receptor signal transduction protocols

As no radioligand binds exclustvely to its receptor, it is also necessary to determine the nonspecific binding of each concentration of radioligand by suppressing its binding to the re

Radiollgand binding is a straightforward technique that measures the bind-

mg of a labeled agonist or antagonist to its receptor It is applicable to a variety

of receptor preparations, ranging from purified receptors to tissue slices, or even whole animals However, membranes or broken-cell preparations are undoubtedly the most widely used Radioligand binding allows the affinity of drugs for their receptors to be determined very readily and It also allows the number of receptors in a tissue or cell to be quantified-something that was impossible before the mtroductlon of binding techniques Furthermore, the technique can be adapted to study the association and dissociation kinetics of llgand binding (1,2), as well as complex allosteric mteractlons between ligands (2) or between receptors and effector molecules, such as guanine nucleotide- binding proteins (G proteins) (2,2)

Despite the ease with which radloligand binding can provide information regarding a wide range of receptors, it does have its limitations There is an absolute requirement for a high-affinity radioligand, selective for the receptor

of Interest Even if such a radioligand exists, it may still be impossible to detect receptors in a particular tissue if receptor abundance is low relative to the non- specific binding of the radioligand As with any technique, radioligand-bind- ing data can be beset with artifacts if experiments are not designed carefully Most importantly, it must always be remembered that radioligand-binding experiments identify ligand-binding sites, which may or may not represent bonafide receptors These points are discussed m more detail m refs 1-3 This chapter provides detailed instructions on how to obtain and analyze equilibrium-binding data for the IP prostanoid (prostacyclin) receptor in human

From Methods m Molecular Bralogy, vol 83 Receptor S/gnal Transducbon Protocols

Edlted by R A J Chalks Humana Press Inc Totowa, NJ

1

platelets and the neuroblastoma x glioma cell line, NGlO%15, using the labeled agonist, [3H]-iloprost, as a radioligand This system illustrates many of the more common problems inherent in the design and analysis of radioligand bind- ing experiments, because [3H]-iloprost is a rather difficult ligand

3 Lysis buffer: 5 mM Tris-HCI, pH 7.4

4 Wash buffer: 50 mMTris-HCl, 0.25 mM EDTA, pH 7.4

5 Assay buffer: 50 mA4Tris-HCI, 5 mMMgCl,, pH 7.4 Note: Items 3-5 may all be prepared in liter quantrties and stored at 4°C until required

6 [3H]-iloprost may be obtained from Amersham International, Amersham, UK The stock should be stored at -20°C and diluted in assay buffer on the day of the experiment

7 Iloprost: Unlabeled iloprost is supplied with [3H]-iloprost from Amersham Inter- national or Schermg AG, Berlin, Germany The stock concentrations supplied may be stored at 4OC, being diluted m assay buffer on the day of the experiment

8 Cell harvester and filters: A 24-place Brandel cell harvester and GF/B filters were used in the experiments presented here Both cell harvester and filters may

be obtained through Semat Technical, St Albans, UK

9 5’-Guanylylimidodiphosphate (GppNHp [Sigma, Poole, UK]): 2.5 mMGppNHp (to give a final concentration of 100 pM) may be made up in I-mL quantities in assay buffer on the day of the experiment and stored on ice Any excess may be frozen at -2O’C and used in a subsequent assay

10 Curve-fitting programs: A wide range of suitable programs are available, but a detailed comparison is beyond the scope of this chapter However, it is useful to choose a program that allows parameters to be constrained to particular values, user-defined equations to be entered, and publication-quality output to be pro- duced, The analysis and figures reproduced here were obtained using FigP (Biosofi, Cambridge, UK)

3 Methods

3.1 Preparation of Platelet Membranes

1 Thaw the bag of frozen platelets in cold water Centrifuge at 80,OOOg for 35 mm

at 4°C Discard the superuatant

2 Resuspend the pellet in roughly 10 mL/U platelets of ice-cold 5 mA4 Tris-HCl,

pH 7.4 Lyse any intact platelets by either stirring the suspension on ice for

30 min or freezing the suspension overnight and then thawing Use the most convenient method

Radioligand-Binding Methods 3

3 Centrifuge the lysed platelets at 80,OOOg for 30 min at 4’C Discard the supema- tam and resuspend the pellet in a convenient volume of ice-cold wash buffer

4 Repeat step 3 once

5 Centrifuge at 80,OOOg for 30 min at 4°C Resuspend the pellet in wash buffer to a concentration of about 20 mg protein/mL.‘As a rough guide, 1 mL of pelleted membranes is equivalent to about 100 mg protein Freeze the final suspension in 1-mL aliquots at -70°C until required

6 On the day of the binding assay, thaw the frozen membranes and dilute 1: 10 with assay buffer Do this a few minutes before startmg the binding assay, so that the membrane suspension just has time to come up to room temperature before pipeting (see Note 2)

3.2 Preparation of NG108-15 Cell Homogenates

1 On the day of the binding assay, thaw the pelleted NGl08- 15 cells and resuspend

in assay buffer, using 20 strokes of a tight-fitting glass Dounce homogenizer (Jencons, Leighton Buzzard, UK) or something similar The pellet from an 80-cm2 cell culture flask homogenized in about 7 mL assay buffer provides sufficient homogenate for a single binding curve in thrs system Prepare the homogenate a few minutes before starting the bmding assay, so that the suspension has Just enough time to come up to room temperature before pipetmg (see Note 2)

2 NG108-15 cells may also be prepared as washed homogenates (see Note 3) or whole cells (see Note 4)

3.3 Direct (Saturation) Binding

This type of assay measures the equilibrium binding of a range of concen- trations of the radioligand As no radioligand binds exclustvely to its receptor,

it is also necessary to determine the nonspecific binding of each concentration

of radioligand by suppressing its binding to the receptor by the inclusion of a saturating concentration of an unlabeled ligand (see Note 5) Specific binding can then be calculated as the difference between total and nonspecific binding (see Section 3.5.)

1 The design of the binding experiment depends largely on the method to be used

in separating bound and free ligand (see Note 6) In the experiments presented here, a Brandel cell harvester was used, which filters 24 samples simultaneously Therefore, each binding curve was designed to use 24 samples; 6 concentrations

of [3H]-iloprost (0.3-100 nM; see Notes 7 and 8), each prepared in the absence

and presence of 10 pM unlabeled iloprost, to define nonspecific binding (see Note 5), with each determination being performed in duplicate

2 Prepare appropriate dilutions of all ligands, and so on, in assay buffer (see Note 9)

3 Pipet the assay components into polypropylene assay tubes (see Note 10) as fol- lows: 30 pL assay buffer, 10 pL [3H]-iloprost, 10 pL assay buffer or iloprost, and

200 & membranes Adding the components in this order minimizes the risk of

cross-contamination (see Note 11)

4 Following the addmon of the membranes, mix each sample well, and leave at room temperature for sufficient time for equilibrium to be achieved, m this case,

30 min (see Note 12) Reserve leftover membranes for protein determination, and dilutions of radioligand for countmg (see step 8, below)

5 If using an unfamiliar radiohgand, it is advisable to perform filter blanks, m which the total and nonspecific bmdmg of the radiollgand is determined m the absence

of any membranes (see Note 13)

6 At the end of the incubatton period, separate bound llgand from free ligand In this case, filter the samples onto GF/B glass fiber filters using a Brandel cell harvester and rapidly (see Note 14) rinse the filters with 3 x 3.5 mL cold buffer (see Note 15)

7 Place the filters m 5-mL scmtillatton vials (see Note 16) and add 4 5 mL emulsi- fying scmtrilant Leave the vials overnight to allow the radiohgand to be extracted from the filters into the scintillant (see Note 17)

8 Prepare radioligand standards Pipet 10 pL of each radroligand dilution mto 5-mL scmtillatron vials and add 4 5 mL scmttllant These provide an accurate estimate of the actual amount of radtoligand added to each sample

9 Count the samples and standards m a scmtillation counter

10 This basic method can be easily modified to examine the effects of modulators

on radiohgand bmding Thus, to examme the effects of 100 p&Y GppNHp on [3H]-tloprost binding, as here, mitially add only 20 pL assay buffer to each tube and then add 10 p.L GppNHp The rest of the assay components are then added as described in step 3, above

3.4 Competition (Displacement) Binding

In a competttion-bindmg assay, the bmding of an unlabeled ligand is measured

by its ability to displace the specific binding of a low, fixed concentration of radioligand This is an extremely useful and versatile techmque, allowing the properties of a wide range of ligands to be investigated rapidly, even those with rather low affinities for the receptor; it 1s widely used as a primary screen- ing technique in drug-discovery programs Furthermore, as the selectivity of the technique is determined by the selecttvtty of the radiohgand, it allows the study of the binding of nonselective hgands to one particular receptor of interest

1 The design of the experiment again depends on the chotce of separation method

In the case of a 24-place cell harvester, as used here, it is convenient to set up each binding curve as follows: 4 nM [3H]-iloprost alone, to determine total bind- ing (see Note 18); 4 nA4 [3H]-iloprost plus 10 concentrations of unlabeled ligand (see Note 19), m the case of the experiments presented here, 0.1-3 pA4 iloprost;

and 4 nA4 [3H]-iloprost plus 10 pMunlabeled iloprost to define nonspecific bind-

mg (each determmation being performed in duplicate)

2 Dilute the ligands and perform the assay exactly as descrtbed for dtrect bmdmg (see Section 3 3.) Remember to count 10 pL of the radioligand dilution as a concentration standard and retain excess membranes for protein determmation

Radioligand-Binding Methods 5 3.5 Data Analysis

A potential danger of radioligand binding is that the very simplicity of the technique may obscure the need for careful experimental design and thought- ful analysis of the data It IS easy to obtain reproducible data, feed it into a computer, and get values out It is much harder to be sure that those values really mean what you hope they mean As with all data analyses, the principle

of rubbish in, rubbish out applies; if the data are highly scattered or riddled with artifacts, any estimates of receptor number or affimty generated by even the most sophisttcated of curve-fittmg techniques will be flawed A detailed consideration of the avoidance of potential artifacts is beyond the scope of this chapter, but is considered in some detail elsewhere (1,3) However, even if the data that go into the analysis are perfectly good, it is still possible to get rub- bish out The hazards of using linearizations of bmdmg data, such as the Scatchard plot, are widely recognized (4,5), and these techniques should be avoided However, the use of nonlinear, curve-fitting techniques also has problems; as the number of variable parameters in the various models of binding increases, the number of combinations of these values that will fit the data more or less equally well also increases greatly There is no guar- antee that the purely mathematical, nonlinear regression analysis will auto- matically converge on the biologically correct combination It is thus helpful to consider this sort of analysis as testing a hypothesis (se> Note 20) rather than automatically yielding precise and meaningful values For effective curve fitting, it is necessary to start with simple models, only increasing their complexity if the data demand; keep the underlying assumptions of the various models m mind, checking that the analysis gives sensible results; and check the results of the analysis for consistency with results from other experiments In order to illustrate these principles, the following sections consider in some detail the processes involved m ana- lyzing some real data-the binding of [3H]-iloprost to platelets and NGlO8-15 cells In the case of [3H]-iloprost, analysis of competition-binding curves is somewhat more straightforward than direct binding, so these will be dealt with first

3.5.1 Analysis of Competition-Binding Curves

I Calculation of radioligand concentration The precise amount of tracer radio- ligand included in each sample is determined from the lo-& standard, as follows: radiohgand (pmol/sample) = (standard dpm)/(specific activity x 2220) where the specific activity of the radioligand is expressed as CMnmol and 2220

is a conversion factor This amount can be converted to a concentration by taking the sample volume into account In this case, 10 pL radioligand was included in

of the same model to the data obtained in the presence of GppNHp (see Section 3.5.1 for details)

a final assay volume of 250 & Thus the radioligand concentration in nM(=pmol/mL)

is given by:

[radioligand] (nM) = (standard dpm x 4)/(specific activity x 2220)

2 Calculation of % inhibition of specific binding The data presented here (Figs 1 and 2) are expressed as % inhibition of specific binding of [3H]-iloprost, which IS calculated as follows:

% inhibition = 100 x { I- [(sample dpm - nsb dpm)/(total dpm - nsb dpm)]) where sample dpm is the mean dpm in each sample containing a particular concentration of the unlabeled lrgand, total dpm is the mean dpm in the samples containing [3H]-rloprost alone, and nsb dpm is the mean dpm in the samples containing 10 p.M unlabeled rloprost to define nonspecific binding This transformation of the data IS useful in that it incorporates total and non-

Radioligand-Binding Methods 7

specific binding into the inhibition curve so that low concentrations of unla- beled ligand should inhibit 0% specific binding and high concentrations inhibit 100% However, the transformation is by no means essential and may incorpo- rate errors, if estimates of total and nonspecific binding are not accurate (see Note 2 1)

3 The effect of GppNHp on iloprost binding to platelet membranes

a The inhibition of the specific bindmg of 7.3 nA4 [3H]-iloprost by unlabeled iloprost in the absence and presence of 100 pM GppNHp is shown m Fig 1 The first step in the analysis of any binding data should be to determine the best fit of the simplest model of binding to the data: the simple Langmuir isotherm, or single-site model This model assumes binding to a single popu- lation of noninteracting sites, so that:

WI = &a, P-W + PI)

where [LR] is the concentration of ligand-receptor complexes (equivalent

to % inhibition in this case), B,,, is the maximal binding capacity or total number of receptor sites (which should be 100% inhibition in the case of competition binding data), and K is an estimate of ligand-binding affinity (see part c, below) Curve fitting is performed by computer assisted, nonlin- ear regression analysis A wide range of suitable programs are available (see Section 2.)

Fitting the single-site model to the binding curves in Fig 1 yielded the following estimates for K and B,,,:

in each case B,,, is close to 1 00%, as expected, and the estimates of K obtained

m the absence and presence of GppNHp are very similar, in accord with the observation that GppNHp appears to have very little effect on iloprost bind- ing under these circumstances (see Fig 1) Thus the fit of the single-site bind- ing model to the data seems satisfactory, and no further curve fitting IS required (or justified) in this case

c Correction for the presence of [3H]-iloprost Competition-binding curves, such as those shown in Fig 1, are necessarily obtained in the presence of a low, fixed concentration of radiohgand The presence of this radioligand affects the position of the binding curve for the unlabeled ligand; it is shifted

to the right of its true position by a factor determined by the radiohgand aftin- ity and concentration Thus, it is necessary to correct estimates of affinity obtained from competition studies to take this rightward shift into account, using the Cheng-Prusoff equation (6):

true Kd = K/{ 1 + ([L*]/K*)}

of the same model, in which B,,, is constrained to 100% (B) The solid lines represent the

where true Kd IS the corrected dissociation constant for the unlabeled ligand, [L*] is the radioligand concentration (determined from the radioligand stan- dard; see step 1, above), andK* is the dissociation constant for the radioligand, determined from a direct-binding assay (see Section 3.5 2 )

In the case of a self-competition assay, as here, where the same ltgand is used as both radioligand and competing ligand, the Cheng-Prusoff equation simplifies to:

4 The effect of GppNHp on iloprost binding to NG108- 15 cell homogenates:

a The inhibition of the specific binding of 2.4 nM[3H]-iloprost by unlabeled iloprost

in the absence and presence of 100 @fGppNHp is shown in Fig 2A Again, the first step in the analysis of these data should be to determine the best fit of a smgle-site model to the data This yielded the followmg estimates for K and B,,,,

no GppNHp + GPPNHP

K 6.90 x 10-9A4 2.31 x lVsh4

B max 100% (fixed) 100% (fixed)

The predicted curves given by this procedure are shown in Fig 2A (solid lines) The curves still fit the data well and are thus likely to represent a better estimate of K than those obtained when B,,.,,, is allowed to converge on a value of less than 100% (see Note 22)

c The close agreement between the data and the single-site binding curves may suggest that any further analysis of this data using more complex models is unJustitied However, the rightward shift in the binding curve obtained in the presence of GppNHp is characteristic of G protein-coupled receptors It seems

to reflect the ability of the guanme nucleottde to disrupt receptor-G protein complexes (RG), which have high affinity for agonists, converting them to uncoupled receptors (R), which have low agomst affinity If the results of the

Radioligand-Binding Methods 11

single-site fits are accepted at face value, they tend to suggest that, in the absence of GppNHp, 100% of the IP receptors exist as RG complexes and that all of these are uncoupled by GppNHp, to give a homogenous population

of low-affinity agonist sites While not impossible, this scenario is unlikely Usually, in the absence of guanine nucleotides, agonists recognize a mixture

of high- and low-affinity sites, reflecting a mixture of RG and R In the pres- ence of guanme nucleotide, the agonist-binding curve is shifted to the right and steepened, because of the conversion of high-affinity RG sites to low- affinity R sites From inspection of the data, it is obvious that GppNHp shifts the iloprost-binding curve to the right The best way to determine the steep- ness of the curve is to fit the Hill equation to the data:

where LSo is the midpoint location parameter of the curve (in this case the concentration of ligand inhibiting 50% of the specific binding) and n is the pseudo-Hill coefficient, or slope factor If the data correspond to binding to a single population of nonmteracting sites, la is equal to 1 and the Hill equation simplifies to the single site model If n > 1, it tends to indicate positive cooperativity between binding sites; if n < 1, it tends to indicate binding to a heterogenous population of sites (see Note 23)

In many curve-fitting programs, the Hill equation is expressed in a slightly different format For example, in FigP:

amount bound = [min + (max - min.)]/{1 + ([L]/LSo)*)

where min is the minimum level of binding, obtained as ligand concentration tends to zero, and max is the maximum level of binding, obtained as ligand concentration tends to 00 This form is particularly useful for fitting untrans- formed data and estimating levels of total and nonspecific binding (see Note 21) In analyzing the data from NG108-15 cells, min and max were con- strained as 0 and loo%, respectively, and the best fit of the model to the data

is shown in Fig 2B:

L50

n

no GppNHp 6.88 x lo-9A!f 0.79

+ GPPNHP 2.30 x 10-8M 0.82 Thus, for both curves, the slope factor is less than one, indicating a degree of binding site heterogeneity Furthermore, the curve obtained in the presence of GppNHp does appear to be slightly steeper in the presence of GppNHp than

in its absence

d Based on the shallow slope factors and the working hypothesis regardmg the effects of GppNHp, the best fit of a two-site model to the data was determined The two-site binding model is simply the sum of two single-site models:

The best fit of this model to the NG108-15 data is shown in Fig 2C and yielded the followmg parameters:

RG sites to R by guanine nucleotides: The expected result would be the same values of K, and K2 m both the absence and presence of GppNHp, with only the relative proportions of the two sites differmg between the two curves Therefore, the next step in the analysts should be to determme how well the data can be described by a model that meets these criteria

e There are curve-fitting packages available that allow the stmultaneous fit- ting of more than one curve, so that various parameters (e.g., KI and K2) can be shared between the various curves (see Note 24) However, the most widely available and user-friendly programs do not have this facility It is, of course, possible to keep trying different combinattons of values for K1 and K2 until you hit on values that provide a good fit to both curves However, in this case, a somewhat simpler approach has been adopted Given that the bindmg curve obtained in the presence of GppNHp can be well described by a single-site binding model, it is assumed that this curve represents binding to a pre- dominantly homogenous population of low-affinity R sites Thus the affin- ity value obtained from this single-site fit should at least approxtmate the affimty of the agonist for R (Kz m the two-site model) This approach gave the following values:

The predicted curves fit the data well (see Fig 2D) and the sum of Bmaxl and

B max2 is close to 100% The data can at least be said to be consistent with a model in which guamne nucleotides promote the conversion of a subpopula- tion of high-affinity RG sites to low-affinity R sites

f The final step is to convert the apparent affinity values from the competi- tion binding experiments to true affinity values, as described above for the platelet data Thus:

Radioligand-Binding Methods 13

Kdl

Kd2

no GppNHp 0.02 x lo-sA4

2.07 x lO-%U

+ GPPNHP

-

2.07 x lO+V

3.5.2 Analysis of Direct-Binding Curves

1 Calculation of radioligand concentrations These are determined as described in Section 3.5.1

2 Calculation of specific binding For each concentration of radtoligand, specttic binding 1s calculated as the difference between total and nonspecific binding This obviously means that any errors in the determination of nonspecific binding will be incorporated mto the estimate of specific binding In the vast majority of cases, nonspecific binding is a linear function of radtohgand concentration, and this can be used to check the accuracy of the nonspecific binding data Many curve-fitting packages include a linear, “nonspecific binding” component in their one- and two-site models (see Note 25), and it is thus possible to estimate both specific and nonspecific binding from the total binding curve However, this adds an extra level of complexity (and therefore variability) to the model and

it is better to eliminate nonspecific binding at the beginning The equation used to determine radiohgand concentration from dpm can then be used to convert the amount of radiohgand bound to pmol/sample However, it is prob- ably easier to leave these values as dpmsample until the end of the analysts, as here (Figs 3 and 4)

3 The effect of GppNHp on the specitic binding of [3H]-rloprost to platelet membranes

a The specrtic binding of 0.3-100 nA4 [3H]-iloprost in the absence and presence

of 100 @f GppNHp is shown in Fig 3A As previously, the first step in the analysis of these data is to determine the best fit of a single-site binding model (indicated by the solid lines in Fig 3A):

are reasonably similar, which fits in with the expected effects of GppNHp in this system However, comparison of the Kd values with those obtained from the competition data (see Section 3.5.1.) suggest that the binding affinity of [3H]-iloprost is somewhat lower than that of unlabeled iloprost in the same membranes

Radioligand-Binding Methods

b A logical next step in the analysis is therefore to see how well the direct- bmding data can be described by the model if Kd is constrained to the values obtained from the competition experiment (indicated by the dotted lines in Fig 3A):

c There is in fact a problem wtth using [3H]-iloprost m direct-binding curves, which underlies the apparent differences between affinity values obtained in direct and competition assays At concentrations any higher than 10 nA4, a significant proportion of the iloprost-displaceable binding of [3H]-iloprost is

to a low-affinity, high-capacity, nonreceptor site, which probably corresponds

to the low-affinity prostanoid site found in many tissues (2,3) The presence

of this second site ts more obvious m the direct-bmding curves obtained in NGl OS-15 cell homogenates (see point 4, below) It would be theoretically possible to account for this second site using a two-site model of binding However, the two-site model has four variable parameters, and it is probably not justifiable to fit a data set of only 6 points (as here) to such a model In practice, direct-binding curves for [3H]-iloprost can be well described by a single-site model, with the addition of a linear component (the component of nonspecific bmdmg included in this equation in many commercially avail- able packages):

[LR] = Bmax [L]/K,, + CL]/+ C[L]

This tends to imply that, over the concentration range used, the occupancy of the second, nonreceptor site by [3H]-iloprost is very low Under these condi- tions, the single-site binding model approximates to a straight line

The best fit of this model to the data is shown in Fig 3B, the fits giving the following parameters:

Fig 3 (previous page) The specific binding of [3H]-iloprost measured in the absence (A) and presence (V) of 100 pMGppNHp in human platelet membranes The lines represent the best fit of various models to the data (A) The solid lines represent the best fit of a single-site model of binding; the dotted lines represent the best fit of the same model, in which Kd has been constrained to the values obtained from the competition experiment (B) The solid lines represent the best fit of a single-site binding model with the addition of a linear component to the same data (see Section 3.5.2 for details)

to obscure this heterogeneity (see Note 26) It IS important to note that ignor- ing the existence of the nonreceptor sate, and takmg the single-site fits to the platelet data at face value, leads to a substantial overestimate of B,,, (the number of receptors in the tissue) and an underestimate of lloprost affinity

d Knowing the specific activity of the [3H]-rloprost (14.7 Wmmol) and the concentration of protein in the homogenate used in the assay (0.8 mg/mL), the values obtained for B,,,= can now be converted from dpmlsample to finol.Jmg protein, as described m Section 3.5.1.:

no GppNHp + GPPNHP

B max 505 fmol/mg 560 fmol/mg

e If the purpose of these experiments was to investigate the effect of GppNHp

on binding m more detail, it might seem appropriate to fit a two-site model of binding to the data, again incorporating a linear, nonreceptor component However, this sort of analysis would require more experiments to be per- formed; there is an insufficient number of data points here to Justify the use of such a complex model It would be necessary to obtain more detailed direct- binding curves, with a greater number of data points incorporated over the same concentration range (see Note 19)

Fig 4 (previous page) The specific binding of [3H]-iloprost measured in the absence (A) and presence (V) of 100 pM GppNHp in NGlOS-15 cell homogenates The lines represent the best fit of various models to the data (A) The solid lines repre- sent the best fit of a single-site model of binding (B) The dotted lines represent the best fit of a single-site binding model, with the addition of a linear component to the same data; the solid lines represent the best fit of the same model to the data, but,

in the case of the data obtained in the presence of GppNHp, Kd has been constrained to the value obtained from the competition experiment (see Section 3.5.2 for details)

4 The effect of GppNHp on the specific binding of [3H]-iloprost to NG 108-l 5 cell homogenates

a The specific binding of 0.3-100 nM [3H]-iloprost in the absence and presence

of 100 @4 GppNHp is shown in Fig 4A The curve-fitting strategy used here

is the same as that outlined above for the data from platelet membranes Ini- tially, a single-site model was fitted to the data:

b Fitting the data to a single-site model, with the addition of a linear compo- nent, gave predicted curves, shown as the dotted lines in Fig 4B, and the following parameters:

of iloprost affinity obtained in the presence of GppNHp does not agree very well with that estimated in the competition experiment

c In order to investigate whether these data are compatible with the competition data, the effect of constraining the affinity of [3H]-iloprost in the presence of GppNHp to the value obtained in the competition study on the fit of the single- site model + linear component was examined:

1 Further details regarding the growth of NGl OS- 15 cells may be found in ref 7

2 Cold membrane suspensions tend to clump together, making accurate pipeting difficult The suspension will be more homogenous if warmed to room tempera- ture before pipeting However, the membranes will deteriorate if kept at room temperature for too long

3 If necessary, washed membrane preparations may be prepared from NGl OS- 15 cells by several subsequent centrifugation and resuspension steps, as described for the preparation of platelet membranes (Section 3.1.) In this case, after the first centrifugation step, it is useful to filter the suspension through muslin, in order to remove the large tangle of DNA that will have formed However, in the system used here, washed membrane preparations appear to offer no advantages over crude homogenates

4 If required, NG108-15 cells can be harvested on the day of assay by agitation in phosphate buffered saline and suspended in an isotonic medium, such as gassed Krebs or cell culture medium, to allow the binding of ligands to intact cells to be measured In the case of [3H]-.iloprost, this presents considerable problems: [3H]-iloprost is a very hydrophobic ligand that readily penetrates intact cells, thus leading to high levels of nonspecific binding However, binding to intact cells can be extremely useful as a means of measuring binding under physiologi- cal conditions (81, or to investigate receptor internalization, using hydrophilic ligands that do not penetrate the cells (9) Nevertheless, the use of intact cells does represent an additional level of complexity compared with membrane prepa- rations, and is more prone to artifacts; in particular, receptor-mediated internal- ization of ligands may give rise to a large component of apparently displaceable binding, and estimates of B,,, that increase with incubation time

5 In order to define nonspecific binding, it is best to use an unlabeled ligand that

is different from the radioligand, and ideally one that is as structurally distinct as possible This minimizes the risk of displaceable nonreceptor binding, to enzymes

or uptake-sites, for example However, this is not always possible, either because

no suitable ligands exist, or because their use would be prohibitively expensive, especially since they need to be used at concentrations sufficient to occupy all the available receptors, even when competing with the highest concentration of radioligand

6 Choice of separation method A wide variety of possible separation methods are available, including equilibrium dialysis, gel filtration, centrifugation, filtration, and so on The choice depends on the receptor preparation (e.g., gel filtration may be most suitable for soluble or solubilized receptors; simple rinsing may be best for receptor autoradiography), as well as the radioligand used For example,

if the ligand has a rather low affinity for the receptor, significant amounts may drssoctate durmg filtratron; centrifugatton may be better, but is not appropriate if the ligand displays a htgh degree of nonspectfic bmdmg Of the various separa- tion methods, filtration is the most wtdely used It is very fast, reproductble, and usually incorporates a washing step to reduce the amount of nonspecific bmdmg caused by loose association of the radtolrgand with membrane fragments, and so on

7 The chotce of radtohgand may be a rather trivial matter; there may only be one available In the event of a choice, high-affinity bgands are generally preferred; they can be used at lower concentrations, which tends to reduce both the cost and the level of nonspecific binding, and they are less likely to dissociate from the receptor durmg the separation procedure However, if the affinity of the radiohgand 1s too high (Kd C 0.1 nM), it may take an impractically long time for equilibrium to be achieved Furthermore, the level of nonspecific binding for different ligands in different tissues may vary enormously; the only way to find the most appropriate ligand may be by trial and error Antagonist hgands are generally preferred to agonists, partly because antagomsts often exhibit higher affinity than agonists, and partly because the binding of agonists tends to be more sensitive to the assay conditions For example, the bmdmg of an agonist to a G protein-coupled receptor will depend on the state of receptor G protein couplmg, and the binding of an antagonist will not It is obviously best if the radrohgand displays a high degree of selecttvtty for the receptor of interest However, if nec- essary, it may be possible to suppress bmdmg to unwanted receptors by mcluding

a saturating concentration of an unlabeled ligand that 1s highly selective for that receptor [3H]-iloprost does not bind exclusively to the IP prostanoid receptor; it also exhibits high affinity for the EP, receptor (20) The more selective IP recep- tor ligand, cicaprost, can be used to determine whether this EP, receptor bmdmg presents a problem in any particular tissue; tf cicaprost inhibits significantly less than 100% of the specrfic bmdmg of [3H]-iloprost, it is hkely that the remamder represents binding to the EP, receptor Fortunately, EPi receptor binding does not appear to be a problem m either platelets or NG108-15 cells

8 The choice of radtoligand concentrations m a saturation experiment depends partly on the separatton method to be used (which dictates the number of samples that can be conveniently incorporated mto a single-bmdmg curve) and partly on the question to be addressed In this case, the different concentrations of [3H]-iloprost have been chosen to be separated by half-decades on a log scale and span the expected affinity of [3H]-tloprost, approx 10 nM If the affinity of the radiobgand is unknown, rt may be necessary to perform range-finding experi- ments, with fewer concentrattons over a wider range In order to tit complex binding models to the data, it would be necessary to include a greater number of ligand concentrations over the concentration range of interest

9 Choice of assay buffer Radiohgand binding can usually be performed m very simple buffer systems The choice depends on the preparation (e.g., whole cells require isotonic oxygenated medmm), the bgand (e.g., it may be neces- sary to perform binding in a high ionic strength buffer to reduce nonspecific

Radioligand-Binding Methods 21

binding to filters; Mg2+ ions are essential for the binding of [3H]-iloprost

[Zl]), and the question to be addressed (e.g., Mg2+ ions are required for the interaction between receptors and G proteins) The precise conditrons of the bmding assay (temperature, ionic composition of buffers, and so on) can affect ligand affinity; it is important to consider this when comparing data from different laboratories

10 It is possible to lose substantial amounts of some particularly sticky ligands (both labeled and unlabeled) that may adhere to the sides of the tubes durmg dilution and during the assay It may be possible to reduce this by using silamzed glass tubes, or by including bovine serum albumin in the buffer for peptide ligands, and so on

11 One of the largest sources of error in radioligand-binding experiments is poor pipetmg technique It is important to add low concentrattons of ligands before high ones, to ensure that all additions reach the bottom of the assay tube, and to mix everything thoroughly

12 Sufficient time needs to be allowed for equtlibrium to be achieved, but this needs

to be balanced against the stability of the receptor preparation and/or the ligands Receptor stability can be improved at low temperatures, but this will also slow the rate of approach to equilibrium Eqmlibrmm is approached most slowly at the lowest concentrations of radioligand and m competition assays

A fuller discussion of the problems of nonequtlibrmrn binding may be found in ref 12

13 Radioligands commonly bind to filters; this contributes to measured level of non- specific binding It is often possible to reduce the level of filter bmdmg by, for example, increasing the ionic strength of the assay buffer or presoaking the fil- ters in 1% polyethylenetmine Filter binding presents a particular problem if it is displaceable by the unlabeled ligand used to define nonspecific binding, m this case it may be mdistmguishable from receptor binding

14 The rapid removal of free ligand during filtration necessarily disrupts the equilibrium and promotes dissociation of ligand from the receptor This loss can be considerable with low-affinity ligands, rendering filtration unsuitable

In all cases, the impact of dissociation should be minimized by filtering and rinsing the membranes as rapidly and reproducibly as possible, usmg ice- cold buffers throughout

15 Ideally, the filters should be rinsed with the assay buffer However, the procedure gets through a great deal of buffer, and buffered tap water (cold tap water with a dash of buffer, to bring its pH to 7.4) usually works Just as well

16 The filters should be put close to the bottom of the vial to ensure that they are covered by scintillation fluid-but not scnmched up, because this slows the rate

at which the radioactivity can be extracted by the scintillant

17 Following absorption of the scintillant, the filter becomes transparent and does not appear to interfere with counting However, make sure that any inhomogene- ity filter on the scintillation counter is switched off, otherwise, all the samples may be reJected

18 The concentration of radioligand in competition studies needs to be kept as low

as possible, to minimize the shift of the competition curves (see Section 3.5 l.), while still obtaining sufficient binding to allow the degree of inhibition of this binding to be accurately determined The concentration of receptors in the assay may also need to be considered When a low concentration of a high-affinity ligand binds to an abundant population of receptors, the free-ligand concentra- tion may he substantially reduced, a phenomenon known as depletion If the free- radioligand concentration is reduced by more than 10% by depletton, the receptor preparation needs to be diluted

19 An advantage of the competition technique is that the unlabeled ligands used can

be of low affinity or rather nonselective The choice of concentration range depends on the same factors as considered above for radioligands in a direct- binding assay (see Note 8)

20 Testing a hypothesis is exactly what curve fitting is, in that It involves determm- ing the best fit of various theoretical models to the data The choice of model is therefore, all important, and it is worth bearing in mind that the underlying assumptions of the model may or may not be completely accurate For example, the two-site model used here to analyze agonist-bindmg assumes the existence of two populations of noninterconverting sites However, RG can clearly be con- verted to R under appropriate conditions, such as the addition of guanine nucle- otides Curve-fitting techniques that do not rely on an underlying model have been described (13)

2 1 Calculation of % inhibition of specific binding assumes that the estimates of total and nonspecific binding are absolutely accurate; but, of course, they may not be

It is possible to obtain independent estimates of total and nonspecific binding by determining the best fit of the Hill equation to the untransformed data, in which case total and nonspecific binding will correspond to min and max respectively (see Section 3.5.2.)

22 If an unlabeled ligand consistently inhibits less (or more) than 100% of the spe- cific binding of the radioligand, it may suggest that something is wrong with the definition of nonspecific binding Inhibition of less than 100% of specific bind- ing may legitimately occur in the case of an unlabeled ligand, highly specific for

a particular receptor subtype, inhibiting binding of a nonselective radioligand to

a heterogeneous population of receptors It may also occur if the unlabeled ligand inhibits radioligand binding by an allosteric mechanism, in which both ligands can bind simultaneously to the receptor (14)

23 Steep curves (n > 1) can indicate that the ligand exhibits positive cooperatlvity How- ever, such curves can also arise artifactually, either because binding did not reach equilibrium (12) or because of ligand depletion (15) Shallow curves (n < 1) usually indicate binding-site heterogeneity, but can also arise from negative cooperativity

24 A generally applicable program that enables the simultaneous fitting of several curves is available as part of BMDP (BMDP Statistical SoRware, Cork, Ireland) However, the use of this package is rather cumbersome and requires a working knowledge of Fortran

25 It is important to constrain this linear, nonspecific binding component to zero in most cases (for example, when nonspecific binding has already been subtracted from the data, or in competition-binding studies, in which the nonspecific bind- ing of the unlabeled ligand is necessarily invisible)

26 Binding-site heterogeneity will be obscured in competition studies in which the radioligand used exhibits selectivity for one of the populations of sites For example,

as an agonist, [3H]-iloprost has higher affinity for RG than for R Thus, the right- ward shift of the high-affinity component of the competition curve for an unla- beled agonist will be greater than for the low-affinity component Hence, the competition curve is steepened In extreme cases, low concentrations of labeled agonists will only bind to high-affinity RG sites In this case, guanine nucleotides appear to produce a decrease in the B,,, for the labeled agonist, with no change

in agonist affinity

References

1 Hulme, E C., ed (1992) Receptor-Ligand Znteractzons A Practical Approach

IRL Press at Oxford University Press, Oxford, UK

2 Keen, M and MacDermot, J (1993) Analysis of receptors by radioligand bind- ing, in Receptor Autoradiography: Principles and Practice (Wharton, J and Polak, J M., eds.), Oxford University Press, Oxford, UK, pp 23-56

3 Keen, M (1995) The problems and pitfalls of radioligand binding, in Methods in Molecular Bzology, vol 41 Szgnal Transduction Protocols (Kendall, D A and Hill, S J., eds.), Humana, Totowa, NJ, pp 1-16

4 Klotz, I M (1982) Numbers of receptor sites from Scatchard graphs: facts and fantasies Science 217, 1247-1249

5 Burgisser, E (1984) Radioligand-receptor binding studies: What’s wrong with the Scatchard analysis? Trends Pharmacol Sci 5, 142-144

6 Cheng, Y C and Prusoff, W H (1973) Relationship between the inhibition con- stant Kl and the concentration of inhibitor which causes 50% inhibition (I&e) of

an enzymic reaction Biochem Pharmacol 22,3099-3 108

7 Kelly, E., Keen, M., Nobbs, P., and MacDermot, J (1990) Segregation of discrete G,,-mediated responses that accompany homologous or heterologous desensiti- zation in two related somatic hybrids Br J Pharmacol 99,309-3 16

8 Nathanson, N M (1983) Binding of agonists and antagonists to muscarimc ace- tylcholine receptors on intact cultured heart cells J Neurochem 41, 1545-1549

9 Sibley, D R and Lefkowitz, R J (1985) Molecular mechanisms of receptor desensitization using the P-adrenergic receptor-coupled adenylate cyclase system

as a model Nature 317, 124-129

10 Wise, H and Jones, R L (1996) Focus on prostacyclin and its novel mimetics

Trends Pharmacol Sci 17, 17-21

11 MacDermot, J., Blair, I., and Cresp, T M (198 1) Prostacyclin receptors of a neu- ronal hybrid cell line; divalent cations and ligand-receptor coupling Biochem Pharmacol 30,2041-2044

12 Motulsky, H J and Mahan, L C (1984) The kinetics of competitive radioligand binding predicted by the law of mass action Mol Pharmacol 25, l-9

13 Tobler, H J and Engel, G (1983) Affinity spectra: a novel way for the evaluation

of equilibrium binding experiments Naunyn-Schmiedeberg’s Arch Pharmacol 322,183-192

14 Stockton, J M., Birdsall, N J M., Burgen, A S V., and Hulme, E C (1983) Modification of the binding properties of muscarinic acetylcholine receptors by gallamine Mol Pharmacol 23,55 l-557

15 Wells, J W., Birdsall, N J M., Burgen, A S V., and Hulme, E C (1980) Com- petitive binding studies with multiple sites: effects arising from depletion of free radioligand Biochim Btophys Acta 632,464-469

Site-Directed Mutagenesis and Chimeric Receptors

in the Study of Receptor-Ligand Binding

Mark E Olah and Gary L Stiles ’

1 Introduction

Prior to the cloning of G protein-coupled receptors (GPCRs), structure-func- tion analysis of the ligand-binding properties of these receptors had for the most part been limited to study of the structure-activity relationships (SAR) of agonists and antagonists at the individual receptors For a particular receptor, these SAR data were obtained via synthesis of series of chemically modified hgands, followed by the determination of their binding activities in tissues or cells that natively expressed the receptor of interest With the molecular clon- ing of a multitude of GPCRs over the last several years, the amino acid sequence of these receptors is now known With this knowledge, structural features of the receptors that are involved in ligand binding may now be explored in mutagenesis studies Indeed, a principal focus of receptor research over recent years has been the detailed structure-function analysis of the ligand-binding properties of genetically engineered receptors heterologously expressed in the appropriate cell systems Through such studtes, receptor regions and even single amino acids involved in ligand recognition have been identified This chapter details the techniques used in our laboratory for the construction and analysis of receptors possessing single ammo acid point mutations and receptors composed of amino acid sequence derived from two parent wild-type receptors, i.e., chimeric receptors Specifically, we have used these techniques to study the structural features responsible for the bind- ing properties of different adenosine receptor subtypes (1-3) Similar approaches have been used by many laboratories focusing on many different GPCRs (reviewed in refs 4-6)

From Methods m Mo/ecu/ar Slology, vol 83 Receptor SIgnal Transduction Protocols

Edlted by R A J Chalks Humana Press Inc , Totowa, NJ

25

1.1 Design of Mutated Receptors

of amino acids to be targeted For example, adrenergic receptors were the first GPCRs to be examined in mutagenesis studies Because catecholamines pos- sessing a nitrogen that can be protonated are the prototypical agonist ligands for this receptor family, it was reasoned that a negatively charged amino acid

of the receptor may serve as a counter ion for this critical functional group (4)

It was subsequently shown that substitution of asparagine for an aspartate resi- due in transmembrane domain 3 of the P-adrenergic receptor nearly abolished agonist binding (7) Amino acid targets may also be suggested by the sensitiv- ity of receptor ligand binding to specific chemical treatments For example, treatment of rat brain membranes with the histidine specific reagent diethyl- pyrocarbonate was demonstrated to perturb agonist and antagonist binding by the A, adenosine receptor (8) Site-directed mutagenesis of histidine residues

in transmembrane domain 6 and transmembrane domain 7 of the A1 adenosine receptor has subsequently shown the rmportance of these residues in ligand binding (r) Computer models of the three-dimensional ligand binding pocket

of GPCRs have been used to select potentially important amino acid residues that have been examined in receptor mutagenesis studies (9) A more indirect rationale for targeting an individual amino acid may include its conservation in

a specific location in all members of a specific receptor family, thus suggesting

a critical function

Once an amino acid has been targeted for site-directed mutagenesis, the replacement residue must be selected Frequently, alanine is chosen as the resi- due with which to replace the wild-type amino acid Alanine is of small mass, making it relatively less likely to substantially disrupt protein structure Addi- tionally, alanine is unhkely to form bonds with the ligand(s) under examina- tion However, it may be found that other residues are more suitable than alanine for the substitution For example, replacement of a bulky amino acid, such as tyrosine, in a transmembrane domain with the smaller alanine may be found to disrupt overall receptor conformation Thus, an amino acid of similar

size, but lacking the propensity to participate m similar chemical interactions with the ligand, may be a more appropriate selection Frequently, it is of inter- est to make reciprocal point mutations in which individual amino acids are swapped among receptor subtypes or species homologs in an attempt to define the structural basis for distinct pharmacologic profiles of the wild-type recep- tors Replacement of wild-type residues with additional amino acids is further discussed in Section 4

7.7 2 Chimeric Receptors

The vast majority of chimeric receptors employed to study ligand binding are constructed from two parent receptors that belong to the same receptor family For example, the initial study employing chimeric proteins to map ligand-binding domains of GPCRs employed sequences derived from the a2- and P2-adrenergtc receptors (10) In selecting the wild-type receptors that will con- stitute the sequence of the chimeric receptor, the degree of similarity in the wild-type receptors regarding pharmacological profiles and amino acid sequences must be considered Ideally, the parent receptors may share nearly identical binding affinity for an individual compound or one class of ligands, but differ substantially in the affinity for a separate class of compounds These distinc- tions may exist between different subtypes m a receptor family or perhaps between species homologs of the same receptor For example, our laboratory has employed chimeric receptors composed of sequences derived from the bovine A, adenosme receptor and the rat A3 adenosine receptor to examine the structural requirements for ligand binding by adenosine receptors The two wild-type receptors, as well as receptor chimeras derived from their sequence, bind the agonist ‘251-AB-MECA with very similar high affinity, thus permit- ting receptor quantification However, the very different affinities of the parent adenosine receptors for certain other agonists, and all antagonists, has allowed for receptor domains recognizing these latter ligands to be identified through the study of chlmeric receptors (2,3) With both chimeric receptors and recep- tors possessing pomt mutations, the ability of constructs to maintain high- affinity binding of at least one radioligand is very advantageous Such binding permits the analysis via competition-binding assays of other classes of ligands that may display decreases in affinity of several orders of magnitude

Additional factors will influence which specific regions of the parent wild- type receptors to exchange in the construction of chimenc receptors As with selection of the parent receptors, the specific receptors and ligands under examination must be considered when identifying the receptor segments to be swapped For example, the binding of most small ligands to their receptors, e.g , catecholamine binding by adrenergic receptors, appears to exclusively involve receptor transmembrane domains; receptor extracellular domains have

been shown to be involved in ligand recognition by several of the GPCRs that are activated by peptide agonists, such as the tachykinins (4,I1) Often, one of two systematic approaches is taken m the construction of chimeric receptors First, a series of chimeras may be created with each construct focusing on a distinct transmembrane domain or extracellular region Alternatively, an ml- tial study may examine one or two chimeric receptors composed of the replacement of multiple receptor regions, with the resulting data used to design subsequent chimeras of more defined substitution,

Once the general receptor regions have been chosen for inclusion in a chimeric receptor, the precise splicing points should be selected Of con- cern is the ability of the receptor to tolerate sequence substitution at these points A major obstacle in the study of chimeric receptors is the lack of detectable expression of certain constructs, which obviously precludes analysis Though rigid guidelines do not exist, it 1s often preferable to avoid initiating sequence replacements in receptor transmembrane domains The probability of success with substitutions at precisely the interface of a trans- membrane domain and an adjoining loop segment is perhaps slightly greater than that of the above option In these situations, the chance of obtaining chi- merit receptor expression may be increased if the sequences of the two parent wild-type receptors are highly homologous in the splicing region Conversely, splicing in the loop regions of GPCRs is often well tolerated and may be the preferred strategy m the construction of initial chimeras Additional factors in the design of receptors possessing point mutations and of chimeric receptors are discussed in Section 4

2 Methods for Creation of Mutant Receptors

The techniques employed in our laboratory for construction of receptors containmg point mutations as well as chimeric receptors utilize the polymerase chain reaction (PCR) For pomt mutations, the design of the oligonucleotide primers and strategy employed in the PCR reactions are based on the method for site-directed mutagenesis described by Nelson and Long (12) The con- struction of chimeric receptors follows a similar approach, but also mcorpo- rates the use of splicing oligonucleotides as described by Yon and Fried (13) Thus, the methods for creating point mutations or chimeric receptors are rela- tively similar, with the construction of the latter requiring additional oligo- nucleotide primers and an additional PCR step Described below in detail are the methods for construction of the two types of mutant receptors The procedure for point mutations is described first because it is relatively simple; the construc- tion of receptor chimeras is an extension of this procedure Alternative tech- niques for the site-directed mutagenesis of GPCRs (14), as well as the creation of chimeric receptors (15), have recently been outlined in excellent reviews

Receptor-Ligand Binding 29 2.7 Point Mutations

The template to be used in the construction of point mutations may be the isolated wild-type receptor cDNA of interest or, more frequently, the receptor cDNA subcloned into a vector such as pBluescript or an expression vector In the latter instances, the construct does not need to be lineartzed prior to the PCR reaction Prior to the design of the oligonucleotide primers, the receptor cDNA must be analyzed for the existence of two unique restriction endonu- clease sites that flank the area selected for mutation Following the PCR gen- eration of the mutated fragment, these sites will ultimately be used for subclonmg the cDNA; thus, they should be absent from the expression vector

to be employed Preferably, sites as close as possible to the target region are selected because this limits the amount of DNA sequencing required to ensure that unwanted PCR errors have not been introduced during the PCR amplifica- tion steps If two convenient unique restriction endonuclease sites are not present in the receptor cDNA, sites used for subcloning the construct into the vector may be employed Infrequently, the limited number of useful unique restriction sites located in the receptor cDNA may also be present in the expres- sion vector typically employed for transfection procedures, making subcloning via these sites inconvenient In these situations, it may be necessary to employ

a shuttle vector that lacks these restriction sites The cDNA subcloned into the shuttle vector is used as the template in the PCR reactions Following the PCR amplification, the mutated cDNA fragment is subcloned via the unique sites into the shuttle vector and then again subcloned into the expression vector of choice using additional convenient restriction sites

2.1.1 Oligonucleotides

Four oligonucleotide primers are required for the creation of point muta- tions, as will be described In terms of expense, it is important to realize that three of the four oligonucleotides may be repeatedly employed to construct subsequent mutations, as long as the receptor region of interest remains the same, i.e., the target sequence is flanked by the same two unique restriction sites selected initially Additionally, these oligonucleotides are useful as primers in the sequencing reactions that follow subcloning and permit the relatively rapid determination of the fidelity of the PCR amplification A schematic outline of the design of the oligonucleotides and the PCR reactions is provided in Fig 1, The first oligonucleotide (primer A) will code for the desired point mutation and should be designed to permit annealing to the coding or sense strand of the cDNA Typically, the oligonucleotide conststs of approx 5-8 bases 5’ of the point mutation to be made and -15-20 bases 3’ to the region of interest Figure 2

is an example of the design of an oligonucleotide specifying the mutation of a

histidine residue to alanine Design of this oligonucleotide and the others employed in this procedure should permit a theoretical melting temperature (T,,,) of at least 50°C, thereby lessening the occurrence of primers annealing to additional segments of the cDNA template and reducing the amplification of unwanted cDNA segments Oligonucleotides consisting of 18-20 bases with

Receptor-Ligand Binding 37

WILD-TYPE RECEPTOR SEQUENCE

5' - AGC TGG CTG CCT TTG CAC ATC CT - 3'

3' - TCG ACC GAC GGA AAC GTG TAG GA - 5'

SER TRY LEU PRO LEU HIS ILE

To replace histidine (CAC) with alanine (GCC):

5'- AGC TGG CTG CCT TTG a ATC CT - 3'

3'- TCG ACC GAC GGA AAC CGG ATC CT - 5'

The oligonucleotide selected for the PCR reaction is:

5'- TC CTA u CAA AGG CAG CCA GCT - 3'

Fig 2 Example of the design of an oligonucleotide coding for a point mutation (HISTIDINE TO ALANINE) Sequence specifying the mutation IS underlined

-60% G/C content should provide for this minimum T, The use of ohgo- nucleotides with segments of self-complementary sequence should be avoided, The second oligonucleotide (primer B) consists of approx 36 bases The

18 bases composing the 3’ end of this oligonucleotide should be complemen- tary to 18 bases of sequence of the antisense strand of the cDNA template upstream of restriction endonuclease site 1 A distance of W-100 bases between the oligonucleotide target sequence and restriction endonuclease site

1 is typically employed The 18 bases composing the 5’ end of primer B are of sequence unrelated to that of the DNA template This dummy sequence should

be of -60% G/C content and again have a T,.,, of at least 50°C Primer C is an oligonucleotide consisting solely of the 18 bases of dummy sequence contained

in primer B Primer D, designed to contain at least 18 bases, should anneal to the sense strand of the DNA template approx 5&100 bases downstream of restriction endonuclease site 2

lent results The amount of the polymerase, the buffer concentratton, and any other additional components, e.g., MgC12, should be employed as directed by the manufacturer

The PCR parameters typically employed for PCR reaction 1 are 95°C x I mm; 50°C x 1 min; 72OC x 1.5 min The program is run for 25 cycles and IS termt- nated with a single 72°C extension conducted for 10 min In almost all instances, we have found 25 cycles of amplification to provide sufficient quan- tities of product Approximately 20 & of the reaction mix is electrophoresed

on an agarose gel (l-l 5% depending on the size of the expected band) and the DNA fragment of interest (fragment 1) purified by standard techniques Our laboratory employs elution of the DNA fragment mto a well that has been cut mto the agarose gel directly below the band of interest For the elution proce- dure, the well contains O.lMammonmm acetate Other rsolation methods, such

as the commercially available Qiaex (Qiagen [Santa Clorita, CA]), also work well The isolated DNA is extracted once with isoamyl alcohol, ammonium acetate/ethanol-precipitated, and resuspended in 50 pL H20 This DNA frag- ment is then used as a primer in a single-cycle PCR extension step as follows: PCR reaction 2A: -100 ng cDNA template (typically in 10 JJL); 10 pL of purt- tied DNA fragment 1; 100 @4 dNTP mix; polymerase buffer; thermostable DNA polymerase; and HZ0 to final volume of 50 &

The PCR parameters are 95°C x 2 min; 45°C x 2 min; and 72°C x 10 min This program is run for a single cycle during which fragment 1 is extended with the receptor cDNA used as template Upon completion of the IO-min extension, the thermal cycler should be programmed to initiate a cycle consist- ing of the same parameters as those used in PCR reaction 1 During the first 95’C denaturation step, the additional components for the reaction (PCR reac- tion 2B) should be added The components are 50 pmol primer C; 50 pmol primer D; 100 w dNTP mix; DNA polymerase buffer; thermostable DNA polymerase; and HZ0 to a final volume of 50 &

Typically, these components are prepared during the lo-min extension of PCR reaction 2A and then introduced in a single 50-a addrtion directly under the mineral oil of the PCR reaction tube, which remains in the thermal cycler PCR reaction 2B is continued for 25 cycles, with a final 72’C extension step conducted for 10 min The product of these two PCR reactions is a DNA frag- ment (fragment 2) containing the desired point mutation, flanked by two unique restriction endonuclease sites, which permit subclonmg mto an expression vec- tor The inclusion of the dummy sequence in primers B and C ensures that only the DNA containing the desired mutation is amplified, and not wild-type receptor cDNA The location of the dummy sequence upstream of restriction endonuclease site 1 permits its removal during subcloning Approximately 10 pL

of the product of PCR reaction 2B may be electrophoresed on an agarose gel to

Receptor-Ligand Binding 33 determine if a fragment of the expected size has been amplified and to check for the presence of any additional bands The remainder of the product should

be extracted with phenokchloroform, ammonium acetate:ethanol-precipitated, and digested with the appropriate restriction endonucleases The fragment may then be ligated into the appropriate vector using standard molecular biological techniques E coli are transformed and plated, colonies grown up in media, and mimprep DNA obtained via standard methods (16) The presence of the desired mutation is determined by DNA sequencing All DNA derived from PCR amplification, i.e., that spanning restriction endonuclease sites 1 and 2 should be sequenced to ensure that unwanted mutations have not been intro- duced durmg the procedures Plasmid DNA may then be isolated on a large scale, e.g., Plasmid Maxi Kit (Qiagen), for transfection into cells of interest 2.2 Chimeric Receptors

As mentioned previously, creation of receptor chimeras via the PCR approach is similar to that described (Section 2.1.) for point mutations, with the requirement of an additional oligonucleotide and an additional PCR step

To illustrate the design of oligonucleotides and the PCR steps, an example will

be employed m which the third and fourth transmembrane domains (and con- necting sequence) of receptor A (acceptor) will be replaced with the analogous regions of receptor D (donor) A schematic outline of the design of the oligo- nucleotides and PCR reactions is provided m Fig 3 Initially, it is often helpful

to obtain a computer-generated alignment of the amino acid sequences of receptor A and receptor D for the determination of splicing junctions

2.2.1 Oligonucleo tides

As described for the creation of point mutations, the cDNA of receptor A must first be analyzed for the existence of two unique endonuclease sites that flank the region of interest, i.e., transmembrane domains 3-4 Again, these sites will ultimately be employed for subclonmg the engineered DNA frag- ment into receptor A Regarding restriction endonuclease sites and receptor D,

it is only a concern that the two sites selected for subcloning do not exist in the sequence of receptor D that is to be used in the substitution Two oligonucle- otide primers are required to amplify the region of receptor D selected for the substitution into a precise region of receptor A These two primers thus define the splicing junctions The first primer (primer A) consists of 36 bases, with

18 bases of the 3’ end designed to anneal to the antisense strand of receptor D These 18 bases should code for the initial sequence of receptor D that will be contained in the chimeric receptor, i.e., the first six amino acids of the third transmembrane domain The 18 bases at the 5’ end of primer A should be complementary to the antisense strand of receptor A in the region after which

sequence substitution will be initiated, i.e., the last six amino acids of the first extracellular loop Primer B consists of 36 bases, with the 18 bases of the 3’ end complementary to the sense strand of receptor D in the region at which substi- tution will be terminated, i.e., the last six amino acids of transmembrane domain 4 The 18 bases composing the 5’ end of primer B should be comple- mentary to the sense strand of receptor A in the region at which receptor A sequence will be reinitiated within the chimeric receptor, i.e., the first six amino acids of the second extracellular loop For both primer A and primer

B, the 18 bases originating from receptor A and receptor D sequence should always provide for a T, of at least 50°C If required to obtain this minimum T,,,, the number of bases may be increased

The remaining three oligonucleotide primers required for construction of the chimeric receptor are analogous to those employed in the creation of receptor point mutations Primer C is composed at the 3’ end of 18 bases complementary to the antisense strand of receptor A upstream of restric- tion endonuclease site 1 and at the 5’ end of 18 bases of dummy sequence Primer D is composed solely of the 18 bases of dummy sequence Primer E

is complementary to the sense strand of receptor A downstream of restriction endonuclease site 2

to 100 pL final volume

The PCR parameters are the same as those employed in PCR reaction 1 in the creation of point mutations After the 25 cycles of PCR, approx 20 @ of the product may be electrophoresed on an agarose gel and the fragment of interest (fragment 1) purified The DNA is extracted once with isoamyl alco- hol, ammonium acetate:ethanol-precipitated, and resuspended in 50 pL of H20 Fragment 1 is composed primarily of the sequence of receptor D, with sequence

of receptor A at both ends, which should permit annealing to the receptor A cDNA template in the succeeding steps Fragment 1 is then used as a primer in the second PCR reaction (PCR reactton 2A): -100 ng receptor A cDNA; 10 & fragment 1; 100 pM dNTP mix; DNA polymerase buffer; thermostable DNA polymerase; and Hz0 to a final volume of 50 pL

The PCR parameters are 95°C x 2 min; 45OC x 2 min; and 72°C x 10 mm This program is run for one cycle only Upon completion of the 10 min exten- sion, the thermal cycler should be programmed to initiate a cycle consisting of the same parameters used in PCR step 1

During the initial 95OC denaturation, additional components are added (PCR reaction 2B): 50 pmol primer B; 50 pmol primer C; 100 @4 dNTP mix; DNA polymerase buffer; thermostable DNA polymerase; and Hz0 to a final volume of 50 pL

As in the constructron of the receptors possessing point mutations, these components are introduced in a single 50-pL addition directly to the PCR reaction tube beneath the mineral oil PCR reaction 2B is run for 25 cycles, with a final lo-min extension Again, an aliquot of the reaction mix is run on

an agarose gel and the fragment of interest (fragment 2) is purified, extracted with isoamyl alcohol, precipitated, and resuspended in 50 pL, of H20 Frag- ment 2 is then used as a primer in the initial part of PCR reaction 3A: - 100

ng receptor A cDNA; 10 pL fragment 2; 100 @I dNTP mix; DNA poly- merase buffer; thermostable DNA polymerase; and Hz0 to a final volume

of 50 pL

The PCR parameters are identical to those used in PCR reaction 2A Upon completion of the extension step, additional components are then added to the reaction tube in a final volume of 50 pL (PCR reaction 3B): 50 pmol primer D;

50 pmol primer E; 100 @! dNTP mix; DNA polymerase buffer; thermostable DNA polymerase; and H,O to a final volume of 50 &

The PCR parameters are the same as those used for PCR reaction 2B (25 cycles) Ten microliters of the product may be run on an agarose gel to determine if a fragment of the expected size has been amplified and to ascer- tain the presence of any additional contaminating bands The remainder of the reaction may be phenol:chloroform-extracted, precipitated, and digested with the appropriate restriction endonucleases for subcloning The remaining pro- cedures for preparation of the construct are exactly as described (Section 2.1.2.) for the creation of pomt mutations For screening purposes, it is often conve- merit and time-saving to perform a restriction endonuclease digestion of miniprep DNA to determine the presence of the chimeric receptor substitutton Frequently, the substitution of DNA sequence results m the addition or re- moval of a restriction site, which is easily recognized relative to the wild-type receptor cDNA Once identified, the isolated plasmid DNA may then be sub- jected to sequencing

2.2.3 Additional Notes

The following are factors that should be considered during the construction

of chimeric receptors as outlined (Section 2.2.) Regarding primer B, it was noted that the sequence composmg the 3’ end of this ohgonucleotide should anneal to the sense strand of receptor D The ability of this sequence to anneal

to the analogous region of receptor A should also be determined If this sequence of primer B is homologous to that of receptor A, we have observed m PCR reaction 2B the unwanted amplification of wild-type receptor A template rather than the desired amphfication of chimeric fragment 1 sequence The greater the sequence homology displayed by receptor A and receptor D in the area near the splicing junction, the more likely this possibility becomes Thus,

if necessary in the design of primer B, the sequence from receptor D selected

to compose the 3’ end of the oligonucleotide should be extended until the possibility of annealing to receptor A has been greatly reduced Increasing the programmed annealing temperature m PCR reaction 2B may also be beneficial

We have used the method outlined (Section 2.2.) for the construction of chimeric receptors containing sequence substitution of as few as 20 amino acids Accounting for the sequence of the acceptor receptor in primer A and primer B, this requires the amplificatton and puriticatton of an approx lOO-bp DNA fragment For amplification of this limited amount of sequence m PCR reaction 1, we decrease the dNTP concentration to 100 @4 and the amount of primer A and primer B to 25 pmol For the sequence substitutton of smaller receptor regions, e.g., 10 or fewer amino acids, we have successfully employed the protocol described previously for construction of point mutations In these situations, a single oligonucleotide may be used to code for the multiple base changes A single oligonucleotide may also be used to delete a limited amount

of nucleotide sequence if required For this purpose, the primer should contain

at least 15 bases complementary to the cDNA template on either side of the omitted sequence

3 Expression of Mutated Receptor Constructs

Following its construction, the binding properties of a mutated receptor, either a point mutation or chimeric receptor, may be studied by subcloning the cDNA into an expression vector and transfecting the construct into a suitable cell line Many expression vectors are available and suitable for this procedure

In our laboratory, both pBC (17) and pCMV5 (18) have been employed It is desirable to select an expression vector and cell line for transfection that per- mit for relatively high-level expression of the constructs This is particularly beneficial for certain mutated receptors that may be much more poorly expres- sed than weld-type receptors A system that permits only low levels of wild- type receptor expression may not be useful for analysis of mutated receptors that are present at several-fold lower levels

Perhaps the most frequently used system for the study of the ligand-binding properties of mutated receptors are COS cells that have been transiently trans- fected with the receptor cDNA-expression vector construct COS cells permit for the replication of DNA vectors that contain the SV40 origin of DNA repli- cation (16) In our laboratory, COS-7 cells are transiently transfected via a DEAE-dextran transfection procedure The DEAE-dextran transfection proce- dure is that commonly used (161, with minor modifications All solutions should be filter-sterilized prior to use

Day 1: COS-7 cells at approx 90% confluence in a T75 flask are washed one time with 10 mL of PBS and 5 mL of a PBS solution containing both 2.5 mg DEAE-dextran and the appropriate amount of DNA (see nextpage) is added to the flask The flask is returned to the 37°C cell culture incubator for 30 min At this time, 20 mL of DMEM/lO% FBS containing 100 @4chloroquine is added

to the flask, which is then incubated at 37°C for 2.5 h At this time, the media is removed and replaced with 5 mL of 10% DMSO in DMEM/lO% FBS After an incubation of 2.5 min at room temperature, the media is removed and a fresh

25 mL of DMEM/I 0% FBS is added

Day 2: Media is removed and a fresh 25 mL of DMEM/lO% FBS is again added The cells may be prepared for ligand-binding analysis on either d 3 or 4 Advantages of employing the DEAE-dextran transient transfection protocol include the relatively short amount of time required between obtaining the mutated receptor construct and determination of its ligand-binding properties Additionally, through titration of the amount of receptor DNA-expression vec- tor construct used in the protocol, it is possible to regulate to some degree the level of receptor expression For example, transfection of cells with 5-l 0 pg of

wild-type receptor DNA may result in membrane expression of 1 pmol/mg receptor protein A similar amount of a mutated receptor that is expressed less efficiently may perhaps be obtained if the amount of receptor cDNA used in the transfection is increased to 25 or 30 g Since with certain mutated recep- tors there may be a plateau at which no increased receptor expression is obtained, even with increasing amounts of transfected cDNA, it may be neces- sary to titrate down the level of expression of wild-type receptors or of those mutants that may express more efficiently than wild-type receptor Such a regu- lation of receptor levels may be particularly desirable when studying agonist ligand binding in which the amount of expressed receptor, and hence the sto- ichiometry of receptor to G proteins in the cell or membrane preparation, may theoretically influence agonist affinity Generally, the quantities of expressed receptor obtained with transiently transfected COS-7 cells are greater than those present in native tissues and are more than adequate for receptor-ligand bmding studies via saturation binding analysis or competition bmding assays Radioligand-binding analysis of mutated receptors heterologously expressed

in mammalian cells may be performed via standard methods Conditions for the assay, such as the use of intact cells vs membrane preparations, incubation temperature, and time, are typically predetermined using the wild-type receptor and are dependent on factors such as ligands available, cell type employed, and factors intrinsic to the receptor under analysis

In certain instances (see Section 4.), analysis of the functional or biochemi- cal activity of a mutated receptor may be required to assess its ligand-binding properties For example, the potency of an agonist ligand in eliciting a bio- chemical response may be used as an indicator of agonist affinity In these situations, the cell type to be used, as well as the transfection procedure (i.e., transient vs stable transfection), is governed by the known signaling properties

of the receptor being examined For example, transient transfection of mutated receptors in COS cells may be used for receptors that couple to phospholipase

C activation via the G, class of G proteins; stable transfection of receptor con- structs may be required to demonstrate the receptor-mediated inhibition of adenylyl cyclase occurring via activation of a member of the G, family Stable expression of

a receptor may also be desired for functional studies because this permits a source

of cells in which the receptor population IS constant over time In transient trans- fections, the expression of receptors may vary from day to day, with disparate expression among cells obtained from a single transfection

4 Analysis and Interpretation of Results

This chapter primarily focuses on the techniques used in receptor mutagen- esis It may, however, be appropriate to conclude with a discussion of factors pertaining to the analysis of data obtained from ligand-binding studies with

Receptor-ligand Binding 39

genetically engineered receptors The goal of these studies is to identify receptor regions or perhaps specific amino acids that are involved in Iigand recognition by the receptor When a mutation is found to alter ligand bind- ing relative to the wild-type receptor, the most pertinent question is whether the effect is direct or indirect The observed response may occur as the result of a substitution of an amino acid(s) that directly binds the ligand via

a chemical interaction such as hydrophobic or hydrogen bonding Alterna- tively, the targeted amino acid may not actually interact with the ligand, but rather the substitution may indirectly affect ligand recognition by altering receptor conformation or architecture At present, these possibilities may not

be differentiated with absolute certainty No single experiment permits the unequivocal assignment of the role of a particular receptor region or amino acid in ligand binding However, the design of the initial mutagenesis experi- ments and the completion of complementary studies may be very valuable in obtaining the most valid insights from the data generated in the mutational analysis of ligand binding

In the study of receptors possessing point mutations and chimeric receptors,

an initial consideration is the specificity of the effect of the sequence substitu- tion For example, disruption of agonist binding while antagonist binding is unaltered is an indication that the mutation may be specifically affecting an agonist-binding domain and is not perturbing overall receptor conformation Observation of the loss of both agonist and antagonist radioligand binding is more difficult to interpret It is not unreasonable to assume that certain amino acids of a receptor may be involved in the coordination of both classes of ligands Thus, a single mutation may perhaps directly perturb the binding of both classes of ligands However, it is also possible that the mutation dis- rupts receptor processing or folding and is preventing membrane insertion

of the receptor In such a situation, it may be possible to examine the activ- ity of an agonist in a functional assay that permits the use of much higher ligand concentrations than those employed in radioligand-bmdmg assays Additionally, many investigators have used immunologic techniques to demonstrate membrane expression of constructs that do not display ligand binding Presumably, the alteration in the ligand-binding pocket produced

by the mutation does not prevent recognition by the appropriate antibody Demonstration of cell surface expression of a mutated receptor, however, does not preclude a disruption of receptor architecture that is nonspecifically alter- ing ligand binding

In the design of genetically engineered receptors, particularly chimeric receptors, it is often advantageous to demonstrate a gain of function as well as

a loss of ligand binding upon sequence substitution For example, when study- ing receptor subtypes with substantially different binding profiles, it may be

possible that a limited amount of sequence replacement in a chimeric receptor may significantly enhance its affinity for a particular class of ligands relative

to the parent wild-type receptor from which the majority of its structure is derived Relative to a loss of binding, such an enhancement may less likely result from an indirect effect on receptor structure In this regard, the demon- stration of reciprocal effects with reciprocal sequence replacements is an addi- tional indication of direct effects of the mutation

In the study of receptors possessing point mutations, additional information regarding the precise role of a specific amino acid may at times be obtained if

a series of amino acids is used to perform the substitution As noted previously,

in initial mutagenesis studies designed to determine if a particular amino acid

is involved in ligand binding, the target residue is frequently replaced with alanine The rationale for this selection is that the physical properties of ala- nine make it less likely than other residues to nonspecifically disrupt protein structure (particularly in transmembrane domains), and this residue has little propensity for forming chemical bonds with the ligand A loss of binding upon alanine substitution is an initial indication of a role for the target residue in ligand binding Additionally, many investigators have used the change in free energy of the binding interaction, i.e., the relationship of d(dG) = -RT ln(K,[mutant]/K,[wild-type], as an indication of the nature of the type of bond lost upon making a point mutation (19-21) Subsequent experiments may attempt to determine if the presence of the wild-type residue may be mimicked

by replacement with an amino acid possessing the ability to provide a similar interaction with the hgand Examples of ammo acids that have been found to functionally substitute for one another to varying degrees in mutagenesis stud- ies of hgand binding to GPCRs include asparagine-glutamine (22), serine- threonine (9,23), histidme-glutamine (24), arginine-lysine (25,261, and glutamate-aspartate (7) Obviously, such relationships may not be found in all receptors because several factors, such as differences in overall receptor archi- tecture, slight differences in amino acid bulk, and variation in their propensity for bond formation, must be considered

The specificity of the mutation may be further explored by studying the binding of ligands in which key functional groups have been replaced or deleted

at both the wild-type and mutated receptor This analysis also may indicate with which functional group of the ligand the targeted amino acid interacts Theoretically, a ligand in which a crucial functional group has been removed will demonstrate similar affinity at a wild-type receptor and a point mutant of this receptor in which the amino acid specifically interacting with the particu- lar functional group has been replaced At such mutants, there should be no further reduction in affinity for modified ligands relative to the wild-type receptor, since the binding interaction does not occur in either situation This