The Na⁺-phosphate cotransport system (NaP_i-II) with a cleaved protein backbone: implications on function and membrane insertion

Abstract

1. Renal handling of inorganic phosphate (P_i) involves a Na⁺-P_i cotransport system which is well conserved between vertebrates. The members of this protein family, denoted NaP_i-II, share a topology with, it is thought, eight transmembrane domains. The transporter is proposed to be proteolytically cleaved within a large hydrophilic loop in vivo.

2. The consequences of an interrupted backbone were tested by constructing cDNA clones encoding different N- (1-3 and 1-5) and C-terminal (4-8 and 6-8) complementary fragments of NaP_i-II from winter flounder. When the cognate fragments were used in combination (1-3 plus 4-8; 1-5 plus 6-8) they comprised the full complement of the putative transporter domains.

3. None of the four individual fragments or the 1-5 plus 6-8 combination when expressed in Xenopus oocytes increased P_i flux. Coexpression of fragments 1-3 plus 4-8 stimulated transport activity identical to that for expressed wild-type NaP_i-II with regard to pH dependency and K_m for Na⁺ and P_i binding; however, the maximal transport rate (v_max) was lower.

4. Immunohistochemistry on cryosections confined the functionally active 1-3 plus 4-8 combination to the oocyte membrane. This was not the case for the 1-5 plus 6-8 combination or any of the individual fragments, all of which failed to induce fluorescence.

5. A second immunohistochemical approach using intact oocytes allowed determination of the extracellular regions of the protein. Epitopes within the loop between transmembrane domains 3 and 4 enhanced fluorescence. Neither N- nor C-terminal tags induced fluorescence.

A constant level of inorganic phosphate (P_i) in the body is maintained by tightly regulated renal excretion. Glomerular filtration or tubular secretion (as in fish and birds) is followed by controlled reabsorption. The net flux is influenced by the P_i blood level, acid-base state and a variety of hormonal factors (e.g. parathyroid hormone and growth factors; Renfro & Gupta, 1990; Berndt & Knox, 1992). Renal Na⁺-dependent P_i cotransport systems, which are crucially involved in P_i homeostasis, have been cloned from a variety of species (Murer & Biber, 1997). The majority of different cotransporters belong to a protein family designated NaP_i-II. Besides the well-documented role in renal P_i handling NaP_i-II homologues seem to be involved in bone remodelling and intestinal P_i absorption in fish (Gupta, Miyauchi, Fujimori & Hruska, 1996; Kohl et al. 1996).

A common model for all members of the NaP_i-II protein family is based on hydropathy analysis. The model predicts eight transmembrane segments (Fig. 1) with both N- and C-termini facing the cytoplasm and a large hydrophilic loop between the third and fourth transmembrane domains which virtually divides the protein in two (Murer & Biber, 1994). This loop carries several potential N-glycosylation sites (2 in rat, 7 in flounder; Magagnin et al. 1993; Werner, Murer & Kinne, 1994; Kohl et al. 1996). However, this model awaits experimental confirmation. The Na⁺-P_i cotransport systems are well conserved between fish and mammals. Rat and flounder NaP_i-II show about 60% identical amino acids, increasing to 80% in the functionally relevant trans-membrane domains (Werner et al. 1994). This close relationship is reflected on a functional level. The kinetic characterization of the rat and fish proteins after expression in Xenopus laevis oocytes revealed only marginal differences with regard to the affinities for Na⁺ and P_i, as well as for the pH dependency of the transport (Forster et al. 1997).

Figure 1. Hypothetical model of the Na⁺-P_i transport protein, flounder NaP_i-II.

Transmembrane spanning segments are numbered (1-8). The glycosylation sites in the large extracellular loop are indicated (♦). Coexpression of N- and C-terminal fragments results in a putative transporter with an interrupted backbone. The two sites in the second and third extracellular loop are shown (‖). The flounder NaP_i-II cDNA is depicted below the main diagram indicating the relative positions of the restriction sites relevant for the subcloning of the different constructs.

There is evidence that rat NaP_i-II protein is cleaved within the hydrophilic loop in vivo (Biber et al. 1996; Boyer, Xiao, Dugré, Vincent, Delisle & Béliveau, 1996). Using SDS- PAGE, under non-reducing conditions rat NaP_i-II migrates as a 85 kDa protein. Under reducing conditions, antibodies against the N- or C-terminus were shown to recognize bands of 50 or 40 kDa, respectively. We decided to test the functional consequences of such a bipartition by expressing N- and/or C-terminal transporter fragments in Xenopus laevis oocytes. Since we expected a drastic reduction of transport activity due to the interrupted protein backbone we chose the Na⁺-P_i cotransport system from winter flounder (Pleuronectes americanus) as a model. The P_i-induced currents in oocytes related to flounder NaP_i-II (also referred to as NaP_i-5) exceed the mammalian NaP_i-II-related currents by an order of magnitude. Furthermore, membrane delivery of the newly synthesized transporter fragments is supposedly improved by using a fish and not a mammalian protein in an amphibian expression system.

The object of this study was to test the consequences of an interrupted protein backbone on the functional integrity of the Na⁺-P_i cotransporter. Furthermore, we wanted to establish a highly reliable system to study topological aspects of NaP_i-II. The expression in oocytes guarantees correct folding of the membrane-integrated transporters. Using two different immunohistochemical approaches we were able to determine whether epitopes were located in the intra- or extracellular compartments. The correct processing of truncated membrane proteins implicates a possibility for high level expression of the different fragments followed by reassociation.

METHODS

Polymerase chain reaction (PCR)

The PCRs were performed using an Omnigene thermocycler (Hybaid, Teddington, UK). The flounder NaP_i-II-specific oligo-nucleotides used were 5′-ACTGATGCATTCCAGAATTACAC-GTTAGTC-3′ to generate fragment 1-3; 5′-GACTAACGTGA-CTTTCTGGAATGC-3′ for the C-terminally tagged fragment 1-3; 5′-TTCAGAGCTCCGCTGCAGTCATGTTCTGGAAT-GCGAC-3′ for fragment 4-8; 5′-CGCGATGCATGCTTAAG-CAGCAGCAAGTATAGC-3′ for fragment 1-5; 5′-TTCAGAG-CTCCGCTGCAGTCATGGCTAGTCCCGCG-3′ for fragment 6-8; 5′-CCTCCTCTTATG-GACTACAAGGACG-A-CGATGA-CAAGCAGAGCTCCGC-3′ plus 5′-GGAGC-TC-T-G-C-TTG-T-C–ATCGTCGTCCTTGTAGTCCATAAGAGGAGGAGG-3′ to generate the N-terminal Flag and 5′-GATGCATTTGACTA–CA-A-GGACGACGATGACAAGGTGACACACCTTTAA-3′ plus 5′-CAATAAAAGGTGTGTCACCTTGTCATCGTCGT-CCT TGTAGTCAATGCATC-3′ for the C-terminal epitope. The exact cycling conditions as well as aliquots of the primers are available upon request.

Site-directed mutagenesis

Point and deletion mutations were carried out by recombinant PCR. In brief, two PCRs were performed to introduce the mutation in two overlapping fragments. The amplicons were analysed by agarose gel electrophoresis. The bands were cut out and transferred to a spin column. The gel slices were spun for 10-15 min at 14 000 g in an Eppendorf centrifuge. 1-5 μl of the samples were used for the overlapping PCR (rTth DNA Polymerease; Perkin Elmer, Weiterstadt, Germany). The resulting fragment was digested with endonucleases for cloning purposes according to the sites introduced into the primers.

Epitope tagging

The NaP_i-II-derived cDNA fragments were generated by PCR and cloned into vector cassettes containing a Flag epitope (Integra Biosciences, Fernwald, Germany) encoding 24mer (GACTAC AAGGACGATGATGACAAG) at the 5′, 3′ or both ends. The insertion mutations were done by overlapping PCR (Ho, Hunt, Horton, Pullen & Pease, 1989). A single Sac I restriction site (position 20) and a Nsi I site (position 1981) flanking the NaP_i-II coding region (see Fig. 1) were used for cloning purposes. This resulted in a N-terminally tagged construct containing the start methionine, the epitope, followed by a linker of three to six amino acids and the NaP_i-II-related protein part. The C-terminal Flag was connected to the transporter sequence by a linker of three to fifteen amino acids.

Xenopus laevis oocyte expression

Female Xenopus laevis were obtained from H. Kähler (Hamburg, Germany). According to national guidelines, full anaesthesia was induced by immersion in 1 g l⁻¹ tricaine (Sigma) and parts of the ovaries were surgically removed. The frogs were allowed to recover without taking special precautions. The handling of oocytes has been described in detail elsewhere (Werner, Biber, Forgo, Palacin & Murer, 1990). The components for in vitro transcription were obtained from Promega (Madison, WI, USA) and Pharmacia and standard protocols were followed (Short, Fernandez, Sorge & Huse, 1988). Routinely, oocytes were injected with 5 ng cRNA either flounder NaP_i-II wild-type or the relevant construct(s) and measured after 2 days. The flux assay for radioactive P_i has been described in detail elsewhere (Werner et al. 1990), the ³²P-labelled orthophosphate was purchased from Amersham (Buckinghamshire, UK).

Electrophysiology

Two electrode voltage-clamp recordings were performed using a Geneclamp 500 amplifier (Axon instruments) and a MacLab D/A converter and software (ADInstruments, Castle Hill, Australia). The oocytes were clamped at -50 mV in a superfusate containing 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂ and 5 mM Hepes (pH 7.4). To study substrate dependency either P_i concentrations were varied as indicated or Na⁺ was gradually replaced with equimolar amounts of choline. With a superfusion flow rate of 20 ml min⁻¹ a complete exchange of the bath solution occurred within 10 s. Currents were induced by adding 1 mM P_i to the superfusate for 30 s if not indicated otherwise. All experiments were repeated at least three times, the results proving identical. Variations in expression were found in different batches of oocytes.

In vitro translation

The rabbit reticulocyte lysate in vitro translation kit, as well as pancreatic microsomes, were obtained from Promega. All experiments were performed strictly according to the supplier's manual. cRNA (1 μg) from each wild-type NaP_i-II and the different constructs were translated in the presence of canine pancreatic microsomes and [³⁵S]-methionine (Amersham) for 1 h at 30°C. Microsomes were enriched in aliquots by centrifugation through a 1 ml cushion of sucrose (250 mM sucrose, 50 mM Tris-HCl pH 7.4) for 1 h at 14 000 g at 4°C in an Eppendorf centrifuge. The pellets were dissolved in 0.4% SDS and denatured at 96°C. Endoglycosidase H cleavage was carried out in 0.75 mM sodium citrate (pH 5.5) containing 1 mU endoglycosidase H (Boehringer) at 37°C for 4 h. The samples were heat denatured and separated over a 8-16% SDS-polyacrylamide gel. The radiolabelled proteins were detected by autoradiography.

Immunohistochemistry

Intact oocytes

Intact oocytes for direct detection of extracellular epitopes were incubated for 1 h in 1% bovine serum albumin. Barth's solution (composition: 88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO₃, 0.82 mM MgSO₄, 0.33 mM Ca(NO₃)₂, 0.41 mM CaCl₂, 10 mM Hepes-Tris (pH 4), 40 μg l⁻¹ gentamicin) was used to dissolve all indicated substances and for washing. After washing three times, the antibody directed against the Flag epitope (Integra Biosciences, Fernwald) was added at a concentration of 17.5 μg ml⁻¹ for 1 h. The washing was performed as above. The secondary anti-mouse antibody coupled to Cy2 or FITC (fluorescein isothiocyanate; Dianova, Hamburg, Germany) was diluted 1:100. After washing the oocytes were analysed using a confocal laser microscope (MRC 600; BioRad, Munich, Germany).

Thin sections

Cryosections of the oocytes were made according to Terada, Saito, Mukai & Inui (1996), for the immunostaining we followed a protocol by Custer, Lötscher, Biber, Murer & Kaissling (1994). In brief, the oocytes were fixed for 1 h in 3% paraformaldehyde, washed with phosphate-buffered saline (PBS; Sigma) and cryo-protected for at least 16 h in 30% sucrose-PBS. The 8 μm cryosections were incubated for 10 min in goat serum, washed with PBS followed by incubation with the primary antibody (mouse, anti-Flag) and the secondary antibody (goat, anti-mouse) coupled to the Cy3-fluorochrome.

RESULTS

Biber et al. (1996) and Boyer et al. (1996) concluded from Western blot experiments that rat NaP_i-II may be cleaved in vivo. We decided to simulate a comparable situation by coexpressing two independent clones encoding either an N- or a C-terminal fragment of NaP_i-II from flounder. Four different constructs of the domains were made (Fig. 1): 1-3, 1-5, 4-8 and 6-8, referring to the encoded transmembrane segments. The combination of 1-3 plus 4-8 is supposed to represent the naturally occurring version, the other constructs served for control purposes. The combinations 1-3 plus 4-8, and 1-5 plus 6-8 both add up to the full-length transporter. A Flag epitope (indicated by a star in Figs 5 and 6) at either end of the transporter was used for immunodetection.

Figure 5. Immunodetection of tagged NaP_i-II in oocytes: cryosections.

A, thin sections of oocytes injected with water (a), Flag epitope-tagged NaP_i-II cRNA (no anti-Flag) (b), and Flag epitope-tagged NaP_i-II cRNA (c) with its corresponding bright-field image (d). The pictogram in c represents the putative translation product with the schematic Flag epitope (⋆). The background fluorescence, predominantly in intracellular regions, varies from oocyte to oocyte. B, thin sections of oocytes injected with the constructs 1-3 plus 4-8 (a) with its corresponding bright-field image (b), fragment 1-3 (c) and fragment 4-8 (d). C, thin sections of oocytes injected with fragments 1-5 plus 6-8 (a), with its corresponding bright-field image (b), fragment 1-5 (c), and fragment 6-8 (d).

Figure 6. Immunofluorescence of intact oocytes incubated with anti-Flag antibodies.

A, NaP_i-II cRNA with (a) or without (b) Flag was injected and the oocytes were assayed after 3 days of incubation. The oocytes show almost no background fluorescence. B, cRNA of fragments 1-3 plus 4-8 untagged (a); 1-3 with a C-terminal tag and 4-8 untagged (b); 1-3 untagged and 4-8 with an N-terminal tag (c); 1-3 plus 4-8 tagged (d).

The different constructs were generated following a PCR-based strategy. A flounder NaP_i-II specific primer was used in combination with a ‘bivalent’ primer. It contained either an initiation or a stop codon defining the length of the construct, as well as a Sac I or a Nsi I restriction site, respectively. The primers complemented a Sac I site at the 5′ end of the NaP_i-II cDNA or a Nsi I site in the 3′ non-coding region. Flounder NaP_i-II cDNA without internal Sac I sites (A. Werner, unpublished data) was used as template DNA. The resulting fragments were cloned using the restriction endonuclease recognizing the bivalent primer and a convenient internal site (Pst I or Sac II). The flounder NaP_i-II non-coding regions were conserved in all the constructs in order to reduce variability in expression levels.

In a first set of experiments the different constructs and flounder wild-type NaP_i-II were transcribed in vitro and the resulting cRNA was translated in the presence of pancreatic microsomes. The membranes were purified through a sucrose cushion, denatured and separated by SDS-PAGE (Fig. 2, lanes labelled M). In parallel, an aliquot of each sample was treated with endoglycosidase H (Fig. 2, lanes labelled E). Wild-type NaP_i-II cRNA served as the positive control. The apparent molecular mass of flounder NaP_i-II is about 74 kDa. A 16 kDa shift to 58 kDa is observed after treatment with endoglycosidase H. The in vitro translation product of fragment 1-3 measures 28 kDa and seems not to be glycosylated. Its cognate partner has an apparent molecular mass of 43 kDa in the glycosylated form and 33 kDa after endoglycosidase H treatment. Additionally, the cRNAs of fragments 1-3 and 4-8 were cotranslated in the same experiment revealing no differences from the results above (data not shown). The in vitro translation products of the other constructs are represented in the right panel of Fig. 2. The NaP_i-II fragments migrated within the expected range and were properly glycosylated with the exception of fragment 1-3 as indicated in Table 1.

Figure 2. In vitro translation of cRNA derived from wild-type NaP_i-II and fragments 1-3, 4-8, 1-5 and 6-8.

The cRNA was transcribed in the presence of pancreatic microsomes. The membranes were purified through a sucrose cushion and then subjected to endoglycosidase H treatment (lanes labelled E). The controls without enzyme are shown in lanes labelled M. Proteins were separated over a 8-16% polyacrylamide gel. A 10 kDa protein standard was used for size calibration.

Table 1.

Molecular properties of the different N- and C-terminal constructs of flounder NaP_i-II

		Molecular mass (kDa)
	No. amino acids	Calculated	Apparent	No. glycosylation sites
Wild-type NaPi-II	636	70.7	74/58	7
Fragment 1-3	285	31.6	28/28	1
Fragment 4-8	351	39.0	43/33	5
Fragment 1-5	445	49.4	60/48	7
Fragment 6-8	191	21.3	24/24	—

The in vitro translation data are derived from Fig. 2. The cloning protocol for fragment 1-3 destroys one of the putative N-glycosylation sites. Apparent molecular mass shows values of the glycosylated/unglycosylated forms.

The different constructs were assayed for Na⁺-P_i cotransport activity using the Xenopus laevis oocyte expression system. The cRNAs (3-5 ng oocyte⁻¹) of the fragments were either injected individually or in pairs resulting, theoretically, in bipartite full-length transporters. Flux measurements were performed to screen for functionally relevant combinations. As shown in Fig. 3 none of the individual NaP_i-II fragments or the 1-5 plus 6-8 combination induced P_i transport in oocytes. The only functionally relevant combination consisted of fragments 1-3 and 4-8 which stimulated Na⁺-P_i cotransport to approximately an order of magnitude above control values (water-injected oocytes). On the other hand, the reassociated transporter always showed lower P_i transport rates (40-60%) than flounder wild-type NaP_i-II, perhaps due to a reduced abundance in the oocyte membrane. The Na⁺-P_i cotransport activity of the two active parts was characterized in detail in voltage-clamp experiments. The oocytes were injected with 3-5 ng of cRNA derived from fragments 1-3 or 4-8, or wild-type NaP_i-II or 50 nl of water as negative controls. The expression of exogenous protein increased linearly for 4 days (data not shown); the assays were performed after 2 days of incubation at 18°C. Unless indicated otherwise, the oocytes were clamped at a potential of -50 mV and currents were induced by the addition of 1 mM P_i (I_P). The values presented were obtained from different oocytes of the same frog. Experiments with oocytes derived from other frogs gave qualitatively similar results.

Figure 3. Expression of NaP_i-II fragments in Xenopus oocytes.

cRNA (5 ng) of each construct was injected either alone or combined with its cognate partner. Wild-type NaP_i-II cRNA or water-injected oocytes were used as controls. After 2 days incubation at 18 °C P_i transport activity was determined.

The coexpression of fragments 1-3 and 4-8 provoked a drastic increase in P_i-induced currents (I_P) from < -0.5 nA (control) to -75 ± 10 nA. Wild-type NaP_i-II cRNA induced a current of -190 ± 30 nA (Fig. 4A). Both currents decreased as a function of the membrane potential as described for other NaP_i-II type transporters (Fig. 4D; Busch et al. 1994). The currents induced by the combined parts and wild-type NaP_i-II were characterized with regard to extracellular concentrations of P_i and Na⁺ (Fig. 4B and C). Sodium ions in the superfusate were gradually replaced by choline and the currents (I_P/I_P,max, where I_P,max is maximal I_P) were fitted to the Hill equation. The K_m for Na⁺ binding for the cleaved transporter (40.1 ± 2.2 mM) and wild-type NaP_i-II (49.1 ± 5.1 mM) were found to be similar. Comparable experiments have been performed applying different concentrations of extracellular P_i. The calculated half-maximal currents were induced at P_i concentrations of 0.031 ± 0.001 mM (wild-type) and 0.026 ± 0.002 mM (fragments). Furthermore, no differences in pH dependency of the two currents could be detected (data not shown). The close similarity of all the tested parameters corroborate the functional identity of the two systems.

Figure 4. Electrophysiological characterization of fragments 1-3 plus 4-8 expressed in oocytes.

A, amplitudes of P_i-induced currents (I_P) in oocytes injected with water or cRNA encoding fragments 1-3, 4-8, 1-3 plus 4-8, or NaP_i-II wild-type. The holding potential was -50 mV, the currents were induced by superfusion of 1 mM P_i for 30 s. B, phosphate dependency of I_P. I_P was induced with varying concentrations of P_i in the superfusate at a holding potential of -50 mV. •, wild-type (K_m= 0.031 ± 0.001 mM). ▴, 1-3 plus 4-8 combination (K_m= 0.026 ± 0.002 mM). The data were fitted according to the Hill equation and represented as I_P/I_P,max. C, Na⁺ dependency of the P_i-induced currents. The functional characterization of I_P was performed using oocytes injected with wild-type NaP_i-II cRNA (•, K_m= 49.1 ± 5.1 mM) or coinjected with the 1-3 plus 4-8 cRNA (□, K_m= 40.1 ± 2.2 mM). Sodium in the superfusate was isosmotically replaced by choline. Data were fitted according to the Hill equation. D, P_i-induced currents as a function of the potential difference over the membrane. I_P was induced by superfusion with 1 mM P_i for 30 s at the indicated holding potentials. •, wild-type; ▴, 1-3 plus 4-8 combination. Data points and error bars represent means ±s.e.m., respectively; n = 4 or more.

The lack of transport activity in the other bipartite transporter (1-5 plus 6-8) could either have structural reasons or reflect impaired membrane delivery. In order to unravel the characteristics of the transport loss we monitored the expression of the different constructs immunologically. For this purpose, Flag epitopes were added to the N- or C-terminus of the four fragments (i.e. C-terminal Flag in 1-3; N-terminal Flag in 1-5, 4-8, and 6-8). Addition of the Flag epitopes did not alter the functional characteristics of the transporters (wild-type or 1-3 plus 4-8, results not shown). Cryosections were immunostained to demonstrate the membrane localization of the different constructs. In parallel, extracellular epitopes were detected by direct labelling of intact oocytes (Figs 5 and 6).

Routinely, oocytes were injected with 5 ng cRNA, incubated for 2-4 days and assayed for P_i transport activity. All the different fragments were assayed both separately and in the combinations resulting in full-length transporters (1-3 plus 4-8, 1-5 plus 6-8). Water-injected oocytes or oocytes expressing an untagged transporter served as negative controls. Cryosections were incubated with a monoclonal antibody recognizing the Flag epitope in combination with a Cy3-coupled secondary antibody. As shown in Fig. 5Ac and Ba the wild-type and 1-3 plus 4-8 combination were inserted properly into the membrane; however, none of the individually expressed fragments nor the 1-5 plus 6-8 combination were delivered to the oocyte membrane. This could have been for several reasons, e.g. a lack of interaction of the two fragments, decreased stability, or an inadequate interruption of the protein backbone.

In order to test the inserted constructs (1-3 plus 4-8, NaP_i-II) for their orientation in the membrane we performed immunohistochemistry using intact Xenopus oocytes. They were injected with different combinations of tagged and untagged cRNAs as indicated in Fig. 6. The oocytes were incubated with primary and secondary antibodies in Barth's solution and analysed by laser microscopy. The intact oocyte membrane prevents the antibodies from interacting with intracellular epitopes. Therefore, only extracellular tags should induce enhanced fluorescence. First, the orientation of wild-type NaP_i-II was determined. The absence of detectable fluorescence proves the intracellular location of both N- and C-termini (Fig. 6A). Then, the membrane insertion of 1-3 plus 4-8 was investigated using differently tagged constructs as indicated in Fig. 6B. Oocytes expressing the untagged constructs resulted in background fluorescence (Fig. 6Ba). If 1-3 and 4-8 were tagged at the C- and N-termini, respectively, and coexpressed, markedly increased fluorescence was observed (Fig. 6Bd). Similar results were obtained by complementing an epitope-carrying fragment with an untagged fragment (Fig. 6Bb and c). The result is in full agreement with the extracellular location of the large hydrophylic loop as proposed by Hayes et al. (1994) and with the initially proposed model. When oocytes were injected with the constructs 1-5 plus 6-8 there was no detectable fluorescence observed (data not shown). The lack of a signal confirms the negative result obtained with the cryosections.

DISCUSSION

We tested the functional consequences of an interrupted protein backbone in the Na⁺-P_i cotransporter type II as proposed by Boyer et al. (1996) and Biber et al. (1996). With this purpose in mind we constructed several truncated NaP_i-II related cDNA clones and expressed the cognate proteins in vitro and in Xenopus laevis oocytes. The fragments were denoted 1-3 plus 4-8, and 1-5 plus 6-8 referring to the putative membrane-spanning segments in the proposed topological model of NaP_i-II. The in vitro translation experiments revealed correct synthesis of all the fragments followed by integration into canine pancreatic microsomes. If translated individually the functional test after expression in oocytes was negative for all the constructs. Immunohistochemical experiments corroborated this finding: none of the proteins was delivered properly to the membrane. If two cRNAs which complemented to a full-length transporter (1-3 plus 4-8; 1-5 plus 6-8) were injected together two different situations were observed. (i) Fragments 1-5 and 6-8 did not reach the membrane. (ii) If 1-3 and 4-8 fragment cRNA were coexpressed a P_i-induced current with identical functional characteristics compared with wild-type flounder NaP_i-II could be measured. The two different immunochemical approaches using either intact oocytes or thin sections allowed the localization of different epitopes with regard to intra- or extracellular space.

The reconstitution of a functional complex from two independent fragments has been shown for a number of membrane proteins. The best described system includes the E. coli lac permease. The backbone of this protein can virtually be cut within any hydrophilic loop without abolishing transport activity completely (Zen, McKenna, Bibi, Hardy & Kaback, 1994). A bipartition of the permease within a membrane-spanning segment results in a loss of function with the protein still detectable in the membrane. A comparable approach with a eukaryotic system has been described for the human red blood cell anion exchanger (band 3, AE1; Groves & Tanner, 1995; Wang, Groves, Mawby & Tanner, 1997). The backbone of the exchanger was interrupted within different intra- and extracellular loops and the fragments were assayed either separately or in combination for function and plasma membrane expression. In agreement with our findings the band 3 single fragments were retained in intracellular compartments. The coexpression of pairs of fragments in Xenopus laevis oocytes resulted in functional anion exchangers if the protein backbone was not interrupted within the first five transmembrane domains. Comparable findings have been reported for rhodopsin (Ridge, Lee & Yao, 1995; Ridge, Lee & Abdulaev, 1995) and the rat m3 muscarinic acetylcholine receptor (Schöneberg, Liu & Wess, 1995). Other examples of eukaryotic bipartitioned proteins, the yeast α-factor transporter (STE 6, Berkower & Michaelis, 1991), an adenylylcyclase (Tang, Krupinski & Gilman, 1991), a sodium channel (Stühmer et al. 1989), the glucose transporter GLUT1 (Cope, Holman, Baldwin & Wolstenholme, 1994), and CFTR (Ostedgaard, Rich, Deberg & Welsh, 1997) have symmetrical structures and the cuts were inserted only in between the different subunits, resulting in functional combinations. The plasma membrane delivery of the non-functional mutants has not been investigated within this context. However, the expression of the N-terminal half of CFTR alone resulted in very low Cl⁻ channel activity (Sheppard, Ostedgaard, Rich & Welsh, 1994).

The in vitro translation experiments prove the integrity of the different cRNAs resulting in correctly translated protein fragments. However, none of the truncated transporters or the 1-5 plus 6-8 combination was efficiently processed in Xenopus oocytes. Residual fragments that supposedly reach the membrane do not amount to a detectable level. Whether incorrect insertion into the membrane, unstable tertiary structure of the fragments or exposed residues cause degradation is unknown. The coexpression experiments revealed that the complementing fragments 1-3 plus 4-8 could stabilize each other resulting in proper membrane delivery. This implies a direct interaction of the two cognate constructs and correct folding of the individual fragments. This assumption is supported by the functional integrity of the combined fragments 1-3 and 4-8. The two fragments aggregate and form a functional protein with identical functional characteristics compared with the flounder Na⁺-P_i cotransporter. Inefficient membrane delivery or incomplete aggregation could account for the observed difference in maximal transport rate (Fig. 4). However, an influence of the cleavage on v_max cannot be excluded. Nevertheless, our data are in agreement with a functional, bipartite transporter in vivo. Taken together our results and the above cited examples of coexpressed fragments indicate a specific interaction between the different hydrophobic domains of membrane proteins, often strong enough to maintain functional integrity. The absence of activity seems mostly due to impaired trafficking of the individual fragments.

The two immunohistochemical approaches for locating the translation products resulted in information about the topology of NaP_i-II. The absence of a fluorescent signal with the intact oocytes expressing NaP_i-II tagged at both the N- and the C-terminus locates both termini in the intracellular space. The highly positive staining obtained with the C-tagged 1-3 and/or N-tagged 4-8 fragment proves the reliability of the method. The large extracellular loop was shown to be glycosylated and, consequently, postulated to face the extracellular space (Hayes et al. 1994). The reported results support the proposed topological model of NaP_i-II so far, but more experimental data are required. Information concerning this question was expected from the combined expression of fragments 1-5 and 6-8. The rapid degradation of the two fragments could indicate a special function of the cut loop in stabilizing the protein. Alternatively, one or both of the parts could be misfolded inhibiting the interaction of the two parts necessary for correct membrane delivery.

Xenopus laevis oocytes represent a versatile expression system for eukaryotic membrane proteins. In combination with the two immunohistochemical methods described above, epitopes can be unambiguously confined to the intra- or extracellular space. The protein is expressed in vivo, therefore avoiding the folding artefacts often observed with in vitro translation systems. The reported strategy represents a highly reliable way of investigating the topology of membrane proteins. The fact, that a correctly folded transporter - here we assume that only proteins with the native structure reach the plasma membrane - may be assembled from different individual fragments opens up a whole new possibility for structural investigations of membrane proteins. So far, one of the bottle-necks in this field has been the difficulty in obtaining high level expression of proteins with several membrane-spanning domains. The observed self-assembly of two fragments into a functional protein may lead to an approach with individually expressed non-functional fragments followed by in vitro reconstitution of the entire protein.