Abstract
Introduction
Two recent models are in use for analysis of allosteric drug action at receptor sites remote from orthosteric binding sites. One is an allosteric twostate mechanical model derived in 2000 by David Hall. The other is an extended operational model developed in 2007 by Arthur Christopoulos’s group. The models are valid in pharmacology, enzymology, transportology as well as several other fields of biology involving allosteric concentration effects.
Results
I show here that Hall’s model for interactions between an orthoster, an alloster, and a receptive unit is the best choice of model both for simulation and analysis of allosteric concentrationresponses at equilibrium or steadystate.
Conclusions
As detailed knowledge of receptors systems becomes available, systems with several pathways and states and/ or more than two binding sites should be analysed by extended forms of the Hall model rather than for instance a Hill type exponentiation of terms as introduced in nonmechanistic (operational) model approaches; yielding semiquantitative estimates of actual system parameters based on Hill’s unlikely simultaneity model for G proteincoupled receptors.
Background
A sizeable decline in development of classical agonists and antagonist for medication [13] has elicited a drughunt to construct and develop allosters in laboratories of academia [48] and industry (e.g., Novasite Pharmaceuticals Inc; Addex Pharmaceuticals), including positive and negative allosters as well as orthoallosters for therapeutic purposes. In doing so, it has become important to simulate and analyse concentrationresponse data for allosters by models that are as close to the systems mechanistic function as possible.
Optimal allosteric models are in great demand, since mechanistic simulations may be combined with structural analysis of alloster binding, receptor multimerization and association of molecules as G proteins, arrestins, and RAMPs into synthesis of QSARs for ligand binding and receptor activation [916].
Data from equilibrium concentrationresponse experiments involving allosteric modulators are presently interpreted by unlike choices of model. Therefore, with such schism in selection of model, especially true for data from cellsystems expressing subtype 7TMRs [17], it seems worth a discussion about which direction analysis of synagics data for allosters should take. For possible outcomes of including allosters consult Figure 1. For definitions of terms related to allostery see Table 1.
Figure 1. Phenotypic behavior of allosters. Panel A. Some concentrationresponse curves with an alloster present demonstrating enhancement and alloinhibition of both a mixed and a competitive type antagonism and with ceiling effects for all three. The red curve represents an orthoster concentrationresponse in the absence of an alloster. Panel B. Concentrationresponse relations with an alloster present, displaying alloagonism as a lifted initial activity with ceiling and allosynergy as a lifted maximal response. Both alloagonism and synergy curves are lifted compared to a concentrationresponse curve with no alloster present as in the green curve. Definitions of phenotypic alloster terms are listed in Table 1.
Two actual allosteric models  ATSM and EXOM. One model is the allosteric twostate model, ATSM, introduced by Hall in 2000, implemented and further discussed by others [5,1725]. Another model we could call the “extended operational model”, EXOM for short [26], is based on combining the original operational model, BLM [27], with the ternarycomplex model, TCM [28], as later further detailed [2931]. EXOM is implemented and presently advocated by several leadmodellers [7,8,3238]. There are other approaches taken to model the behaviour of allosters in the field of 7TMRs [20,33,3942].
ATSM is a mechanistic model. ATSManalysis with extracted numbers for model parameters supposes direct information about mechanical interactions between allosters, receptors and orthosters at a molecular scale. Thus, one might gain a quantitative and dynamic handle on molecular processes per se within receptors. The other model, EXOM, a nonmechanistic model, is a close relative of ATSM and has the same number of independent parameters to be determined. EXOM is used assuming that individual physical parameters of multistep processes as such cannot be extracted, as they are composite. EXOM may give quantified estimates on elicited cooperative binding and efficacy for orthosters and allosters interacting at receptors [26,34]. By selecting similar assumptions for ATSM as for EXOM, ATSM may cover the EXOMscenario and yield estimates of parameters for lumped multisteps rather than single steps, and thus become a blackbox model as the EXOM.
In both ATSM and EXOM, allosters may behave as enhancers with ceiling and as competitive antagonists without ceiling. Furthermore, they are also efficient in simulating alloagonism and allosynergy both with ceiling effects; observed as lifts of concentrationresponse curves by allosters at low and high orthoster concentrations [17,26,37]. However, EXOM lacks ATSM’s advantage of being a mechanistic model and for describing spontaneous activity of receptive units. Additionally, from a theoretical point of view, a parameter in EXOM to describe cooperative activity is amputated, yielding illogic results. For this latter conclusion, see details in the next to last sections of Methods and Results and Discussion.
Here I focus on ATSM and EXOM and compare them for simulation and analysis of experimental data. It is demonstrated that there are no arguments as posited [8,17] for employing EXOM instead of ATSM, quite the other way about. Therefore, my goal is to convince future modellers to use ATSM and possible extended forms for analysis and simulation of allosteric concentrationresponse relations rather than EXOM.
Methods
One basic model  cTSM
In simulation of synagics for orthosters and allosters, the basis of most models is often two simple reaction schemes; the cyclictwostate model, cTSM, and the ternarycomplex model, TCM. Since this paper is about modelling as opposed to general statements about ligandreceptor interactions it is paramount with precise definitions including aspects of cTSM and TCM. This has been discussed before [22] and may seem superfluous. However, in order to validate and compare newly derived ATSM and EXOM in a coherent fashion, concepts related to cTSM and TCM must be brought together and systematized. cTSM is dealt with first.
The gist of the cTSM, Figure 2A, is its explicit description of a conformational switch between an inactive and active state of a nonbound receptor. It specifically includes spontaneous activity in form of nonliganded receptor R*. The behaviour of cTSM has been scrutinized [43,44]. cTSM has two interesting parameters. L describes the distribution between unliganded inactive and active receptor states, R ⇌ R*, such that L = R*/R, Figure 2A. Deriving cTSM’s distribution equation for activity, the free nonactive receptor state R is equated with “1”. Thus, the unliganded, active receptor state R* is equal to L. The second parameter, a, is a concomitant constant for activation of receptor forms bound with ligand S, RS ⇌ R*S. This step has a·L as its efficacy constant. By assuming multisteps, a·L is identical to Stephenson's efficacy constant [45] and Black & Leff’s transducer ratio τ[27]. A_{s} is the equilibrium affinity constant for S binding to nonactive forms of R, Figure 2A. Therefore, a is also a concomitant constant for binding of S to already activated receptors. The affinity constant for S+R* ⇌ R*S is thus a·A_{s}.
Figure 2. Two simple reaction schemes. Panel A. The cyclic twostate model, cTSM, with selection and induction arrows indicating two separate but simultaneous pathways from an inactive and nonliganded receptor conformation R to an active and agonist S liganded receptor conformation R*S. A_{s} is an equilibrium association constant for S, L is a conformational efficacy constant for nonbound receptors, and parameter a is an efficacy constant for ligand bound receptor conformations from RS to R*S. Panel B. The ternarycomplex model, TCM, in which symbol M represents the term and concentration for an additional alloster ligand. A_{m} is an equilibrium association constant for M, and parameter c is a cooperativity coefficient for twoligand binding.
Arguments still appear on how to understand activation of protein molecules when ligands are applied  is it by induction after ligands bind or is it rather by ligand selection and stabilization of already activated molecules? Jacques Monod early on favoured a selection process [46] and this understanding crystallized in the famous MWCmodel [47]. The MWC explicitly introduces an unliganded switch R⇌R* as the “allosteric transition” [48]. Contrary, Koshland argued for induction after binding [49]. “Selection” follows one leg of cTSM while “induction” follows another [50], Figure 2A. They are two views on a single process [18] chapter 5. Below, when either “induction” or “selection” is used on activation of receptive units as ligands bind, it covers both pathways in cTSM.
Another basic model  TCM
The TCM, Figure 2B, looks fairly simple, but possesses surprising allosteric regimes. Depending on which of the liganded complexes are included for activity, TCM can simulate enhancement with ceiling and competitive (“surmountable”) inhibition, besides alloagonism without ceiling and “mixed competitive inhibition”. TCM with tacit active conformations has no allosynergy or spontaneous activity. Ten submodels derived from TCM are characterized in Table 2. Three of these submodels are further described in the Results section and some simulations by these three models are shown in a figure in the Results section.
Figure 3. Reaction schemes of the allosteric twostate model, ATSM, and the extended operational model, EXOM. Panel A. The ATSM. Panel B. The EXOM. The models are presented with their basic simpler reactions schemes as the cTSM and TCM from Figure 2. The cubic ATSM has eight receptor conformations while the EXOM only has seven of those, as the spontaneous active represented by receptor conformation R* is excluded. The two models have the same total number of parameters, seven in all. Besides parameters defined in Figure 2, ATSM and EXOM have parameter b, an efficacy constant when the alloster Mbound receptor is activated, and parameter d a cooperativity efficacy constant involving two ligands. The constants L, A_{s}, A_{m} , a and c are given as in Figure 2, and EXOM has a slope factor n, not shown.
Table 2. Phenotypic concentrationresponses for allosters in 10 submodels from TCM
Operational models
To understand the present use of “stimulus”, “efficacy” and “intrinsic efficacy” in operational models as EXOM, it is necessary to go back to their definitions [45,51,52]. Stephenson’s stimulus concept seems obsolete today by accepting twostep receptor schemes with straightforward derived distribution equations [18] chapter 2; [50] and when needed, apt assumptions of more than two steps. Twostep schemes yield equations identical to initially derived operational models based on the stimulusresponse idea [27,51,53]. Concepts as “stimulus”, “transducer ratio” and “fitting parameter” are of course justified in selecting operational model approaches rather than mechanistic ones. Spontaneous activity often seen in studies with 7TMRs is not included in the realm of operational models, although recently serious attempts have appeared [54,55].
Meanwhile, users of operational models should recognize that their assumptions for derivation put a veil over underlying physical systems and that any involved “operational” assumption may just as well be applied to the ATSM. For instance, as mentioned, a·L can be conceived as equal to transducer ratio τ.
Distribution equation for ATSM and EXOM
Reaction schemes of ATSM and EXOM are depicted in Figure 3A and 3B. The intention with EXOM was to derive a stimulusequation for activating receptors, including allosteractivated units, while explicitly excluding nonliganded active conformations [26]. Thus, three bound species RS, MR, and MRS in EXOM can switch to active forms R*S, MR*, and MR*S. But, in order to exclude constitutive activity, nonliganded R is not allowed a switch to active R*, Figure 3B. Thus, EXOM is a pure “induction” reaction scheme in Koshlandsense, as free forms of receptor R must be bound before activation. The three bound and active forms of the receptor are equated as “stimulus” and transformed through a hyperbolic expression for activity, as for the BLM. The result is a distribution equation with three active conformations to a total of seven conformation, as even a possible inactive R*conformation is considered nonexistent [26].
Figure 4. Simulations from four submodels of the ternarycomplex model, TCM. For submodel definitions see Table 2. Parameters A_{s} and A_{m}, equilibrium association constants for ligands S and M, are kept at unity. Parameter c, the cooperativity constant for binding, is varied by a factor 10^{3} in three steps for each submodel as indicated in the panels. Red curves indicate orthoster concentrationresponse curves in the absence of an alloster. In all panels the alloster M concentration is varied in four steps: in panels AI by a factor 10^{2} from 1x10^{2} to 1x10^{4}; in panels GK by a factor 10 from 1x10^{2} to 1x10^{1} and in panel L by a factor 10^{2} from 1x10^{3} to 1x10^{3}. Green curves with circles show the actual EC_{50} and the black circle represents the position of a limiting EC_{50} for M → ∞.
To simplify a comparison of EXOM with ATSM, distribution equations for both are expressed parallel to earlier expressions for ATSM [18] chapter 7.
This yields for activity in EXOM:
and for activity in ATSM:
Deviations between the two models are marked by bracketed and bolded symbols. Definitions of symbols listed below are followed by symbols in parenthesis from Leach [26] and Hall [22]: E = actual response; E_{m} = maximal activity; S = orthoster (A; A); M = alloster (B; B); A_{s} = equilibrium association constant for ligand S (1/K_{A} , K); A_{m} = equilibrium association constant for ligand M (1/K_{B}, M ); a = efficacy constant for S (τ_{A}; α); b = efficacy constant for M (τ_{B}; β); c = binding cooperativity constant (α; γ); and d = activation cooperativity constant (β; δ). Parameter β for EXOM is only defined for cooperativity of an alloster on orthoster activation, but not reciprocally as in ATSM. Further, unlike ATSM, EXOM has a Hill type exponentiation parameter, n, for terms of summed activity and inactivity. The benefits of including such a Hill exponentiation may be questioned as discussed earlier [18] chapter 10. Indeed, Hilltype exponentiation may also be applied to ATSM. However, as ATSM is a mechanistic approach, it seems more logical to derive equations based on formulation for an extended ATSM with more than two binding sites [18,25].
In absence of an orthoster the initial efficacy, IntEff, for ATSM is given by: L/[L+(1+ A_{m}·M)/ (1+b·A_{m}·M)], and for EXOM, assuming n = 1, by: 1/[1+(1+A_{m}·M)/(1+b·A_{m}·M)].
For high values of the orthoster, S⇒∞, maximum activity, MaxEff, as a function of alloster concentration for ATSM is given by: L/[L + (1 + c · A_{m} · M)/(a · (1 + b · c · d · A_{m} · M))], and for EXOM, assuming n = 1, by: 1/[1 + (1 + c · A_{m} · M)/(a · (1 + c · d · A_{m} · M))]. Differences between ATSM and EXOM expressions are indicated with bolded types.
Bestfit analyses to experimental data for ATSM and EXOM
The analyses were performed in the following manner. Selected allosteric effects were obtained from datafigures in the literature, datafigure 1 ([38], Figure 2B), datafigure 2 ([37], Figure 2B), and datafigure 3 ([56], Figure 3). Model parameters a and A_{s} were first evaluated by fitting the distribution equations for ATSM and EXOM to response data at zero alloster concentration. The obtained values for a and A_{s} were then inserted into the distribution functions for the two models and used for an ensuing fitting of the remaining parameters listed in the last Table, parameters b, c, d, and A_{m}. By varying the initial values for each parameter in three steps, at least 12 fits were performed on each curve for every alloster concentration in all three datafigures. Only fitted parameter values with convergence to a tolerance of 10^{10} in SigmaPlot software were accepted.
Thus, concentrationresponse curves at three different alloster concentrations yields three bestfit values for each of the four parameters. Obtained results for the single parameter in the last Table represent a ratio between the two bestfit values with the largest mutual difference of the three determinations for each parameter at different alloster concentrations. A global fit to data sets for all four parameters [57] was not possible.
A fourth data set, datafigure 4 ([36], Figure 1C), was also analysed but neither ATSM nor EXOM fitted well to these data with a 44% spontaneous activity and a 56% alloster/ orthoster response. The failure of fitting was mostly due to a lack in obtaining a reasonable determination of maximal response for several of the concentrationresponse curves.
Results and discussion
TCM  three and ten variants
Three functional variants of TCM are briefly described below and examples of their simulations shown in Figure 4, while characteristics of ten different forms derived from TCM are listed in annotated Table 2.
In a first form, complex RS tacitly moves to R*S as the sole source of activity. Simulation of this alloscheme can resemble classical noncompetitive antagonism for orthosters in functional assays, where only the maximal effect attenuates as the concentration of an alloster increases while the dissociation constant for the agonist stays constant. This happens for activity when constant c is unity. An example is shown in Figure 4B. Note, that in TCM occupancy, alloster effects can never be noncompetitivelike, i.e., with reduced activity and fixed EC_{50}.
In a second form, Sliganded conformations, RS and MRS, move tacitly to R*S and MR*S as source of activity. This reaction scheme gives us models of activity and occupancy that behave in an identical manner as their distribution equations are identical. This reaction scheme includes enhancement for constant c > 1 and with ceiling when A_{m}·M > 1 and competitive inhibition when c < 1, but with a ceiling effects for both binding and activation by an alloster when c·A_{m}·M > 10, Figure 4D and 4F. This model is identical to the uncompetitive reaction scheme.
In a third form, all liganded conformations, i.e., RS, MR, and MRS, are sources of activity, Figure 2B. In EXOM, this is the basic TCM. TCM subtype 3 may simulate alloagonism for activity, but without ceiling effects as indicated by black circles for limiting EC_{50} values as M → ∞, Figure 4GI.
Since the term “competitive inhibition”, according to an informative review [48], meant inhibition through an overlap or steric hindrance at binding sites [58], the term “allosteric inhibition” was used from the start of the 1960s merely to indicate negative feedback different from competitive inhibition. Nothing more. TCM with its two remote binding sites has no mutual exclusion by steric hindrance or by overlap. Meanwhile, TCM may still simulate “competitive inhibition”, either by its uncompetitive form as shown in Figure 3F, or by mutual exclusion of triple complex MRS through remote or intermolecular conformational changes, not shown. Thus, TCM has allosteric inhibition in the MWCsense. “Competitive inhibition” by mutual exclusion in TCM requires that the cooperative binding constant c goes to insignificantly small values, thus preventing detectable levels of MRS and of its tacitly active form, MR*S. Such allosteric mutual exclusion, as one type II competitive inhibition ([18], chapter 2) has been cartooned ([58], Figure III1, panel 5). Thus, as “allosteric” solely refer to ligand binding at remote, nonoverlapping binding sites and without steric hindrance, “allosteric” becomes a pleonasm in “allosteric ternary complex model”, ATCM, as TCM is defined by having two, nonoverlapping binding sites without steric hindrance. As both acronyms cover the exact same model, it remains a matter of taste using either ATCM or TCM. Contrary, the signifier “allosteric” in “allosteric transition” [48] becomes indicative for twostate models as MWC and ATSM, involving cTSM.
Figure 5. Simulations of concentrationresponse relations for ATSM and EXOM. The parameters A_{s} and A_{m} are both kept at unity, while parameter L is 10^{2} for all ATSM simulations in order to keep spontaneous activity insignificant and n for all EXOM simulations is = 1. Parameter c, the binding cooperativity constant, is varied in three steps by multiplying with a factor 10^{3} from 10^{3} to 10^{3} as indicated in the panels. Parameter a is 5000 in all ATSM panels except for panels MP where it is 500. For EXOM, parameter a is 50 in all panels except for panels QS where it is 5. For ATSM, parameter b is 1 in panels AC, and 50 in the rest of panels GV. For EXOM, parameter b is 0.01 in panels DF, and 0.5 in the rest of panels JZ. Parameter d is 1 in all panels except in panels TZ where it is 3x10^{3}. All red curves have no alloster present, i.e., concentration of M = 0. M is varied in four steps. In panels AF by a factor 100 from 2x10^{4} to 2x10^{2}; in panels GS by a factor 10 from 2x10^{3} to 2x10^{0}; and in panels TZ by a factor 10 from 2x10^{1} to 2x10^{2}. Green curves with circles show the actual EC_{50} and the black circle represents the position of a limiting EC_{50} for M → ∞. The black circle falls outside the orthoster concentration range, 10^{6} to 10^{2}, in panels S and Z with limiting EC_{50} values of 250 and 1304.
Comparison of simulations from ATSM and EXOM
A comparison is made between ATSM and EXOM simulations of concentrationresponses of activity with orthoster concentration as independent variable and with varying alloster concentration M. Thus, the following are principal statements about parameter influences on initial and maximal efficacies, on ceiling effects for enhancement, competitive and mixed inhibition, on alloagonism and synergy, as well as on apparent dissociation constant EC_{50}. To simplify the comparison, EXOM slope factor n is assumed unity. The results reveal a few crucial differences between the two models even based on homologous parameters as A_{s}, A_{m}, a, c, and d.
As indicated above, IntEff for EXOM is dependent on parameter b, while for ATSM it is dependent on both b and L. For ATSM, MaxEff is dependent on L·a, whilst EXOMMaxEff is only dependent on a. Thus, when comparing ATSM and EXOM, choice of values for a and b in EXOM should match with values for L·a and L·b in ATSM. Accordingly, in selection of parameter values for compared simulations with L for ATSM chosen as 0.01 in order to suppress spontaneous activity, values for a and b in EXOM are chosen 100 fold higher in ATSM, Figure 5.
IntEffs for both ATSM and EXOM are always completely independent of A_{s}, a, c, and d. ATSMIntEff is dependent on L and b·A_{m}·M. For more details see annotated Table 3. EXOMIntEff only depends on b·A_{m}·M. Alloagonism is a lift in the IntEff when supplying an alloster even before an orthoster is added. Various forms of alloagonism are shown in Figure 5GZ and with ceiling effects indicated by black circles for the limiting EC_{50} values as M → ∞. Alloagonism is often seen in studies with small molecule allosters [59]. Alloagonism takes effect in both models when both b and b·A_{m}·M are larger than unity. Furthermore, ATSM may simulate spontaneous activity before any ligand is added. Simulation of detectable spontaneous activity starts at values of L above 10^{2}. This possibility is excluded from the EXOM theory.
Table 3. Conditions for alloster effects on initial efficacy and maximal efficacy in ATSM
MaxEff in ATSM is dependent on L·a and b·c·d·A_{m}·M, Table 3, while MaxEff in EXOM is dependent on a and c·d·A_{m}·M. In comparison, EXOMMaxEff demonstrates complete independence of b, which is somewhat inconsistent. The independence is due to the definition of parameter d (β) in EXOM, where an alloster only affects the efficacy of an orthoster with no reciprocity. Thus, synergy and mixed inhibition are different between ATSM and EXOM, since the MaxEffATSM has both parameter b and d involved while EXOM only depends on d.
As already indicated, more details on parameter influences on IntEff, enhancement, alloagonism, allosynergy, MaxEff, and mixed inhibition are given in comments to Table 3.
Ceiling effects of enhancement and alloagonism by positive allosteric modulators (PAMs) are hallmarks and often detected in experiment [17,3537]. These ceiling effects appear for A_{m}·M > 1, panels A, D, G, J, M, Q, T, and X in Figure 5. Ceiling effects for competitive inhibition are determined by cooperative binding constant c < 1 and appears for c·A_{m}·M > 10, and best seen for b·d = 1, panels C, F, I, L, P, S, V, and Z in Figure 5.
The ATSM was rejected as model for allocompetitive inhibition by gallamine at muscarinic subtype M2 receptors [20]. Meanwhile, both ATSM and EXOM can nicely simulate competitive inhibition with values of c low enough to keep the parameter products b·c·d·A_{m}·M for ATSM and c·d·A_{m}·M for EXOM less than 10, exemplified in Figure 5C and F.
Allosynergy, seen in the presence of allosters as a lift in MaxEff above MaxEff for othosters alone, is now commonly described for agonisticPAMs as well [5,8,25,36]. In ATSM, these characteristics of PAMs with MaxEff above maximal response for endogenous ligands alone may be simulated with values of b and d when their product is > 1, Figure 5MN, while EXOM can simulate allosynergy for d > 1, not shown. Mixed inhibition, appearing as values of MaxEff lower than MaxEff with orthosters alone in the presence of NAMs, including pure noncompetitive inhibition, may be simulated for b·d < 1 in ATSM, Figure 5U, and for d < 1 in EXOM, Figure 5Y. Published examples of negative allosteric effects are now increasing as more interest is invested in development of NAMs [12,32,60].
In both allosynergy and alloinhibition, parameter c, as its value is lowered, will narrow the gap between MaxEff in the presence and absence of an alloster; compare panels MP and panels TZ in Figure 5.
The lack of effect of parameter b on MaxEff in EXOM clearly weakens the theory, even though additional details have been presented on the behaviour of EXOM [34]. A variant of EXOM has been developed with lumped parameters thus avoiding the problem of a missing effect of parameter b in MaxEff [24].
Comparison of bestfit analyses to experimental data for ATSM and EXOM
Results from analysis of experimental data with ATSM and EXOM are listed in Table 4. Ideally parameters in a theory should manage to stay constant when the theory is fitted to different data sets of the same experimental concentrationresponse system; for instance at increasing alloster concentrations. Therefore, the more the ratios in Table 4 for each single parameter deviate from unity in the present analysis, the worse is its model’s credibility.
Table 4. Parameter ratios from bestfits with ATSM and EXOM on three data sets
Both ATSM and EXOM have problems with a convincing determination of parameters fitted to data in datafigure 2. However, ATSM still seems to give the best result based on an overall evaluation of ratios for all four parameters from the three data sets of datafigure 2, Table 4.
Although exponentiation in form of a Hill coefficient may also be invoked for both models, such exponentiation was omitted in the present analysis. Also, an interpretation and detailed discussion of the actually obtained parameter values are beyond the scope of this paper.
Thus, based on the ratios in Table 4, it may be concluded that ATSM seems to be better than EXOM at evaluating possible parameter values with a requirement of consistency when determined at 3 different alloster concentrations, since in general most of the ratios are closer to unity when employing the ATSM.
Conclusion
In a beautiful review, nonmechanistic EXOM against mechanistic ATSM is debated and further contrasted with an empirical general description of synagic behaviour of allosters in different experimental setups [17]. When system information is limited, analyses of allosteric behaviour by operational, empirical and mathematical approaches as Hill’s exponentiation are still valid. Meanwhile, analysing systems of allosteric synagics as discussed here, the best description of allosteric effects is by Hall’s millennium milestone mechanical model [22] due to shortcomings of EXOM. Limitations of mechanistic models as the ATSM are given with its assumptions, which usually both exclude more than two binding sites and multisteps or parallel pathways. The ATSM may still replace the EXOM as a phenomenological model by applying assumptions similar to those for EXOM. For the future, allosteric models should be developed based on ATSM and implicating multibinding and diverse pathways of receptor activation when needed. Thus, instead of switching to nonmechanistic approaches as EXOM or reduce requirements for the basic TCM to analyse such systems [20,26], phenomenological or extended forms of the ATSM should be preferred (e.g., [25]).
Abbreviations
ATSM: Allosteric twostate model;EXOM: Extended operational model;cTSM: Cyclic twostate model;BLM: The Black & Leff operational model;7TMRs /GPCRs: 7 transmembrane helix G proteincoupled receptors;TCM: Ternarycomplex model;ATCM: Allosteric ternarycomplex model;EC50: Apparent dissociation constant at 50% activity;IntEff: Initial efficacy;MaxEff: Maximal efficacy;PAMs and NAMs: Positive and negative allosteric modulators;QSAR: Quantitative structureactivityrelationship
Competing interest
The author declares no conflicts of interest.
Authors’ contribution
NB developed and wrote the MS.
Acknowledgments
I thank Dr. David A. Hall for helpful discussion on twostate and operational model approaches and for significant comments on previous versions of the MS.
References

Nature 2011, 474:433434. PubMed Abstract  Publisher Full Text

Macilwain C: Pharmaceutical industry must take its medicine.
Nature 2011, 470:141. PubMed Abstract  Publisher Full Text

Scannell JW, Blanckley A, Boldon H, Warrington B: Diagnosing the decline in pharmaceutical R&D efficiency.
Nat Rev Drug Discov 2012, 11:191200. PubMed Abstract  Publisher Full Text

Elsinghorst PW, Härtig W, Gündisch D, Mohr K, Tränkle C, Gütschow M: A hydrazide linker strategy for heterobivalent compounds as ortho and allosteric ligands of acetylcholinebinding proteins.
Curr Top Med Chem 2011, 11:27312748. PubMed Abstract  Publisher Full Text

Gao ZG, Verzijl D, Zweemer A, Ye K, Göblyös A, Ijzerman AP: Functionally biased modulation of A(3) adenosine receptor agonist efficacy and potency by imidazoquinolinamine allosteric enhancers.
Biochem Pharmacol 2011, 82:658668. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Jensen PC, Thiele S, Steen A, Elder A, Kolbeck R, Ghosh S: Reversed binding of a small molecule ligand in homologous chemokine receptors  differential role of extracellular loop 2.
Br J Pharmacol 2012, 166:258275. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Melancon BJ, Hopkins CR, Wood MR, Emmitte KA, Niswender CM, Christopoulos A: Allosteric modulation of seven transmembrane spanning receptors: theory, practice, and opportunities for central nervous system drug discovery.
J Med Chem 2012, 55:14451464. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Valant C, Felder CC, Sexton PM, Christopoulos A: Probe dependence in the allosteric modulation of a G proteincoupled receptor: Implications for detection and validation of allosteric ligand effects.
Mol Pharmacol 2012, 81:4152. PubMed Abstract  Publisher Full Text

Audet M, Lagacé M, Silversides DW, Bouvier M: Proteinprotein interactions monitored in cells from transgenic mice using bioluminescence resonance energy transfer.
FASEB J 2010, 24:28292838. PubMed Abstract  Publisher Full Text

Chung KY, Rasmussen SG, Liu T, Li S, Devree BT, Chae PS: Conformational changes in the G protein Gs induced by the ß2 adrenergic receptor.
Nature 2011, 477:611615. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

CompsAgrar L, Kniazeff J, NørskovLauritsen L, Maurel D, Gassmann M, Gregor N: The oligomeric state sets GABA(B) receptor signalling efficacy.
EMBO J 2011, 30:23362349. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Henderson BJ, Orac CM, Maciagiewicz I, Bergmeier SC, McKay DB: 3DQSAR and 3DQSSR models of negative allosteric modulators facilitate the design of a novel selective antagonist of human a4ß2 neuronal nicotinic acetylcholine receptors.
Bioorg Med Chem Lett 2012, 22:17971813. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Nygaard R, ValentinHansen L, Mokrosinski J, Frimurer TM, Schwartz TW: Conserved watermediated hydrogen bond network between TMI, II, VI, and VII in 7TM receptor activation.
J Biol Chem 2010, 285:1962519636. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Peeters MC, Wisse LE, Dinaj A, Vroling B, Vriend G, Ijzerman AP: The role of the second and third extracellular loops of the adenosine A1 receptor in activation and allosteric modulation.
Biochem Pharmacol 2012, 84:7687. PubMed Abstract  Publisher Full Text

Schelshorn DW, Joly F, Mutel S, Hampe C, Breton B, Mutel V: Lateral Allosterism in the Glucagon Receptor Family: GLP1 Induces GPCR Heteromer Formation.
Mol Pharmacol 2012, 81:309318. PubMed Abstract  Publisher Full Text

Van Eps N, Preininger AM, Alexander N, Kaya AI, Meier S, Meiler J: Interaction of a G protein with an activated receptor opens the interdomain interface in the alpha subunit.
Proc Natl Acad Sci USA 2011, 108:94209424. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Keov P, Sexton PM, Christopoulos A: Allosteric modulation of G proteincoupled receptors: a pharmacological perspective.
Neuropharmacology 2011, 60:2435. PubMed Abstract  Publisher Full Text

Bindslev N: DrugAcceptor Interactions. Modeling Theoretical Tools to Test and Evaluate Experimental Equilibrium Effects. 1st edition. Stockholm: CoAction Publishing; 2008.

De Amici M, Dallanoce C, Holzgrabe U, Tränkle C, Mohr K: Allosteric ligands for G proteincoupled receptors: a novel strategy with attractive therapeutic opportunities.
Med Res Rev 2010, 30:463549. PubMed Abstract  Publisher Full Text

Ehlert FJ, Griffin MT: Twostate models and the analysis of the allosteric effect of gallamine at the m2 muscarinic receptor.
J Pharmacol Exp Ther 2008, 325:10391060. PubMed Abstract  Publisher Full Text

Gomes I, Ijzerman AP, Ye K, Maillet EL, Devi LA: G proteincoupled receptor heteromerization: a role in allosteric modulation of ligand binding.
Mol Pharmacol 2011, 79:10441052. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Hall DA: Modeling the functional effects of allosteric modulators at pharmacological receptors: an extension of the twostate model of receptor activation.
Mol Pharmacol 2000, 58:14121423. PubMed Abstract  Publisher Full Text

Jäger D, Schmalenbach C, Prilla S, Schrobang J, Kebig A, Sennwitz M: Allosteric small molecules unveil a role of an extracellular E2/transmembrane helix 7 junction for G proteincoupled receptor activation.

Kenakin TP: '7TM receptor allostery: putting numbers to shapeshifting proteins.
Trends Pharmacol Sci 2009, 30:460469. PubMed Abstract  Publisher Full Text

Stahl E, Elmslie G, Ellis J: Allosteric modulation of the M3 muscarinic receptor by amiodarone and Nethylamiodarone: application of the fourligand allosteric twostate model.
Mol Pharmacol 2011, 80:378388. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Leach K, Sexton PM, Christopoulos A: Allosteric GPCR modulators: taking advantage of permissive receptor pharmacology. Supplementary data.
Trends Pharmacol Sci 2007, 28:382389. PubMed Abstract  Publisher Full Text

Black JW, Leff P: Operational models of pharmacological agonism.
Proc R Soc Lond B 1983, 220:141162. PubMed Abstract  Publisher Full Text

Ross EM, Maguire ME, Sturgill TW, Biltonen RL, Gilman AG: Relationship between the betaadrenergic receptor and adenylate cyclase.
J Biol Chem 1977, 252:57615775. PubMed Abstract  Publisher Full Text

De Lean A, Stadel JM, Lefkowitz RJ: A ternary complex model explains the agonist specific binding properties of the adenylate cyclase coupled beta adrenergic receptor.
J Biol Chem 1980, 255:71087117. PubMed Abstract  Publisher Full Text

Stockton JM, Birdsall NJ, Burgen AS, Hulme EC: Modification of the binding properties of muscarinic receptors by gallamine.
Mol Pharmacol 1983, 23:551557. PubMed Abstract  Publisher Full Text

Ehlert FJ: Estimation of the affinities of allosteric ligands using radioligand binding and pharmacological null methods.
Mol Pharmacol 1988, 33:187194. PubMed Abstract  Publisher Full Text

Bradley SJ, Langmead CJ, Watson JM, Challiss RA: Quantitative analysis reveals multiple mechanisms of allosteric modulation of the mGlu5 receptor in rat astroglia.
Mol Pharmacol 2011, 79:874885. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Canals M, Lane JR, Wen A, Scammells PJ, Sexton PM, Christopoulos A: A MonodWymanChangeux mechanism can explain G proteincoupled receptor (GPCR) allosteric modulation.
J Biol Chem 2012, 287:650659. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Kenakin TP: Biased signaling and allosteric machines; new vistas and challenges for drug discovery.
Br J Pharmacol 2012, 165:16591669. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Leach K, Davey AE, Felder CC, Sexton PM, Christopoulos A: The role of transmembrane domain 3 in the actions of orthosteric, allosteric, and atypical agonists of the M4 muscarinic acetylcholine receptor.
Mol Pharmacol 2011, 79:855865. PubMed Abstract  Publisher Full Text

Smith NJ, Ward RJ, Stoddart LA, Hudson BD, Kostenis E, Ulven T: Extracellular loop 2 of the free fatty acid receptor 2 mediates allosterism of a phenylacetamide agoallosteric modulator.
Mol Pharmacol 2011, 80:163173. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Suratman S, Leach K, Sexton P, Felder C, Loiacono R, Christopoulos A: Impact of species variability and 'probedependence' on the detection and in vivo validation of allosteric modulation at the M4 muscarinic acetylcholine receptor.
Br J Pharmacol 2011, 162:16591670. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Wootten D, Savage EE, Valant C, May LT, Sloop KW, Ficorilli J: Allosteric modulation of endogenous metabolites as an avenue for drug discovery.
Mol Pharmacol 2012, 82:281290. PubMed Abstract  Publisher Full Text

Birdsall NJ: Class A GPCR heterodimers: evidence from binding studies.
Trends Pharmacol Sci 2010, 31:499508. PubMed Abstract  Publisher Full Text

Jakubík J, Janícková H, ElFakahany EE, Doležal V: Negative cooperativity in binding of muscarinic receptor agonists and GDP as a measure of agonist efficacy.
Br J Pharmacol 2011, 162:10291044. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Kiselyov VV, Versteyhe S, Gauguin L, De Meyts P: Harmonic oscillator model of the insulin and IGF1 receptors' allosteric binding and activation.

Rovira X, Roche D, Serra J, Kniazeff J, Pin JP, Giraldo J: Modeling the binding and function of metabotropic glutamate receptors.
J Pharmacol Exp Ther 2008, 325:443456. PubMed Abstract  Publisher Full Text

Birnbaumer L, Bearer CF, Iyengar R: A twostate model of an enzyme with an allosteric regulator site capable of metabolizing the regulatory ligand.
J Biol Chem 1980, 255:35523557. PubMed Abstract  Publisher Full Text

Leff P: The twostate model of receptor activation.
Trends Pharmacol Sci 1995, 16:8997. PubMed Abstract  Publisher Full Text

Changeux JP: Allosteric proteins: from regulatory enzymes to receptors  personal recollections.
Bioessays 1993, 15:625634. PubMed Abstract  Publisher Full Text

Monod J, Wyman J, Changeux JP: On the nature of allosteric transitions: a plausible model.
J Mol Biol 1965, 12:88118. PubMed Abstract  Publisher Full Text

Changeux JP: 50th anniversary of the word "allosteric".
Protein Sci 2011, 20:11191124. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Koshland DE Jr: Application of a theory of enzyme specificity to protein synthesis.
Proc Natl Acad Sci 1958, 44:98104. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Katz B, Thesleff S: A study of the desensitization produced by acetylcholine at the motor endplate.
J Physiol 1957, 138:6380. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Furchgott RF: Receptor mechanisms.
Ann Rev Pharmcol 1964, 4:2150. Publisher Full Text

Furchgott RF: The use of βhaloalkylamines in the differentiation of receptors and in the determination of dissociation constants of receptoragonist complexes.

Kenakin TP, Beek D: Is prenalterol (H133/80) really a selective beta 1 adrenoceptor agonist? Tissue selectivity resulting from differences in stimulus–response relationships.
J Pharmacol Exp Ther 1980, 213:406413. PubMed Abstract  Publisher Full Text

Slack RJ, Hall DA: Development of operational models of receptor activation including constitutive receptor activity and their use to determine the efficacy of the chemokine TARC at the CCchemokine receptor CCR4.
Br J Pharmacol 2012, 166:17741792. PubMed Abstract  Publisher Full Text

Ehlert FJ, Suga H, Griffin MT: Analysis of agonism and inverse agonism in functional assays with constitutive activity: estimation of orthosteric ligand affinity constants for active and inactive receptor states.
J Pharmacol Exp Ther 2011, 338:671686. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Perdona E, Costantini VJ, Tessari M, Martinelli P, Carignani C, Valerio E: In vitro and in vivo characterization of the novel GABAB receptor positive allosteric modulator, 2{1[2(4chlorophenyl)5methylpyrazolo[1,5a]pyrimidin7yl]2piperidinyl}ethanol (CMPPE).
Neuropharmacology 2011, 61:957966. PubMed Abstract  Publisher Full Text

Hall DA, Langmead CJ: Matching models to data: a receptor pharmacologist's guide.
Br J Pharmacol 2010, 161:12761290. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Segel IH: Enzyme kinetics. Behavior and analysis of rapid equilibrium and steadystate enzyme systems. New York: Wiley & Sons; 1975.
reissued 1993

Holst B, Frimurer TM, Mokrosinski J, Halkjaer T, Cullberg KB, Underwood CR: Overlapping binding site for the endogenous agonist, smallmolecule agonists, and agoallosteric modulators on the ghrelin receptor.
Mol Pharmacol 2009, 75:4459. PubMed Abstract  Publisher Full Text

Mueller R, Dawson ES, Meiler J, Rodriguez AL, Chauder BA, Bates BS: Discovery of 2(2benzoxazoyl amino)4aryl5cyanopyrimidine as negative allosteric modulators (NAMs) of metabotropic glutamate receptor 5 (mGlu5): from an artificial neural network virtual screen to an in vivo tool compound.
ChemMedChem 2012, 7:406414. PubMed Abstract  Publisher Full Text  PubMed Central Full Text
Prepublication history
The prepublication history for this paper can be accessed here: