To isolate thermostabilizing mutations by systematic scanning mutagenesis, point mutations are
first introduced one by one at every position throughout the target receptor by substituting non-alanine
residues with alanine ones and alanine residues with leucines. The mutated receptors are expressed in
DH5α E. coli or in mammalian cells and membrane fragments collected, solubilized in DDM, and
supplemented or not with a radioactive ligand. Thermostability is assessed by incubating the samples
at the desired temperature for 30 min. Samples are then placed on ice and radioligands added if not
already present and equilibrated. Receptor-bound and free radioligand are separated by gel filtration
(for details on the procedure, see Magnani et al. 2008). Alanine/leucine-scanning mutagenesis was
initially performed on three GPCRs, the β 1 adrenergic and A 2A adenosine receptors (Serrano-Vega
et al. 2008; Lebon et al. 2011a, b; Miller and Tate 2011) and a neurotensin receptor (NTS1) (Shibata
et al. 2009). The thermostability of each mutant was determined using either radiolabeled inverse
agonists (β 1 and A 2A receptors) or radiolabeled agonists (A 2A and NTS1 receptors). In each case,
approximately 5–9% of the mutations tested were thermostabilizing, of which 60–90% were found in
the TM region rather than in the loops. Each thermostabilizing point mutation typically improved the
thermostability by only 1–3
C, but within each scan there were generally one or two significantly
better mutations. This first round of mutations permits to identify positions at which stabilizing
mutations can be introduced. Further improvement can be obtained by substitution with residues
other than alanines. Once individual thermostabilizing mutations have been identified, a selection of
them are combined to generate an even more stable receptor. At the end of the day, stabilization,
measured as the extension of lifetime in DDM as compared to wild type, was 125Â for the β 1 receptor
(six point mutations) and 38Â (eight mutations) or 260Â (four mutations) for two different constructs
of the A 2A receptor. These two receptors were crystallized and yielded structures at resolutions varying
between 2.3 and 3.4 Å, depending on the receptor and on the ligand used for co-crystallization
(reviewed in Tate 2012).
Two points worth noting are (i) that the mutations lie outside the ligand-binding site and (ii) that,
when the binding of various types of ligands to thermostabilized receptors is tested, it is found that
there is little change of affinity for the ligand used during the selection, e.g. an inverse agonist, and a
drop of affinity for ligands with the opposite pharmacology, e.g. an agonist. What the selection
procedure achieves, therefore, is to stabilize one of the conformations the receptor can adopt by
preference to the others, hence the term of “conformational thermostabilization.” Another remarkable
point is that, even though the screening process takes place in DDM, the resulting constructs are also
more resistant to the short-chain detergents used for crystallization (see Lebon et al. 2011a, b).
The reasons why thermostabilized GPCRs are more stable than the wild type are complex and
probably multifarious. In a recent review, analysis of MD data led to the conclusion that receptors are
stabilized through a combination of factors including an increase in rigidity, a corresponding decrease
in collective motions, reduced stress at specific residues, and the presence of ordered water molecules
(Vaidehi et al. 2016; see also Lee et al. 2015). Given that the selection is carried out in the presence of
lipids and DDM, it is probably difficult to exclude that reinforcement of their affinity for the surface of
the native receptor may also play a role. An interesting open question is whether stabilization by
mutations selected in the presence of detergent and stabilization by APols are or not additive.
2.5.3
What About Keeping Membrane Proteins Water-Soluble Without
Using Detergents?
In the previous section, we have reviewed some attempts at increasing the stability of detergentsolubilized MPs by improving either the detergent or the protein. Can one, instead, just dispense with
using any detergent? This is of course achieved when transferring the detergent-extracted protein to
2.5 Solutions to the Instability Problem
87
first introduced one by one at every position throughout the target receptor by substituting non-alanine
residues with alanine ones and alanine residues with leucines. The mutated receptors are expressed in
DH5α E. coli or in mammalian cells and membrane fragments collected, solubilized in DDM, and
supplemented or not with a radioactive ligand. Thermostability is assessed by incubating the samples
at the desired temperature for 30 min. Samples are then placed on ice and radioligands added if not
already present and equilibrated. Receptor-bound and free radioligand are separated by gel filtration
(for details on the procedure, see Magnani et al. 2008). Alanine/leucine-scanning mutagenesis was
initially performed on three GPCRs, the β 1 adrenergic and A 2A adenosine receptors (Serrano-Vega
et al. 2008; Lebon et al. 2011a, b; Miller and Tate 2011) and a neurotensin receptor (NTS1) (Shibata
et al. 2009). The thermostability of each mutant was determined using either radiolabeled inverse
agonists (β 1 and A 2A receptors) or radiolabeled agonists (A 2A and NTS1 receptors). In each case,
approximately 5–9% of the mutations tested were thermostabilizing, of which 60–90% were found in
the TM region rather than in the loops. Each thermostabilizing point mutation typically improved the
thermostability by only 1–3
C, but within each scan there were generally one or two significantly
better mutations. This first round of mutations permits to identify positions at which stabilizing
mutations can be introduced. Further improvement can be obtained by substitution with residues
other than alanines. Once individual thermostabilizing mutations have been identified, a selection of
them are combined to generate an even more stable receptor. At the end of the day, stabilization,
measured as the extension of lifetime in DDM as compared to wild type, was 125Â for the β 1 receptor
(six point mutations) and 38Â (eight mutations) or 260Â (four mutations) for two different constructs
of the A 2A receptor. These two receptors were crystallized and yielded structures at resolutions varying
between 2.3 and 3.4 Å, depending on the receptor and on the ligand used for co-crystallization
(reviewed in Tate 2012).
Two points worth noting are (i) that the mutations lie outside the ligand-binding site and (ii) that,
when the binding of various types of ligands to thermostabilized receptors is tested, it is found that
there is little change of affinity for the ligand used during the selection, e.g. an inverse agonist, and a
drop of affinity for ligands with the opposite pharmacology, e.g. an agonist. What the selection
procedure achieves, therefore, is to stabilize one of the conformations the receptor can adopt by
preference to the others, hence the term of “conformational thermostabilization.” Another remarkable
point is that, even though the screening process takes place in DDM, the resulting constructs are also
more resistant to the short-chain detergents used for crystallization (see Lebon et al. 2011a, b).
The reasons why thermostabilized GPCRs are more stable than the wild type are complex and
probably multifarious. In a recent review, analysis of MD data led to the conclusion that receptors are
stabilized through a combination of factors including an increase in rigidity, a corresponding decrease
in collective motions, reduced stress at specific residues, and the presence of ordered water molecules
(Vaidehi et al. 2016; see also Lee et al. 2015). Given that the selection is carried out in the presence of
lipids and DDM, it is probably difficult to exclude that reinforcement of their affinity for the surface of
the native receptor may also play a role. An interesting open question is whether stabilization by
mutations selected in the presence of detergent and stabilization by APols are or not additive.
2.5.3
What About Keeping Membrane Proteins Water-Soluble Without
Using Detergents?
In the previous section, we have reviewed some attempts at increasing the stability of detergentsolubilized MPs by improving either the detergent or the protein. Can one, instead, just dispense with
using any detergent? This is of course achieved when transferring the detergent-extracted protein to
2.5 Solutions to the Instability Problem
87