SEARCH

SEARCH BY CITATION

Keywords:

  • arrestin;
  • C5a;
  • constitutive activation;
  • endocytosis

Abstract

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

Serpentine receptors relay hormonal or sensory stimuli to heterotrimeric guanine nucleotide-binding proteins (G proteins). In most G protein-coupled receptors (GPCRs), binding of the agonist ligand elicits both stimulation of the G protein and endocytosis of the receptor. We have begun to address whether these responses reflect the same sets of conformational changes in the receptor using constitutively active mutants of the human complement factor 5a receptor (C5aR). Two different mutant receptors both constitutively activate G protein-mediated responses, but one (F251A) is endocytosed only in response to ligand stimulation, while the other (NQ) is constitutively internalized in the absence of ligand. Both the constitutive and ligand-dependent endocytosis are accompanied by recruitment of beta-arrestin to the receptor. An inactivating mutation (N296A) complements the NQ mutation, producing a receptor that is activated only upon exposure to agonist; this revertant receptor (NQ/N296A) is nevertheless constitutively endocytosed. Thus one mutant (F251A) requires agonist for triggering endocytosis but not for activation of the downstream G protein signal, while another (NQ/N296A) behaves in the opposite fashion. Dissociation of two responses normally dependent on agonist binding indicates that the corresponding functions of an activated GPCR reflect different sets of changes in the receptor's conformation.

Serpentine receptors, a family of ligand-activated molecular switches, relay many different extracellular stimuli to heterotrimeric (αβγ) G proteins located on the cytoplasmic face of the plasma membrane. These receptors promote exchange of GDP for GTP bound to the α subunit of the heterotrimer, allowing the Gα-GTP and βγ subunits to separate and subsequently to activate intracellular effectors (1). Patterns of evolutionarily conserved amino acids distinguish six separate families of serpentine receptors (2), of which the rhodopsin-like family (several hundred members) is by far the largest (3). Mammalian serpentine receptors share with their orthologs in plants and yeast a conserved three-dimensional architecture with seven transmembrane α-helices arranged in a bundle (4). The activation mechanism of all these receptors must also be conserved, because mammalian receptors can activate yeast G proteins (5–7). The switch itself resides in the transmembrane helices: swapping extra- or intracellular loops between receptors changes specificity of ligand-binding or G protein-coupling, respectively, but leaves intact the capacity for ligand-dependent receptor activation, as summarized in (8).

Despite reports of the functional effects of an enormous number of mutations in many serpentine receptors, it has proved difficult to draw strong inferences about the conserved switch mechanism, in part because relatively few positions have been mutated in any one receptor (9). In addition, a receptor switch can exist in more than two positions, ‘off’ and ‘on’(10–13). The traditional view was that the active receptor conformation that stimulates one effector is identical to the conformation that stimulates a second effector or produces other agonist-dependent events, such as receptor endocytosis. This view is almost certainly an oversimplification; observations suggesting multiple active conformations have been reported from experiments with the beta-2 adrenergic receptor (14–16), the angiotensin II receptor (17), the N-formyl peptide receptor (18), and several chemokine receptors (18–22). Moreover, studies with the mu opioid receptor (MOR) have indicated that the MOR may have different ligand-selective activated conformations and that these conformations may be clinically relevant (23, 24).

In an attempt to identify G protein-coupled receptor (GPCR) residues that play key roles in the agonist-regulated switch, we subjected transmembrane helices of the receptor for complement factor 5a (the C5aR) to a comprehensive genetic analysis in yeast (25). This approach identified an essential cluster of evolutionarily conserved and functionally important residues in helices III, V, VI, and VII, located in the transmembrane core of the receptor and in close proximity to one another. Now a more detailed characterization of several mutations at the core positions, performed in the context of mammalian cells, shows that two different functions of an activated GPCR – agonist dependence of signaling to the downstream G protein and agonist dependence of endocytosis – can be completely dissociated. We therefore infer that these two signaling events depend on different sets of conformational changes in the GPCR.

Results

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

Constitutively active C5aRs in yeast and mammalian cells

We assayed activities of wild-type (WT) and mutant C5aRs in the yeast strain BY1142 in the presence or absence of the agonist, C5a. As previously described (25), endogenous expression of both the C5a ligand and the WT C5aR allows cells to grow in selective medium with up to 5 mm AT, while cells expressing the WT C5aR alone do not proliferate at concentrations of 0.5 mm AT or higher (Table 1). With agonist present, the activity of the F251A mutant C5aR was equal to that of the WT receptor, while the I124N/L127Q mutant (‘NQ’, with mutations that replace two amino acids) was more active. Both the F251A mutant and the NQ mutant showed markedly elevated activity in the absence of C5a (Table 1); in the absence of the other mutation, neither the I124N nor the L127Q mutation produced a constitutively active receptor, although both signaled robustly in response to C5a.

Table 1. : Analysis of signaling strength for C5aR constructs expressed in yeast. Yeast strain BY1142 was cotransformed with a plasmid encoding the WT or indicated C5aR mutants together with a plasmid encoding either vector alone (– C5a) or an α-factor prepro/C5a ligand (+ C5a), exactly as described (25). Receptor signaling was assayed by growth on histidine-deficient media in the presence of AT: + + + + +, growth on 10 mm AT; + + + +, growth on 5 mm AT; + + +, growth on 2 mm AT; + +, growth on 1 mm AT; +, growth on 0.5 mm AT; 0, no growth on 0.5 mm AT
Construct+ C5a– C5a
  • a

    WT: wild type;

  • b

    NQ: I124N/L127Q.

WTa+ + + +0
F251A+ + + ++ + +
NQb+ + + + ++ + + +
I124N+ + + + ++
L127Q+ + + + ++
N296A00
NQ/N296A+ + + +0

Mapping the amino acid sequence of the C5aR onto the crystal structure of a homologous GPCR, rhodopsin, shows that F251 (in helix VI) and I124 and L127 (both in helix III) point toward one another in the receptor's transmembrane core (Figure 1). All three residues are located at approximately the same depth in the lipid bilayer, at a level about two-thirds of the way from the receptor's extracellular to its intracellular face.

image

Figure 1. Locations of mutated residues in the C5aR. a. A model of the transmembrane, helical domains of the C5aR based on the crystal structure of rhodopsin (50) as seen from the extracellular side. Helices are gray, and the mutated residues are colored (I124 purple, L127 red, F251 yellow, and N296 cyan). All four mutated residues are buried deeply within and oriented inside the transmembrane pocket, which is formed by the seven helices. A semi-transparent surface is shown for the C5aR. b. The same as in panel a, except that the C5aR is turned by 90 degrees, so the viewer sees the receptor from the side with the extracellular ends of the helices at the top.

Download figure to PowerPoint

As environments for C5aR expression and function, yeast and mammalian cells differ in many respects, including membrane phospholipid composition, receptor-G protein stoichiometry, and specific G protein subunits with which the receptor must interact. Accordingly, we assessed total inositol phosphate (IP) production in COS-7 cells coexpressing recombinant Gα16 and a WT or a mutant C5aR (Figure 2a). Relative activities of WT and mutant receptors were similar in yeast and COS-7 cells (Table 1 vs. Figure 2a). Acting on the WT C5aR, C5a (10 nm) induced an eight-fold increase in intracellular IP (Figure 2a). As in yeast, both the NQ and the F251A mutants induced constitutive, agonist-independent IP responses (3- and 2-fold greater, respectively, than that seen with WT in the absence of C5a). Both mutants were further activated by ligand to stimulate IP accumulation similar to (NQ) or even higher than (F251A) that seen with the WT receptor in response to C5a. Although Gα16 levels were not measured directly, data obtained from duplicate determinations from multiple transfections suggest that the differences in constitutive activity seen in COS-7 cells were not due to variable levels of Gα16 expression but reflect the intrinsic properties of the C5aRs.

image

Figure 2. Activity and binding affinity of WT and mutant C5aRs. a. Constitutive (gray bars) or ligand-dependent (black bars) inositol phosphate production mediated by receptor mutants. We coexpressed a plasmid encoding Gα16 with and without the receptor constructs in COS-7 cells and measured accumulated phosphoinositides. Bars represent means ± range of duplicate determinations. Each result is representative of three independent transfection experiments. b. Saturation binding isotherms of the WT C5aR, and the NQ and F251A mutants. Determinations were performed in triplicate. Each curve is representative of at least three independent experiments.

Download figure to PowerPoint

C5a binding and receptor location in mammalian cells

As assessed by measuring binding of [125I]-C5a to membrane preparations of COS-7 cells transiently expressing the recombinant receptors (Figure 2b), agonist affinities of the WT and the F251A mutant receptors were virtually identical (Kd ∼ 1 nm; summarized in Table 2); this value is in good accordance with previous observations with the WT C5aR (26,27). The NQ mutant displayed an eight-fold increase in agonist affinity, consistent with the effects of constitutively activating mutations in other serpentine receptors (28,29).

Table 2. : Analysis of saturation binding for C5aR constructs. Apparent binding affinities (Kd) and receptor plasma membrane expression (Bmax) were obtained from binding curves for [125I]-C5a (see Experimental Procedures and Figure 1). Values represent means ± SE from three independent experiments performed in duplicate
ConstructKdBmax
nMpmol/mg
  • a

    WT: wild type;

  • b

    NQ: I124N/L127Q; nd: not detectable.

WTa1.05 ± 0.119.4 ± 2.9
F251A0.84 ± 0.289.2 ± 3.9
NQb0.13 ± 0.021.8 ± 0.7
N296Andnd
NQ/N296A0.92 ± 0.590.3 ± 0.1

As estimated by Bmax values for [125I]-C5a binding (Table 2), the F251A mutant when expressed transiently was expressed on membranes at a concentration comparable to that of the WT C5aR. In contrast, the NQ mutant receptor was consistently expressed at a reduced level when the same concentration of DNA was used for the transfection (Table 2, Figure 2b). These results suggested that the WT, F251A and NQ receptors may differ in cellular distribution and/or trafficking.

Accordingly, we assessed cellular distribution of WT and mutant C5aRs. HEK293 cells were stably transfected with HA-tagged versions of WT, F251A or NQ mutant receptors. Selection of stable clones allowed cell lines to be selected that stably expressed the wild-type and mutant receptors at comparable levels (within two-fold). Cells were treated with C5a (or left untreated), permeabilized, and stained for receptor using antibodies to the epitope tag (see Experimental Procedures). Both the WT and the F251A mutant receptor were expressed predominantly on the cell surface in the absence of agonist, with a small but detectable amount of receptor visible intracellularly (Figure 3a, top panels). Both WT and F251A rapidly redistributed to small, punctate intracellular compartments following C5a treatment, suggesting the receptors had been internalized in response to agonist (Figure 3a, bottom panels). In contrast, in both the presence and absence of agonist, the NQ mutant receptor was expressed throughout the cell primarily in both small and large intracellular structures (Figure 3a, right panels). Although these data suggested that the NQ receptor was undergoing constitutive endocytosis, staining permeabilized cells for receptor does not definitely distinguish potentially misfolded NQ receptor retained in the secretory pathway from receptor that had been internalized in the absence of agonist.

image

Figure 3. Receptor distribution and endocytosis. a. Distribution of total receptor was assessed in permeabilized cells stably expressing HA-tagged WT, F251A and NQ receptors in the presence or absence of agonist (1 μm C5a, 30 min). Cells were permeabilized prior to treatment with HA antibody. Therefore, all receptors were detected. Bar represents 10 μm. b. Cells stably expressing HA-tagged WT, F251A and NQ receptors were treated with antibody to the extracellular epitope tag (antibody ‘feeding’) for 30 min, fixed and permeabilized, and receptor distribution was assessed. Cells were permeabilized after treatment with HA antibody to detect only receptors that had been on the cell surface. c. Receptor endocytosis was evaluated in the presence or absence of agonist using the biotin protection assay, where internalized receptors are protected from cleavage by membrane-impermeant reducing agent. Experiments were performed in at least two independent cell lines for each receptor type. HEK293 cells were used for this analysis because receptor distribution was easier to assess than in COS-7 cells; receptor distributions in COS-7 generally resembled those seen in HEK293 (data not shown).

Download figure to PowerPoint

The NQ receptor is constitutively internalized

To ask whether the NQ receptor was internalized constitutively, we utilized both a modified version of the above immunofluorescence assay and a quantitative biochemical assay to directly examine constitutive endocytosis. For the immunofluorescence assay, live cells stably expressing epitope-tagged WT, F251A or NQ receptors were incubated with antibody directed to the N-terminal epitope tag. After 30 min, cells were fixed, permeabilized, and stained with secondary antibody for receptor (see Experimental Procedures). This experimental design allowed us to assess distributions of receptors that were accessible to the HA antibody during the 30-min incubation. We found that both the WT and F251A receptors remained almost exclusively on the plasma membrane during the 30-min incubation with antibody (Figure 3b), consistent with their plasma membrane distribution in permeabilized cells (Figure 3a). In contrast, the antibody-tagged NQ receptor was distributed into intracellular compartments within this 30-min time-period (Figure 3b, right panel), indicating that the receptors were endocytosed in the absence of agonist. This ‘antibody feeding’ experiment could not be utilized to assess agonist-induced endocytosis, because our antibody interfered with ligand binding to the N-terminal receptor domain (30) (data not shown).

To further assess endocytosis of the three receptors, we used a previously described biotin protection assay (31). Briefly, cells stably expressing receptor were incubated with thio-cleavable biotin to selectively label membrane proteins, including receptors, on the cell surface. Cells were treated with agonist or left untreated, and then residual biotin was stripped using a membrane-impermeant reducing agent. Cells were lysed and receptors immunoprecipitated with antibodies directed to the epitope tag. Receptors that were protected from cleavage because they had been endocytosed were visualized by streptavidin overlay. This assay showed that both WT and F251A mutant receptors were significantly internalized in response to C5a (Figure 3c); in the absence of agonist (NT), neither of these receptors demonstrated significant constitutive endocytosis (Figure 3c, compare NT lane to strip lane). In contrast, the NQ mutant receptor showed significant constitutive endocytosis (Figure 3c). Indeed, the amount of NQ receptor endocytosed in the presence of agonist was indistinguishable from that internalized in the absence of agonist (Figure 3c, compare NT to + C5a lanes).

Recruitment of arrestin

We next asked whether the cellular mechanism that mediates constitutive endocytosis of the NQ receptor resembled that responsible for agonist-dependent endocytosis of the WT and F251A receptors. Following their activation, many serpentine receptors are regulated by associating with beta-arrestin, which has been proposed to act as a scaffolding protein for numerous signal transduction pathways (32,33), as well as an adapter protein to facilitate the recruitment of activated receptors to clathrin-coated pits (34). In fact, recruitment of GFP-tagged arrestin has served as an accurate marker of receptor activation for numerous serpentine receptors (35). Consequently, we asked whether arrestin was recruited during endocytosis of the WT and mutant C5aRs.

We generated stable cell lines that expressed both a GFP-tagged version of beta-arrestin-2 (35) and either WT, F251A or NQ receptors, each tagged with the HA epitope. Cells were treated with agonist or left untreated, and receptor and arrestin distributions were examined in permeabilized cells (see Experimental Procedures). In the absence of agonist, arrestin was distributed throughout the cytoplasm in cells expressing the WT (Figure 4a) or the constitutively activated F251A receptor (Figure 4b), consistent with the plasma membrane localization of these receptors in the absence of agonist. Application of agonist caused a rapid recruitment of arrestin to both the WT and F251A receptors (Figure 4a,b), and arrestin appeared to travel with the receptor into endosomal compartments (see merged panels in Figure 4a,b,c). Exposure of agonist causes some other serpentine receptors, such as the V2 vasopressin receptor (36), to accumulate in endosomes in association with arrestin; this phenotype contrasts with that of a number of other receptors, including the beta-2 adrenergic (37) and opioid receptors (38), which recruit arrestin at the coated pit but do not appear to associate with arrestins after translocation to endosomes. The constitutively active, constitutively internalized NQ receptor mediated rapid recruitment of arrestin even in the absence of agonist (Figure 4c). These results suggest that arrestin recruitment can mediate C5a-induced endocytosis of WT and F251A receptors as well as constitutive endocytosis of NQ receptors.

image

Figure 4. Arrestin and C5aRs. a, b, c. Cell lines stably expressing a GFP-tagged arrestin (35) and either WT (a), F251A (b) or NQ (c) receptor were treated with agonist for 30 min, fixed, permeabilized, and stained for receptor; distributions of arrestin and receptor were assessed by immunofluorescence microscopy. Bar represents 10 μm. d. Cells stably overexpressing GFP-arrestin and F251A receptor or F251A receptor alone (2 independent cell lines for each) were analyzed for receptor endocytosis in the absence (NT) or presence (+ C5a) of agonist (1 μm C5a, 60 min) using the biotin protection assay. A representative blot is shown. e. Cells overexpressing an EE-tagged beta-arrestin-1 and the mu opioid receptor (MOR) (31), or the MOR alone were analyzed for receptor endocytosis is the absence (NT) or presence of the agonists D-Ala2-MePhe4-Gly5-ol (DAMGO, DG) or morphine (MS).

Download figure to PowerPoint

The recruitment of arrestin to many GPCRs is mediated by phosphorylation by G protein-coupled receptor kinase (GRK). Although we hypothesize that the phosphorylation profile of the C5a receptor mutants would mirror arrestin recruitment, we did not specifically assess this in our cell lines. Hence it is possible that arrestin recruitment to the C5a receptor is phosphorylation independent.

Arrestin overexpression does not enhance endocytosis of F251A

Taken together, these results suggest that the portions of NQ and F251A receptors that interact with G proteins versus the endocytic machinery differ in conformation from one receptor to the other, even though both receptors signal constitutively. In the absence of agonist, the NQ receptor's conformation allows it to recruit the endocytic machinery, includingarrestin, while that of the F251A receptor does not. An alternative explanation is that the NQ receptor simply spends more time in an active conformation than does the F251A receptor; in this scenario, NQ would recruit arrestin more efficiently because it has more time to do so. To distinguish between these possibilities, we asked whether overexpression of arrestin could facilitate constitutive endocytosis of F251A by increasing the local concentration of arrestin, thereby decreasing the amount of time that the receptor would be required to be in the active conformation for efficient recruitment. The data in Figure 4(b) had suggested that arrestin overexpression did not, in fact, increase F251A constitutive endocytosis. We confirmed this observation by quantifying the amount of F251A endocytosis in the presence or absence of stable arrestin overexpression, using the biotin protection assay (see Experimental Procedures). F251A endocytosis in the presence or absence of agonist was quantitatively indistinguishable in cells that stably overexpressed arrestin vs. cells that expressed only endogenous arrestin (Figure 4d). In contrast, as previously reported (24), arrestin overexpression facilitated mu opioid receptor (MOR) endocytosis in response to morphine (MS), which did not promote receptor endocytosis under conditions where arrestin is expressed at endogenous levels (23) (Figure 4e).

Pertussis toxin does not block either constitutive or ligand-mediated endocytosis

Our signaling assays showed that the NQ mutant receptor was more constitutively active than was the F251A receptor (Table 1 and Figure 2a). To ask whether this difference in signaling capacity per se was responsible for the different susceptibilities of the NQ and F251A receptors to agonist-independent endocytosis, we pretreated cells stably expressing these receptors with pertussis toxin, to inhibit Gi-dependent signaling. The toxin inhibited C5a-induced signaling, as assessed by whole cell cAMP accumulation (data not shown), but did not inhibit either ligand-induced or constitutive endocytosis of the receptors, assessed by the biotin protection assay (Figure 5). In fact, PTX slightly enhanced the amount of protected receptor in this assay. We believe that this increase in signal may be attributed to a decrease in postendocytic degradation of the NQ receptor in the PTX-treated cells and are investigating this observation further.

image

Figure 5. Effect of pertussis toxin (PTX) on receptor endocytosis. Cells stably expressing F251A or NQ mutant receptor were treated with PTX (100 ng/ml, 18 h), or left untreated. Receptor endocytosis was assessed using the biotin protection assay, in which internalized receptors are protected from cleavage by membrane-impermeant reducing agent. These results are representative of at least two independent experiments.

Download figure to PowerPoint

Reversion of NQ constitutive activity does not alter constitutive endocytosis

The asparagine at position 296 in helix VII of the C5aR is conserved at the corresponding position in almost all rhodopsin-like serpentine receptors (3), and was also essential for signaling function in our genetic analysis of the C5aR (25). Replacing N296 with alanine abrogated receptor signaling in yeast (Table 1). The N296A receptor was poorly expressed when transiently expressed in mammalian cells, and [125I]-C5a binding to this receptor could not be measured. Combining the N296A and the NQ mutations, however, produced a receptor with activity qualitatively similar to that of the WT receptor both in yeast (Table 1) and in COS-7 cells (Figure 6a). Specifically, this mutation reverted the NQ phenotype so that constitutive activity was abolished, while ligand-induced activity remained intact. As with NQ, the Bmax of C5a binding in the double mutant (NQ/N296A) was significantly lower than that of WT C5a receptor when it was transiently expressed; hence, it is possible that the low expression of this receptor could account for its failure to be detected as constitutively active. However, mutations at this position adversely affect the activity of several GPCRs, suggesting that it is critical for receptor activation. The binding affinity of NQ/N296A was similar to that of the WT receptor (Table 2 and Figure 6b). This result, paired with the decreased expression seen in the transiently transfected cells, raised the possibility that the mutation that reverted the constitutive activity of NQ did not revert the constitutive endocytic phenotype.

image

Figure 6. Phenotype of the NQ/N296A mutant receptor. a. Constitutive (gray bars) or ligand-dependent (black bars) inositol phosphate production mediated by receptors. We coexpressed a plasmid encoding Gα16 with the receptor constructs in COS-7 cells and measured accumulated phosphoinositides. Bars represent means ± range of duplicate determinations. Each result is representative of three independent transfection experiments. b. Saturation binding isotherm of NQ/N296A. Determinations were performed in triplicate. Curve is representative of at least three independent experiments. c. Cells stably expressing HA-tagged WT, NQ and NQ/N296A receptors were treated with antibody to the extracellular epitope tag for 30 min (antibody ‘feeding’), fixed, and permeabilized, and receptor distribution was assessed. Cells were permeabilized after exposure to HA antibody to ensure that only receptors that were once on the cell surface would be detected. The cells expressing NQ/N296A that also expressed GFP-arrestin (far right panel) showed arrestin recruitment in an agonist-independent manner. d. Receptor endocytosis was evaluated in the presence or absence of agonist using the biotin protection assay, in which internalized receptors are protected from cleavage by membrane-impermeant reducing agent. Experiments for panels c and d were performed in two independent cell lines for each receptor.

Download figure to PowerPoint

To ask whether reverting constitutive signaling activity of the NQ receptor by combining it with the N296A mutation concomitantly affected the receptor's endocytosis, we examined NQ/N296A receptor distribution in HEK293 cells stably expressing an HA-tagged version of this receptor, with the same antibody feeding protocol used in Figure 3 (see Experimental Procedures). Cell lines were selected that expressed NQ/N296A receptor at levels comparable to the other receptors (within two-fold). Despite its lack of constitutive activity, the NQ/N296A receptor was still efficiently internalized in the absence of agonist (Figure 6c). This receptor also recruited GFP-arrestin in the absence of agonist (Figure 6c, far right panel). The biotin protection assay confirmed that constitutive endocytosis of the NQ/N296A receptors was indistinguishable from that of the NQ receptor (Figure 6d). Moreover, despite the agonist dependence of activation for this receptor, C5a did not significantly stimulate endocytosis (Figure 6d; compare NT and + C5a lanes).

Discussion

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

Some protein signaling machines work as simple off/on switches, but many can assume multiple different activated conformations, which allows them to generate more complex patterns of response to input stimuli. Examples include combinatorial integration of multiple stimuli [e.g. coincidence detectors like N-WASp (39) or NMDA-sensitive ion channels (40, 41)] and the qualitatively distinct output signals generated by entry of chemically different ligands into the binding pocket of a receptor [e.g. the estrogen receptor (42, 43)]. Our experiments with mutant C5aRs and observations of several other GPCRs (discussed further below) suggest that GPCRs belong in the last category: that is, GPCRs can assume multiple conformations to generate qualitatively different outputs.

In the absence of ligand, the WT C5aR and the three mutants present all four possible phenotypic combinations of constitutive signaling via G proteins and constitutive endocytosis: WT receptors show neither constitutive G protein signaling nor constitutive endocytosis (S/E), while the NQ, NQ/N296A combination, and F251A phenotypes are S+/E+, S/E+, and S+/E, respectively. The mutations that dissociate constitutive activation from constitutive endocytosis do not grossly affect receptor folding and transport to the plasma membrane, reduce affinities for binding C5a, or inhibit ability to signal to G proteins. Moreover, the mutants show maximal rates and extents of endocytosis (constitutive or in the presence of agonist, depending on the receptor) comparable to those seen with the agonist-bound WT receptor, and all recruit arrestin.

We imagine that the C5aR WT and mutant phenotypes represent four distinct molecular conformations. These conformations are determined by amino acid substitutions in the core of the helix bundle (Figure 1), distant (∼ 12–18 Å) from the receptor's cytoplasmic loops. These loops contact the trimeric G protein (8) and the endocytic machinery, including beta-arrestin (44), but the mutated residues do not. Thus the mutations – like agonist binding to a pocket oriented toward the extracellular fluid – effect conformational change indirectly, and at a distance. It is not clear how mutations at positions close to one another in the helix bundle produce such strikingly distinct functional phenotypes. It is worth noting, however, that amino acid residues cognate to F251, I124, L127, and N296 are highly conserved in other GPCRs of the rhodopsin family, and mutations at these positions often produce constitutive signaling activity (like NQ and F251A) or inactivate the receptor (like N296A) [(2,3) and references therein]. Conservation of primary structure and sensitivity to mutation in this region are consistent with the idea that this core region plays an important role in flipping the conformational switch, and/or modulating its output, in most GPCRs.

Our results and accumulating evidence from several laboratories suggest that the simple two-state (off/on) model of the GPCR switch is not correct. Experiments with opioid (23,24), beta-2 adrenergic (14–16), angiotensin II (17), and the N-formyl peptide (18) receptor have suggested that a single receptor can take on multiple distinct conformations. Strictly speaking, the C5aR mutant phenotypes do not by themselves indicate that the WT C5aR itself can take on multiple conformations in response to binding C5a; rather, each C5aR mutation may selectively promote a conformation that represents a subset of the overall conformation that normally accompanies activation by agonist. The fact that subtle alterations in the presumptive core switch of the C5aR produce different activated conformations does suggest, however, that different agonist ligands for a GPCR might flip the activation switch to produce multiple conformations.

Indeed, differential signaling upon stimulation by different ligands has been documented for several chemokine receptors, which, together with the N-formyl peptide receptor, are the closest relatives of the C5aR (45). Thus experiments with monoclonal antibodies raised against CC-chemokine receptor 5 that recognize different epitopes show profound differences in terms of receptor signaling, endocytosis, dimerization, competitive binding, and HIV-coreceptor function (22). The chemokines themselves are also able to provoke differing responses: CC-chemokine ligand 21 (CCL21) and CCL19 activate CCR7 signaling with comparable efficiency, but only CCL19 promotes internalization of CCR7 (21). Similar findings were reported for the actions of CCL5 and an N-terminal derivative thereof on another receptor, CCR1 (19,20). Arguments for two-state vs. multistate models of receptor activation have been thoroughly discussed (10–12).

Although our constitutively activating mutants provide useful experimental tools to dissect receptor activation from endocytosis, no naturally occurring, activating C5aR mutations have been reported. In other receptors, however, activating mutations can produce diseases. For example, mutations in the thyrotropin and luteotropin receptors cause thyrotoxicosis (46) and premature puberty (47), respectively. Curiously, the apparently activating R137H mutation in the V2 vasopressin receptor induces a loss-of-function phenotype, familial nephrogenic diabetes insipidus (48). In this case the mutant receptor – which shows constitutive signaling activity in vitro– undergoes constitutive arrestin-dependent desensitization and internalization. This V2 mutant phenotype thus resembles that produced by the NQ mutation in the C5aR, except that constitutive internalization of the V2 receptor in vivo removes so much receptor from the cell surface that signaling is abolished. In principle, mutations of serpentine receptors – like F251A in the C5aR – that activate G protein signaling without promoting constitutive regulation by arrestins and endocytosis could be even more dangerous physiologically, because their ‘on’ signal would not be counteracted by endocytosis and disappearance from the cell surface.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

Construction of receptor mutants

Mutated C5a receptors (C5aRs) were created using the polymerase chain reaction with one mutagenizing and one wild-type (WT) sequence primer flanking either end of sequences coding for helices III, VI, and VII. Products were digested with appropriate restriction enzymes, subcloned into the C5aR gene in pBS-SK Bluescript vector and sequenced. For yeast or COS-7 cell assays, the various C5aR sequences were subcloned into plasmid p1303 (p1303ADE2, PGK-hC5aRADE2 REP3 2 μm-ori AmpR flori, Cadus Pharmaceuticals) or into plasmid pCDM8 (Invitrogen, Carlsbad, CA, USA), respectively. For generation of HEK293 stable cell lines for endocytosis assays, receptors were subcloned into pCDNA3.1-Zeo and tagged with HA on the N-terminus.

Yeast strains, transformation and signaling assays

Analysis of receptor signaling was performed by replica plating onto different concentrations of 3-aminotriazole (AT), as previously described (25).

Mammalian cell culture and transfection

HEK293 cells were maintained in Dulbecco's modified Eagle's medium with 10% fetal calf serum. Stable cell lines expressing receptor and/or GFP-beta-arrestin (35) were generated by calcium phosphate coprecipitation followed by selection for single colonies with Zeocin (Invitrogen) for receptors and G418 for GFP-beta-arrestin. COS-7 cells were maintained in Dulbecco's modified Eagle's medium with 10% fetal calf serum, 100 μg/ml streptomycin sulfate, 100 units/ml penicillin G, and 10 μg/ml gentamicin. Transient transfections of WT and mutant receptors were performed using a DEAE-dextran/adenovirus method as described (49).

Membrane preparations

Membranes of COS-7 cells transfected with WT or mutant C5aR were prepared by a modification of a previously described method (49). Cells were harvested and lysed in 1 ×PBS pH 7.4, with 50 mm EDTA 2 mm DTT, protease inhibitors (phenylmethylsulfonyl fluoride, bacitracin and leupeptin), and homogenized by passing 10 times through a 27-gauge needle. The supernatant fraction of two successive centrifugations at 900 ×g for 10 min was centrifuged at 100 000 ×g for 30 min and the particulate membrane fractions were resuspended in lysis solution. All manipulations were performed on ice or at 4 °C, and membrane preparations were stored at − 70 °C.

Binding assays

Membrane preparations expressing WT, NQ and F251A (3 μg total protein) or NQ/N296A (20 μg total protein) C5aR were incubated at 37 °C for 1 h with 25 pm (NQ) or 100 pm (all others) [125I]-C5a (2200 Ci/mmol, NEN, Boston, MA, USA) in 250 μl binding buffer [Hank's balanced salt solution supplemented with 25 mm HEPES pH 7.4 and 0.1% (wt/vol) bovine serum albumin]. Recombinant, non-radioactive C5a (Sigma, St Louis, MO, USA) was added to the indicated concentrations. Nonspecific binding was defined as the amount of radioactivity bound in the presence of 100 nm non-radioactive C5a. Incubations were terminated by vacuum filtration through presoaked GF/C filters (Whatman, Clifton, NJ, USA) and rapid washing with 6 ml ice-cold binding buffer. To determine binding affinities and capacities, binding data were analyzed by nonlinear regression analysis using Prism 2.0 (GraphPad Software, San Diego, CA, USA).

Activation assays

5 × 105 COS-7 cells cotransfected via adenovirus with 0.25 μg plasmid encoding Gα16 and with 0.25 μg plasmid encoding WT or mutant C5aRs were incubated overnight with 2 μCi [3H]-inositol (21 Ci/mmol, NEN, Boston, MA, USA), washed with assay medium (RPMI-1640 supplemented with 20 mm HEPES pH 7.4 and 5 mm LiCl), incubated with or without 10 nm C5a for 1 h at 37 °C, aspirated and incubated with 750 μl 20 mm cold formic acid at 4 °C for 30 min, and adjusted with 100 μl of solution I (6 ml concentrated ammonium hydroxide/l). Poly-Prep chromatography columns (BIO-RAD, Hercules, CA, USA) with 1 ml AG 1-X8 resin 100–200 mesh (BIO-RAD, Hercules, CA, USA) were equilibrated with 10 ml of solution II (4 m ammonium formate, 0.2 m formic acid), followed by 5 ml solution III (10-fold dilution of solution II). Samples were loaded, columns eluted with 1 ml solution III and the resulting inositol fractions collected. Columns were washed with 4 ml solution IV (40 mm ammonium formate, 0.1 m formic acid, eluted with 1 ml of solution II and the resulting inositol and inositol phosphate (IP) fractions collected. Activation of Gα16 was assessed as [IP]/([IP] + [total inositol]) ×100%.

Biotin protection assay

Assays were performed as previously described (31), including some modifications for the C5aRs. Briefly, cells stably expressing WT or mutant C5aR were grown to 80% confluency in 10-cm plates, washed with PBS, and treated with 3 μg/ml disulfide-cleavable biotin (Pierce) in PBS at 4 °C for 30 min. Cells were washed in PBS and a plate designated as ‘100%’ and a plate designated as ‘strip’ remained at 4 °C. The other plates were placed in prewarmed media for 15 min, prior to treatment with C5a (1 μm, 60 min) or no treatment (60 min). Following ligand treatment, plates were washed in PBS, and remaining cell surface biotinylated receptors stripped from all plates except the 100% plate in 50 mm glutathione, 0.3 m NaCl, 75 mm NaOH, and 1% fetal bovine serum at 4 °C for 30 min. All plates were quenched, cells extracted in 0.1% Triton X-100, 150 mm NaCl, 25 mm KCl, and 10 mm Tris HCl, pH 7.4, and cellular debris removed by centrifugation at 10 000 ×g for 10 min at 4 °C. Lysates were immunoprecipitated with anti-HA (HA-11 Covance), rabbit anti-mouse linker antibody (Jackson Immunoresearch, Westbrove, PA, USA), and Protein A sepharose (Pharmacia) overnight. Immunoprecipitates were washed 5 times and incubated with PNGase F (Amersham for 2 h. Samples were then denatured in SDS sample buffer with no reducing agent and resolved by SDS/PAGE, transferred to nitrocellulose, biotinylated proteins exposed to Vectastain ABC immunoperoxidase reagent (Vector Laboratories), and developed with ECL reagents (Amersham).

Immunofluorescence assays

Receptor distribution was evaluated in two ways. First, to assess total receptor distributions, cells stably expressing HA-tagged receptors were treated with agonist (C5a 1 μm, 30 min) or left untreated. Cells were fixed in 4% formaldehyde in PBS, permeabilized in 0.1% TX100, and receptors stained with antibodies directed against the epitope tag (HA-11, Covance, 1 : 1000, 30 min), followed by staining with Cy3-conjugated donkey anti-mouse secondary antibody (Jackson Immunoresearch, 1 : 1000, 30 min). When assessing both receptor and GFP-arrestin distribution, cells were treated as above, except that receptors were stained with Texas-red conjugated donkey anti-mouse antibody (Jackson Immunoresearch, 1 : 500), to avoid bleed-through into the GFP channel. Second, to assess distribution/redistribution of cell surface receptor, an antibody feeding protocol was used. Cells stably expressing HA-tagged receptor were fed the antibody to the extracellular tag (HA-11, Covance, 1 : 1000) for 30 min at 37 °C in DMEM to label a pulse of extracellular receptors in living cells. Cells were then fixed and permeabilized as above and receptor was stained with Cy3-conjugated donkey anti-mouse secondary antibody (Jackson Immunoresearch 1 : 1000).

Molecular modeling

Our model of the C5aR was based on the crystal structure of rhodopsin (50). Only the transmembrane helices were used; side chains corresponding to the C5aR sequence were substituted for the rhodopsin side chains with the program SCWRL (51), which uses a backbone-dependent rotamer library. Otherwise, the original rhodopsin coordinates were retained. Protein display was performed in Swiss Pdb Viewer (52), using a semi-transparent surface representation.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

We thank Maria Waldhoer and Dorit Ron for critical reading of the manuscript. This work was supported in part by National Institutes of Health Grant GM-27800 (to H.R.B.), National Institutes of Health Grant DA10711 (to M.vZ.), and National Research Service Award DA05844 (to J.L.W.). B.O.G. is an Advanced Career Postdoctoral Fellow of the Swiss National Science Foundation. T.J.B. is a Howard Hughes Medical Institute Physician Postdoctoral Fellow.

References

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  • 1
    Hamm HE. The many faces of G protein signaling. J Biol Chem 1998;273: 669672.
  • 2
    Bockaert J, Pin JP. Molecular tinkering of G protein-coupled receptors: an evolutionary success. EMBO J 1999;18: 17231729.
  • 3
    Baldwin JM, Schertler GF, Unger VM. An alpha-carbon template for the transmembrane helices in the rhodopsin family of G-protein-coupled receptors. J Mol Biol 1997;272: 144164.
  • 4
    Josefsson LG. Evidence for kinship between diverse G-protein coupled receptors. Gene 1999;239: 333340.
  • 5
    Price LA, Kajkowski EM, Hadcock JR, Ozenberger BA, Pausch MH. Functional coupling of a mammalian somatostatin receptor to the yeast pheromone response pathway. Mol Cell Biol 1995;15: 61886195.
  • 6
    Price LA, Strnad J, Pausch MH, Hadcock JR. Pharmacological characterization of the rat A2a adenosine receptor functionally coupled to the yeast pheromone response pathway. Mol Pharmacol 1996;50: 829837.
  • 7
    Erickson JR, Wu JJ, Goddard JG, Tigyi G, Kawanishi K, Tomei LD, Kiefer MC. Edg-2/Vzg-1 couples to the yeast pheromone response pathway selectively in response to lysophosphatidic acid. J Biol Chem 1998;273: 15061510.
  • 8
    Bourne HR. How receptors talk to trimeric G proteins. Curr Opin Cell Biol 1997;9: 134142.
  • 9
    Horn F, Weare J, Beukers MW, Horsch S, Bairoch A, Chen W, Edvardsen, O, Campagne, F, Vriend G. GPCRDB: an information system for G protein-coupled receptors. Nucl Acids Res 1998;26: 275279.
  • 10
    Kenakin T. Pharmacological proteus? Trends Pharmacol Sci 1995;16: 256258.
  • 11
    Leff P, Scaramellini C, Law C, McKechnie K. A three-state receptor model of agonist action. Trends Pharmacol Sci 1997;18: 355362.
  • 12
    Clark RB, Knoll BJ, Barber R. Partial agonists and G protein-coupled receptor desensitization. Trends Pharmacol Sci 1999;20: 279286.
  • 13
    Schwartz TW, Rosenkilde MM. Is there a ‘lock’ for all agonist ‘keys’ in 7TM receptors? Trends Pharmacol Sci 1996;17: 213216.
  • 14
    Cheung AH, Dixon RA, Hill WS, Sigal IS, Strader CD. Separation of the structural requirements for agonist-promoted activation and sequestration of the beta-adrenergic receptor. Mol Pharmacol 1990;37: 775779.
  • 15
    Gether U, Lin S, Kobilka BK. Fluorescent labeling of purified beta 2 adrenergic receptor. Evidence for ligand-specific conformational changes. J Biol Chem 1995;270: 2826828275.
  • 16
    Seifert R, Gether U, Wenzel-Seifert K, Kobilka BK. Effects of guanine, inosine, and xanthine nucleotides on beta (2)-adrenergic receptor/G (s) interactions: evidence for multiple receptor conformations. Mol Pharmacol 1999;56: 348358.
  • 17
    Hunyady L, Bor M, Balla T, Catt KJ. Identification of a cytoplasmic Ser-Thr-Leu motif that determines agonist-induced internalization of the AT1 angiotensin receptor. J Biol Chem 1994;269: 3137831382.
  • 18
    Bennett TA, Maestas DC, Prossnitz ER. Arrestin binding to the G protein-coupled N-formyl peptide receptor is regulated by the conserved ‘DRY’ sequence. J Biol Chem 2000;275: 2459024594.
  • 19
    Rodriguez-Frade JM, Vila-Coro AJ, Martin A, Nieto M, Sanchez-Madrid F, Proudfoot AE, Wells TN, Martinez AC, Mellado M. Similarities and differences in RANTES- and (AOP)-RANTES-triggered signals: implications for chemotaxis. J Cell Biol 1999;144: 755765.
  • 20
    Elsner J, Mack M, Bruhl H, Dulkys Y, Kimmig D, Simmons G, Clapham PR, Schlondorff D, Kapp A, Wells TN, Proudfoot AE. Differential activation of CC chemokine receptors by AOP-RANTES. J Biol Chem 2000;275: 77877794.
  • 21
    Bardi G, Lipp M, Baggiolini M, Loetscher P. The T cell chemokine receptor CCR7 is internalized on stimulation with ELC, but not with SLC. Eur J Immunol 2001;31: 32913297.
  • 22
    Blanpain C, Vanderwinden JM, Cihak J, Wittamer V, Le Poul E, Issafras H, Stangassinger M, Vassart G, Marullo S, Schlondorff D, Parmerrtier M, Mack M. Multiple active states and oligomerization of CCR5 revealed by functional properties of monoclonal antibodies. Mol Biol Cell 2002;13: 723737.
  • 23
    Keith DE, Murray SR, Zaki PA, Chu PC, Lissin DV, Kang L, Evans CJ, Von Zastrow M. Morphine activates opioid receptors without causing their rapid internalization. J Biol Chem 1996;271: 1902119024.
  • 24
    Whistler JL, Chuang HH, Chu P, January LY, Von Zastrow M. Functional dissociation of mu opioid receptor signaling and endocytosis: implications for the biology of opiate tolerance and addiction. Neuron 1999;23: 737746.
  • 25
    Baranski TJ, Herzmark P, Lichtarge O, Gerber BO, Trueheart J, Meng EC, Iiri T, Sheikh SP, Bourne HR. C5a receptor activation. Genetic identification of critical residues in four transmembrane helices. J Biol Chem 1999;274: 1575715765.
  • 26
    Mollison KW, Mandecki W, Zuiderweg ER, Fayer L, Fey TA, Krause RA, Conway RG, Miller L, Edalji RP, Shallcross MA. Identification of receptor-binding residues in the inflammatory complement protein C5a by site-directed mutagenesis. Proc Natl Acad Sci USA 1989;86: 292296.
  • 27
    Raffetseder U, Roper D, Mery L, Gietz C, Klos A, Grotzinger J, Wollmer A, Boulay F, Kohl J, Bautsch W. Site-directed mutagenesis of conserved charged residues in the helical region of the human C5a receptor. Arg2O6 determines high-affinity binding sites of C5a receptor. Eur J Biochem 1996;235: 8290.
  • 28
    Gether U. Uncovering molecular mechanisms involved in activation of G protein-coupled receptors. Endocr Rev 2000;21: 90113.
  • 29
    Samama P, Cotecchia S, Costa T, Lefkowitz RJ. A mutation-induced activated state of the beta 2-adrenergic receptor. Extending the ternary complex model. J Biol Chem 1993;268: 46254636.
  • 30
    Siciliano SJ, Rollins TE, DeMartino J, Konteatis Z, Malkowitz L, Van Riper G, Bondy S, Rosen H, Springer MS. Two-site binding of C5a by its receptor: an alternative binding paradigm for G protein-coupled receptors. Proc Natl Acad Sci USA 1994;91: 12141218.
  • 31
    Whistler JL, Von Zastrow M. Morphine-activated opioid receptors elude desensitization by beta-arrestin. Proc Natl Acad Sci USA 1998;95: 99149919.
  • 32
    Luttrell LM, Ferguson SS, Daaka Y, Miller WE, Maudsley S, Della Rocca GJ, Lin F, Kawakatsu H, Owada K, Luttrell DK, Caron MG, Lefkowitz RJ. Beta-arrestin-dependent formation of beta2 adrenergic receptor-Src protein kinase complexes. Science 1999;283: 655661.
  • 33
    McDonald PH, Chow CW, Miller WE, Laporte SA, Field ME, Lin FT, Davis RJ, Lefkowitz RJ. Beta-arrestin 2. a receptor-regulated MAPK scaffold for the activation of JNK3. Science 2000;290: 15741577.
  • 34
    Goodman OB Jr, Krupnick JG, Santini F, Gurevich VV, Penn RB, Gagnon AW, Keen JH, Benovic JL. Beta-arrestin acts as a clathrin adaptor in endocytosis of the beta2- adrenergic receptor. Nature 1996;383: 447450.
  • 35
    Barak LS, Ferguson SS, Zhang J, Caron MG. A beta-arrestin/green fluorescent protein biosensor for detecting G protein-coupled receptor activation. J Biol Chem 1997;272: 2749727500.
  • 36
    Oakley RH, Laporte SA, Holt JA, Barak LS, Caron MG. Association of beta-arrestin with G protein-coupled receptors during clathrin-mediated endocytosis dictates the profile of receptor resensitization. J Biol Chem 1999;274: 3224832257.
  • 37
    Zhang J, Barak LS, Anborgh PH, Laporte SA, Caron MG, Ferguson SS. Cellular trafficking of G protein-coupled receptor/beta-arrestin endocytic complexes. J Biol Chem 1999;274: 1099911006.
  • 38
    Whistler JL, Tsao P, Von Zastrow M. A phosphorylation-regulated brake mechanism controls the initial endocytosis of opioid receptors but is not required for post-endocytic sorting to lysosomes. J Biol Chem 2001;276: 3433134338.
  • 39
    Prehoda KE, Scott JA, Mullins RD, Lim WA. Integration of multiple signals through cooperative regulation of the N-WASP-Arp2/3 complex. Science 2000;290: 801806.
  • 40
    Mayer ML, Westbrook GL, Guthrie PB. Voltage-dependent block by Mg2+ of NMDA responses in spinal cord neurones. Nature 1984;309: 261263.
  • 41
    Nowak L, Bregestovski P, Ascher P, Herbet A, Prochiantz A. Magnesium gates glutamate-activated channels in mouse central neurones. Nature 1984;307: 462465.
  • 42
    Shiau AK, Barstad D, Loria PM, Cheng L, Kushner PJ, Agard DA, Greene GL. The structural basis of estrogen receptor/coactivator recognition and the antagonism of this interaction by tamoxifen. Cell 1998;95: 927937.
  • 43
    Burger HG. Selective oestrogen receptor modulators. Horm Res 2000;53: 2529.
  • 44
    Lefkowitz RJ. G protein-coupled receptors. III. New roles for receptor kinases and beta-arrestins in receptor signaling and desensitization. J Biol Chem 1998;273: 1867718680.
  • 45
    Murphy PM. The molecular biology of leukocyte chemoattractant receptors. Annu Rev Immunol 1994;12: 593633.
  • 46
    Corvilain B, Van Sande J, Dumont JE, Vassart G. Somatic and germline mutations of the TSH receptor and thyroid diseases. Clin Endocrinol (Oxf) 2001;55: 143158.
  • 47
    Kosugi S, Van Dop C, Geffner ME, Rabl W, Carel JC, Chaussain JL, Mori T, Merendino JJ Jr, Shenker A. Characterization of heterogeneous mutations causing constitutive activation of the luteinizing hormone receptor in familial male precocious puberty. Hum Mol Genet 1995;4: 183188.
  • 48
    Barak LS, Oakley RH, Laporte SA, Caron MG. Constitutive arrestin-mediated desensitization of a human vasopressin receptor mutant associated with nephrogenic diabetes insipidus. Proc Natl Acad Sci USA 2001;98: 9398.
  • 49
    Garcia PD, Onrust R, Bell SM, Sakmar TP, Bourne HR. Transducin-alpha C-terminal mutations prevent activation by rhodopsin: a new assay using recombinant proteins expressed in cultured cells. EMBO J 1995;14: 44604469.
  • 50
    Palczewski K, Kumasaka T, Hori T, Behnke CA, Motoshima H, Fox BA, Le Trong I, Teller DC, Okada T, Stenkamp RE, Yamamoto M, Miyano M. Crystal structure of rhodopsin: a G protein-coupled receptor. Science 2000;289: 739745.
  • 51
    Bower MJ, Cohen FE, Dunbrack RL Jr. Prediction of protein side-chain rotamers from a backbone-dependent rotamer library: a new homology modeling tool. J Mol Biol 1997;267: 12681282.
  • 52
    Guex N, Peitsch MC. SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling. Electrophoresis 1997;18: 27142723.