biol.Chem. 271: 796–806.

21. Schmitz, G., Robenek, H., Lohmann, U., and Assmann, G. (1985). Interaction of high den-sity lipoproteins with cholesteryl ester-laden macrophages: biochemical and morpho-logical characterization of cell surface receptor binding, endocytosis and resecretion ofhigh density lipoproteins by macrophages. EMBO J. 4: 613–622.

22. Takahashi,Y., and Smith, J. D. (1999). Cholesterol efflux to apolipoprotein AI involvesendocytosis and resecretion in a calcium-dependent pathway. Proc. Natl. Acad. Sci. USA96: 11358–11363.

23. Pajukanta, P., et al. (1998). Linkage of familial combined hyperlipidaemia to chromosome1q21-q23. Nat. Genet. 18: 369–373.

24. Brown, M. S., and Goldstein, J. L. (1997). The SREBP pathway: regulation of cholesterolmetabolism by proteolysis of a membrane-bound transcription factor. Cell 89: 331–340.

25. Cao, G., et al. (1997). Structure and localization of the human gene encoding SR-BI/CLA-1. Evidence for transcriptional control by steroidogenic factor 1. J. Biol. Chem. 272:33068–33076.

26. Armesilla, A. L., Calvo, D., and Vega, M. A. (1996). Structural and functional character-ization of the human CD36 gene promoter: identification of a proximal PEBP2/CBF site.J. Biol. Chem. 271: 7781–7787.

27. Glass, C. K., Pittman, R. C., Keller, G. A., and Steinberg, D. (1983). Tissue sites of degra-dation of apoprotein A-I in the rat. J. Biol. Chem. 258: 7161–7167.

28. van’t Hooft, F. M., and van Tol, A. (1985). The sites of degradation of purified rat lowdensity lipoprotein and high density lipoprotein in the rat. Biochim. Biophys. Acta 836:344–353.

29. Hargrove, G. M., Junco, A., and Wong, N. C. (1999). Hormonal regulation of apolipopro-tein AI. J. Mol. Endocrinol. 22: 103–111.

30. Williams, P. T., Haskell, W. L., Vranizan, K. M., and Krauss, R. M. (1995). The associa-tions of high-density lipoprotein subclasses with insulin and glucose levels, physicalactivity, resting heart rate, and regional adiposity in men with coronary artery disease:the Stanford Coronary Risk Intervention Project baseline survey. Metabolism 44: 106–114.

31. Barlage, S., et al. (2001). ApoE-containing high density lipoproteins and phospholipidtransfer protein activity increase in patients with a systemic inflammatory response. J. Lipid Res. 42: 281–290.

32. Roheim, P. S., Carey, M., Forte, T., and Vega, G. L. (1979). Apolipoproteins in humancerebrospinal fluid. Proc. Natl. Acad. Sci. USA 76: 4646–4649.

33. Christensen, E. I., and Birn, H. (2001). Megalin and cubilin: synergistic endocytic recep-tors in renal proximal tubule. Am. J. Physiol. Renal Physiol. 280: F562–F573.

34. Leheste, J. R., et al. (1999). Megalin knockout mice as an animal model of low molecularweight proteinuria. Am. J. Pathol. 155: 1361–1370.

35. Crook, R. B., Kasagami, H., and Prusiner, S. B. (1981). Culture and characterization ofepithelial cells from bovine choroid plexus. J. Neurochem. 37: 845–854.

36. Gath, U., Hakvoort, A., Wegener, J., Decker, S., and Galla, H. J. (1997) Porcine choroidplexus cells in culture: expression of polarized phenotype, maintenance of barrier prop-erties and apical secretion of CSF-components. Eur. J. Cell Biol. 74: 68–78.


ata Libraries under accession numbers AJ315849 and AJ344092 (human and murine

Cloning and Characterization of a Novel Apolipoprotein A-I Binding Protein, AI-BP, Secreted by Cells...

Documents

Transcript of Cloning and Characterization of a Novel Apolipoprotein A-I Binding Protein, AI-BP, Secreted by Cells...