Use of Concatemers of Ligand Gated Ion Channel Subunits to Study Mechanisms of Steroid Potentiation

Introduction

Most eukaryotic membrane channels are composed of several subunits assembled to make the functional channel, with the notable exceptions of the voltage-gated sodium and calcium channels that are translated as tandem repeats of four modules. The modular basis of the channels increases the diversity of possible functional properties, by allowing mixing and matching of various numbers and types of subunits. However, it also increases the difficulty in determining the molecular mechanisms for drug actions in two ways. First, it becomes necessary to control the number and position of particular subunits in the assembled channel. Otherwise, the concern is always present that a mixture of different subunit arrangements might result in a mixture of pharmacological effects. Second, if multiple copies of a subunit involved in a drug effect are present, it is difficult to distinguish the role(s) of the recognition or transduction regions in the positionally distinct copies. For many anesthetics there is a third difficulty. Many anesthetics act to potentiate responses to transmitter. This mechanism requires that the transmitter and the anesthetic interact with each other through their separate interactions with the receptor. Does this mean that both must bind to the same subunit, or is the interaction mediated by a more global effect of the potentiator on the receptor?

How is it possible to create a receptor which has defined subunits in defined places, given the problem that independent subunits assemble to create the final product? An experimental approach to controlling the number and position of subunits in a channel is to generate larger constructs that contain two or more subunits concatenated together. In an concatemer the DNA coding for two or more separate subunits is covalently linked to result in the translation of a single, larger protein comprising the subunits in a defined linear order (see fig. 1 for schematic views of single subunits, the pentameric receptor, and concatemers). This approach was first used to generate dimers of voltage-gated potassium channel subunits¹. Relatively soon after that it was extended to concatemers of γ-aminobutyric acid type A (GABA-A) receptor subunits², and now has been used in studies of other members of the voltage-gated channel family^3–5, the nicotinic (or ligand gated) ion channel family^6–9 and some other channel types^10,11.

We have been examining the ability of drugs to potentiate the agonist-elicited responses of receptors, with emphasis on the actions of neuroactive steroids. We will discuss the use of concatemers of subunits from two members of the ligand-gated ion channel (LGIC) family, a GABA-A receptor and a neuronal nicotinic receptor. Concatemers have allowed us to selectively manipulate steroid- or transmitter-binding sites and thereby determine the role(s) of particular sites and possible interactions among sites in a receptor.

Research overview

A LGIC gene family receptor is composed of 5 subunits, arranged in a pseudosymmetric rosette around the centrally located ion channel (fig. 1). There is an extensive N-terminal extracellular domain, which contains the binding sites for transmitter and several drugs. There are then 3 closely spaced transmembrane domains, followed by a variable length cytoplasmic loop. The subunit then finishes with a fourth transmembrane domain and a short extracellular C-terminal sequence (fig. 1). The major portion of the ion channel is formed by the second transmembrane domains from each of the five subunits. Channel gating is initiated by binding of transmitter to sites in the N-terminal extracellular domain and communicated to the gating region located in the transmembrane domains. Each of the subunits seems to be strongly coupled to the others in the receptor, in the sense that the process of channel opening appears to occur as a single, global step involving all subunits.

The agonist-binding site is located at an interface between two subunits, with each participating subunit contributing three loops. The assembled receptors we will discuss contain two agonist-binding sites per receptor. The conventional nomenclature is that the positive (primary or +) subunit contributes the A, B, and C Loops, while the negative (complementary or −) subunit contributes the D, E and F Loops (fig. 1). To confuse the nomenclature a bit, for neuronal nicotinic receptors the primary side is contributed by the α subunit and the complementary side by the β subunit. For the GABA-A receptor the primary side is contributed by the β subunit and the complementary by the α. This discrepancy in nomenclature arose because subunits were initially named based on electrophoretic mobility.

We will discuss the actions of steroids to potentiate the responses of a prototypical GABA-A receptor composed of α1, β2 and γ2L subunits, and a neuronal nicotinic receptor composed of α4 and β2 subunits. These particular receptors are, respectively, the most common form of GABA-A receptor and the most common heteromultimeric neuronal nicotinic receptor in the brain. (We note that some steroids and analogues can inhibit GABA-A and/or nicotinic receptors. We will not consider this action.)

Steroids and analogues are well known anesthetic drugs^12–14, and indeed an anesthetic comprising a mixture of alphaxalone and alphadolone was used clinically (Althesin® Glaxo Laboratories Ltd., Greenford, England). A variety of studies have demonstrated that a number of neuroactive steroids potentiate the responses of GABA-A receptors to low concentrations of the transmitter, γ-aminobutyric acid (GABA)^13,15,16. Further, there is a good correlation between the ability of steroids and analogues to anesthetize amphibian tadpoles and their ability to modulate GABA-A receptors¹⁶, supporting the idea that the anesthesia produced by these agents reflects actions at the GABA-A receptor. The mechanism of action has been shown to result from an increase in the stability of the open-channel state of the receptor, rather than an increase in binding affinity or channel opening rate¹⁷. Steroids are very hydrophobic molecules, and studies have indicated that the interaction between a steroid and a GABA-A receptor occurs in the cell membrane^18,19. A major advance was made when a binding site for steroids was identified in the transmembrane regions of the α1 subunit²⁰, and subsequently in other α subunits²¹. It is possible to completely ablate steroid potentiation by mutations to a specific residue, α1 Q241, indicating the importance of this residue^20,22. There are two specific questions we have addressed using subunit concatemers. The first is concerned with the fact that the GABA-A receptor includes two copies of the α1 subunit, each containing a steroid-binding site: does steroid need to bind to both sites to potentiate and are the sites equivalent? The second is concerned with the fact that potentiation reflects binding of both steroid and agonist: does the same subunit need to participate in both binding interactions, or is potentiation a distributed response of the entire receptor?

The nicotinic α4β2 receptor is potentiated by the endogenous steroid, 17β-estradiol (although most steroids and analogues inhibit this receptor, by a distinct mechanism^23,24). The mechanism for potentiation is not known, although it seems likely that it reflects stabilization of the open state²⁵. However, it is known that 17β-estradiol interacts with a specific, 4 amino acid sequence at the extreme C-terminus of the α4 subunit^24,25. We asked similar questions in studies of this receptor: does the C-terminal domain need to be on a specific subunit, does the subunit need to participate in agonist binding, and does the copy number of domains affect potentiation?

Concatemers

The basic portions of a concatemer are the two or more subunits that are joined together, and the linker(s), or new protein sequence that is introduced to connect the subunits.

How to make a concatemer? Our approach has been to use constructs first made by others with only minor modifications, so these comments reflect the insights gained in their work. It is very helpful that both ends (the amino- and carboxy-termini) of these subunits are on the same side of the membrane (fig. 1), the extracellular side, which avoids problems of introducing (or removing) a membrane spanning region. However, when initially translated each subunit has a “signal sequence” at the N-terminal end. The signal sequence assists in membrane insertion of the subunit during translation in the endoplasmic reticulum, and is subsequently cleaved so it is not present in the mature receptor. In the first concatemers, the signal sequence(s) were retained for all subunits in the concatemer, which resulted in a hybrid linker in which part was new protein and part was the existing signal sequence^2,7,9. Most subsequent work has removed all the internal signal sequences (that is, from all subunits except the first)^6,8,26. One comparison reported that removing internal signal sequences resulted in more normal pharmacological properties in pentameric concatemers²⁶.

The next question is what linker sequence should be used - what secondary structure should it adopt? In general, the goal is to achieve a random coil, with the idea that this would place the least structural constraint on the concatemers as they assemble. The first linkers were simple repeats of glutamine residues². Subsequently, linkers incorporating repeats of other amino acids have been used (e.g., alanineglycine-serine triplets)⁸. The rationale for avoiding long repeats of identical amino acids is that such tracts might result in local depletion of transfer RNA molecules and possible early termination of synthesis.

Finally, how long should the linking sequence be? A linker that is too short reduces expression of functional receptors, as might be expected if a short linker caused structural deformations in the subunit that prevent correct assembly^6,8,27. However, it appears that too long a linker can allow more variability in subunit arrangement in the pentamer^8,27. The best length seems to depend on the particular subunits being linked. Typically the total length is estimated from the end of the fourth transmembrane region to the start of the following subunit mature sequence, and usually is in the range of 20 to 40 amino acid residues for well-behaved constructs.

A number of concatemers of differing length have been produced and used in various laboratories, dimers^2,6–8, trimers^27,28 and full pentamers of subunits^26,27,29,30. The full pentamer allows most complete definition of subunit composition, but shorter concatemers have value in terms of ease of generation and manipulation and can be preferred for some experiments. In our laboratories we have used dimers and trimers of subunits, as we will describe.

Two general types of studies have used concatemers. The first type has the objective of defining the properties of receptors with a defined number and arrangement of subunits. In this case, concatemers are produced which incorporate the subunits in particular orders. This approach has been particularly valuable in providing models for possible endogenous receptors whose stoichiometry is unknown^27,31,32. For example, in the case of neuronal nicotinic receptors it has provided the first evidence that receptors containing the α6 and β3 subunits can be studied in defined receptors²⁷. It has also produced some surprising results indicating that so-called “accessory” subunits (that is, subunits that are not thought to contribute to transmitter-binding interfaces) may assemble in unpredicted ways^31,32. In the case of the GABA-A receptor the γ subunit is thought to assemble with one copy per receptor, and to not contribute to a GABA-binding site. However, both the δ and ε subunits apparently can replace not simply the γ subunit in the receptor, but also an α or a β subunit^31,32.

The second type of experiment has the objective of defining the physiological or pharmacological properties of specific subunits in a receptor (for example, the roles of the two primary subunits in transmitter binding)³³. In this case, mutations are made in specific subunits in the concatemer to probe the similarities or differences in consequences for the overall receptor function. Our work so far has focused on this second area. One previous report has been made using concatemers to study receptor modulation - the interaction between the benzodiazepine-binding site and the GABA-binding sites in GABA-A receptors³⁴. In this study, benzodiazepine binding could potentiate responses elicited by GABA binding to either GABA-binding site.

The end result is that several laboratories have succeeded in expressing functional receptors using subunit concatemers. The most successful expression has been in Xenopus oocytes. In general concatemers express more poorly in other systems, such as HEK293 cells, although it has been possible to study some concatemers of GABA-A subunits in nonoocyte systems^9,29,35. The mechanisms determining the surface expression of membrane channels in different expression systems are not fully understood. Xenopus oocytes seem to be particularly adapted for the efficient expression of many proteins, as might be expected based on the extensive burst of initial development they undergo upon fertilization. In contrast, somatic cells are differentiated, to a greater or lesser extent. Indeed, it is known that coexpression of chaperone proteins can enhance expression of particular receptors – for example the RIC3 protein in the case of the nicotinic α7 receptor^36,37 or 14-3-3 protein for the nicotinic α4β2 receptor^26,38. In other cases, mutation of so called “retention sequences” (that tend to reduce receptor trafficking from the initial site of synthesis to the surface membrane) can enhance surface expression³⁹.

Drawbacks to using concatemers

The first possible drawback to the use of concatemers is that the resulting receptors may not actually include the expected complement of subunits. This was recognized very early on, in studies of voltage-gated potassium channels⁴⁰. One obvious example of this problem is the finding that expression of a dimer or a trimer construct of LGIC subunits by itself can result in the expression of functional channels⁸. This is a problem, as the functional assembled receptor contains five subunits. Examination of the expressed protein, using immunoblotting, indicated that the concatemers are not degraded to separate monomeric subunits, so the question arises as to the location of excluded subunits. For nicotinic α4β2 receptors, it appeared that at least some of the receptors formed when a two-subunit concatemer was expressed were actually dimers of pentamers with a bridge contributed by one of the concatemers⁸. In other cases, it has been proposed that some subunits may be in an undefined conformation that is “looped out” from the assembled receptor⁷ – that is, a functional receptor is formed by five of the subunits assembling while the extra subunit is outside the pentamer. Other examples of unexpected assemblies have been identified by using “reporter” mutations, which result in a defined change in receptor function or pharmacology. The reporter mutation is made in a specific subunit in a concatemer to determine whether that subunit is functionally present in the receptor. In some cases, it appears that a subunit may not be expressed in the assembled receptor⁷. The properties of defined receptors with these proposed abnormal subunit locations have not been extensively studied, so it is not clear how different they are from normal receptors. Instead, the focus of most research has been on defining conditions in which biochemical and functional tests indicate that the assembled receptors have the expected subunit stoichiometry.

The second possible drawback is that placing subunits in a concatemer may alter the physiological or pharmacological properties of the resulting receptor. It has been reported that concatemers of GABA-A receptors often have a shift in the concentraton dependence for activation by GABA, requiring a higher concentration^28,41. For nicotinic α4β2 receptors, making a concatemer linked to the carboxy-terminus of the α4 subunit removed potentiation by 17β-estradiol^8,42. This is because the action of 17β-estradiol requires a free carboxy-terminal. However, in both of these cases other properties of the receptors were normal (see next sections, “Neuronal nicotinic receptors: potentiation by 17β-estradiol” and “GABA-A receptors: potentiation by neurosteroids”).

What these observations require, then, is that any investigator who uses concatemers must perform the necessary controls to demonstrate that the concatemers are structurally intact, have reasonable functional and pharmacological properties, and behave in a consistent fashion.

The generation, properties and uses of concatemers have been reviewed elsewhere, for those interested in pursuing the topic further^43–46. Although there are some clear artifacts that can arise in the use of concatemers, overall they provide the most direct means to answer some types of question, as we will discuss in the rest of this article.

Neuronal nicotinic receptors: Potentiation by 17β-estradiol

The endogenous steroid 17β-estradiol potentiates activation of the nicotinic α4β2 receptor. Previous work has shown that the final 4 amino acid residues of the α4 subunit are required^24,25, and that potentiation is very sensitive to the location of these residues. For example, adding a single residue at the end abolishes potentiation, as does insertion or removal of a single residue lying between the end of the fourth transmembrane region and the 17β-estradiol domain²⁴. This spatial sensitivity suggested that the domain had to be located in a precise location in the assembled receptor, and led to the hypotheses that (1) the domain had to be on the α4 subunit and (2) the α4 subunit that bound steroid had to participate in binding of acetylcholine (contributing the primary side of the binding site).

We used concatemers containing two subunits, one α4 and one β2 subunit, to test these hypotheses⁴². Two concatemers were produced, one with a β2 subunit at the amino-terminal end (β2-α4), and the other with α4 (α4-β2). The concatemers were based on the work of Zhou et al⁸, and we measured voltage-clamped responses from receptors expressed in Xenopus oocytes⁴². Receptors were expressed by injecting a concatemer with a free subunit. We first confirmed that the receptors formed in this way had the properties expected from studies of free subunits (for example, by expressing a concatemer with free α4 subunit the resulting receptor had properties of a receptor containing three copies of the α4 subunit). Furthermore, we demonstrated that we could place a mutated subunit selectively in either an acetylcholine-binding or the structural position (fig. 2), using the two concatemers. The data clearly demonstrated that the functional, pentameric receptors assembled when a dimer was expressed with a free subunit behaved in the fashion expected for a receptor with two copies of the concatemer and one copy of the monomer.

The α4 and β2 subunits have one or two proline residues at the end of the fourth transmembrane region, which allowed a reference point for the transfer of the carboxy-terminal β-estradiol domain.

The results show that neither of our hypotheses is correct. We found that we could move the β-estradiol domain to a single β2 subunit, and eliminate it on all α4 subunits, and have strong potentiation by 17β-estradiol (fig. 2B), so hypothesis (1) is disproven. We also found that the domain could be on an acetylcholine-binding subunit or the structural subunit (either a β2 or an α4 subunit) with essentially equivalent potentiation (fig. 2B), so hypothesis (2) is disproven. We could then construct receptors with various numbers of domains included in the pentamer: 0, 1, 2, 3, or 5. The estimated maximal potentiation increased more than linearly with the numbers of domains (fig. 2C).

What interpretations do we make of these data? The first is that 17β-estradiol does not have to interact with a subunit which also binds acetylcholine. This observation indicates that potentiation involves a process which affects the receptor as a whole. The second is that, although the domain is spatially very constrained, potentiation does not seem to require a particular subunit on either side of the subunit containing the domain. This suggests that the mechanism of potentiation only involves the single subunit which contains the estradiol domain. Finally, the maximal potentiation increases about 1.6-fold with each added copy of the domain. A geometric increase is consistent with a simple idea for the mechanism of potentiation. This idea is that when 17β-estradiol binds to the C-terminal domain on a specific subunit, it results in an alteration of the structure of that subunit which stabilizes the open-channel state of the entire receptor by adding a stabilizing energy. If 17β-estradiol binds to another subunit in the same receptor, then there is the addition of an identical energy to stabilize the open-state. Each additive energy contribution results in a multiplicative change in potentiation, because the total energy change is exponentiated to result in the change in equilibrium constant. That is, the multiplicative increase in potentiation results from independent energetic contributions from each subunit. Because of the concerted nature of gating for the LGIC family channels, these independent contributions from subunits have a global effect on receptor activation.

GABA-A receptors: Potentiation by neurosteroids

A number of steroids potentiate activation of GABA-A receptors^15,16. The work by Hosie et al²⁰ has shown that the amino acid residue at position 241 in the α1 subunit (α1 Q241) forms an essential part of the steroid-binding site. We initially explored the consequences of the fact that there are two copies of the α1 subunit in the assembled receptor⁴¹. Does steroid need to bind to more than one subunit to potentiate the receptor, and do the two steroid-binding sites have different pharmacological properties?

We constructed two concatemers, following the work of Baumann et al²⁸. One contained β2-α1 subunits, and the other γ2L-β2-α1 subunits (fig. 3A). The two concatemers were expressed together, and assembled to form functional receptors. The results indicate that the presence of a single normal site conferred steroid potentiation (fig. 3B). When both sites were mutated, potentiation was lost. We tested the pharmacological properties of the sites by comparing the abilities of a 5α-reduced steroid and a 5β-reduced steroid to potentiate, and found that there was no selective effect on potentiation (fig. 3B). We then extended these observations by expressing receptors containing concatemers in HEK cells and examining single channel currents elicited by GABA and modulated by steroids³⁵. The single channel properties of receptor containing concatemers of wild-type subunits were qualitatively identical to those seen with receptors formed from free subunits. The only quantitative difference was a reduction in the affinity of the resting receptor for GABA, which is also reflected in the finding that the concentration of GABA needed to produce a half-maximal whole-cell response is also increased for receptors containing these concatemers^28,35,41. For the wild-type concatemers, potentiation by the steroid allopregnanolone showed the same kinetic effects that we had seen in receptors composed of free subunits. We also obtained results for the receptors with the α1Q241L mutation in the γ-β-α concatemer; again potentiation by allopregnanolone showed the same kinetic effects as for free subunits with intact steroid-binding sites. (Several of the concatemers expressed at such low levels that it was not possible to obtain adequate numbers of single channel events.) Overall, the studies of single channel currents indicate that the receptors composed of concatemers have normal functional properties and modulation by neurosteroids. The studies of whole cell responses demonstrate that the sites in the two α1 subunits have indistinguishable pharmacological properties and that either can support potentiation by steroids. The results indicate that there is only a small decrease in the maximal potentiation produced by steroids when one site is removed.

We then set out to answer a second question⁴⁷. Potentiation requires binding of both transmitter and potentiator. Is there a“privileged” relationship for particular subunits, for example is potentiation enhanced when steroid binds to a subunit which also binds transmitter, or does potentiation apply equally to activation elicited by binding to either transmitter-binding site? For these studies we used a different pair of concatemers, one containing β2-α1 and the other β2-α1-γ2L (fig. 3C). We selectively ablated the GABA-binding site by the mutation β2Y205S⁴⁸ and the steroid site by α1Q241L. The mutated constructs were expressed in all possible combinations (fig. 3D) to examine interactions between steroid- and transmitter-binding sites, using the neurosteroid allopregnanolone to modulate GABA-elicited responses. The results did not show any significant coupling between the GABA- and steroid-binding sites on a particular subunit (fig. 3D). Potentiation was seen whether the GABA- and steroid-binding occurred in the same α subunit or not. These results indicate that neurosteroid potentiation of GABA-A receptors is mediated by a generalized effect on the entire receptor.

It is interesting to note that it was possible to activate receptors in which both GABA-binding sites were ablated by using pentobarbital⁴⁸. The data⁴⁷ indicated that either steroid site was able to potentiate responses elicited by pentobarbital, as in the case of GABA. Since the binding site for pentobarbital is presently unknown it was not possible to examine interactions between pentobarbital and allopregnanolone.

Overall conclusions

Steroids, either neurosteroids acting on the GABA-A receptor or 17β-estradiol acting on the nicotinic α4β2 receptor, appear to act by affecting the gating properties of the receptor as a whole. Potentiation does not require that the subunit that binds steroid also binds agonist. For the nicotinic receptor, the steroid-binding domain is a remarkably discrete element that can be moved from one subunit to another. This may be analogous to our finding that the two copies of the α1 subunit in the GABA-A receptor are essentially equivalent to each other. One difference between the nicotinic and the GABA-A receptors is that potentiation increases steadily as we increase the number of copies of the binding domain in the nicotinic receptor, but does not change consistently when we reduce the number of intact domains in the GABA-A receptor from two to one.

Steroid potentiation is mediated by interactions in or very near to the membrane-spanning regions of these receptors. It is already known that there are mutations in the second transmembrane region which enhance the stability of the open channel state^49–51. It is interesting that these mutations appear to act irrespective of the subunit in which they are placed - transmitter-binding or structural - and that the energetic contributions to stabilizing the open state appear to add. It is possible that the mechanism of steroid potentiation is to directly stabilize the channel in the open channel conformation by an effect on transmembrane domain interactions either with other channel domains or with the surrounding lipid.

The results confirm that concatemers of subunits can be used to define the number and position of subunits in assembled and functional LGIC family receptors. By using concatemers we have been able to ask and answer questions about the nature of steroid interacting sites on these receptors and how interactions between potentiating drugs and transmitters may occur.

Potentiation of GABA-A receptors by steroids underlies their anesthetic actions. Neurosteroid potentiation is potent and efficacious, particularly for responses resulting from relatively low activation levels (that is, by low concentrations of GABA or by receptors which have a low maximal activation). Potentiation reflects a global, receptor-wide effect, and the presence of multiple homologous binding sites in a single receptor provides an efficient and redundant mechanism for mediating steroid effects.

Future directions

We are quite excited by the possibilities for future work which are opened up by the use of concatemers. In studies of anesthetic actions, there are several general areas we think are ripe for study. All of these will focus on the GABA-A receptor, as the more relevant target for studies of anesthetics.

The first is the study of additional classes of anesthetics. Other anesthetics have identified binding regions in the GABA-A receptor^52–56 which are present on subunits which have two copies in the assembled receptor. Analogous studies to the ones we have performed with anesthetic steroids will help to clarify the properties of individual anesthetic sites and interactions with GABA-binding sites.

Another is to extend studies using voltage-clamp fluorometry⁵⁷. By introducing the reporter fluorophore into specific locations in particular subunits, we hope to be able to examine the propagation of conformational changes induced by anesthetics.

Finally, concatemers provide the opportunity to study receptors of defined subunit stoichiometry and arrangement. Recent experiments^31,32,58 have demonstrated that some subunits, notably the δ and ε subunits, can assemble in different positions in the pentamer and may be able to incorporate with differing numbers of copies per receptor. Receptors containing these subunits often contribute to nonsynaptic, “tonic” GABAergic responses. We plan to extend our studies of anesthetic actions to non-synaptic receptors, but to do so we need to have a defined population of receptors. The use of concatemers provides a path to this goal.