Intramolecular allosteric communication in dopamineD2 receptor revealed by evolutionary aminoacid covariationYun-Min Sunga, Angela D. Wilkinsb, Gustavo J. Rodrigueza, Theodore G. Wensela,1, and Olivier Lichtargea,b,1

aVerna and Marrs Mclean Department of Biochemistry and Molecular Biology, Baylor College of Medicine, Houston, TX 77030; and bDepartment ofMolecular and Human Genetics, Baylor College of Medicine, Houston, TX 77030

Edited by Brian K. Kobilka, Stanford University School of Medicine, Stanford, CA, and approved February 16, 2016 (received for review August 19, 2015)

The structural basis of allosteric signaling in G protein-coupledreceptors (GPCRs) is important in guiding design of therapeuticsand understanding phenotypic consequences of genetic variation.The Evolutionary Trace (ET) algorithm previously proved effective inredesigning receptors to mimic the ligand specificities of functionallydistinct homologs. We now expand ET to consider mutual informa-tion, with validation in GPCR structure and dopamine D2 receptor(D2R) function. The new algorithm, called ET-MIp, identifies evolu-tionarily relevant patterns of amino acid covariations. The improvedpredictions of structural proximity and D2R mutagenesis demon-strate that ET-MIp predicts functional interactions between residuepairs, particularly potency and efficacy of activation by dopamine.Remarkably, although most of the residue pairs chosen for mutagen-esis are neither in the binding pocket nor in contact with each other,many exhibited functional interactions, implying at-a-distance cou-pling. The functional interaction between the coupled pairs corre-lated best with the evolutionary coupling potential derived fromdopamine receptor sequences rather than with broader sets of GPCRsequences. These data suggest that the allosteric communication re-sponsible for dopamine responses is resolved by ET-MIp and bestdiscerned within a short evolutionary distance. Most double mutantsrestored dopamine response to wild-type levels, also suggesting thattight regulation of the response to dopamine drove the coevolutionand intramolecular communications between coupled residues. Ourapproach provides a general tool to identify evolutionary covariationpatterns in small sets of close sequence homologs and to translatethem into functional linkages between residues.

allostery | G protein-coupled receptors | residue covariation |Evolutionary Trace

Identifying residues that coevolved to maintain or acquire fit-ness properties is critical for understanding protein structure,

function, and evolution (1). Previous studies have shown thatcovarying residue pairs, those that exhibit correlated amino acidchanges in large multiple sequence alignments, tend to formstructural contacts (2–7), enhancing predictions of protein 3Dstructures (8–11). Covariation can also involve distal residues,but the function of these at-a-distance couplings is elusive andhas been attributed to background noise, alternative proteinconformations, or subunit interactions of protein homooligomers(5, 7, 12). Alternately, distal covarying residue pairs could in-dicate allosteric couplings (6, 13–18).The possibility of capturing intramolecular allosteric communi-

cation by amino acid covariation analysis of protein family se-quences has not been extensively explored. Nonproximalthermodynamic coupling between correlated residue pairs wasnoted in 274 PDZ domains (14), but the relationship to allo-stery is still debated (19, 20). It may be that distinctive allostericmechanisms, even among close homologs, limit the extractionof allosteric couplings from sequences (13). Our previous identi-fication of residues important for allosteric signaling within Gprotein-coupled receptors (GPCRs) using Evolutionary Trace (ET)(21–24) and strong conservation of some of the residues implicated

led us to ask whether ET could also uncover couplings amongprotein sequence positions not in direct contact.ET estimates the relative functional sensitivity of a protein to

variations at each residue position using phylogenetic distances toaccount for the functional divergence among sequence homologs(25, 26). Similar ideas can be applied to pairs of sequence positionsto recompute ET as the average importance of the couplings be-tween a residue and its direct structural neighbors (27). To measurethe evolutionary coupling information between residue pairs, wepresent a new algorithm, ET-MIp, that integrates the mutual in-formation metric (MIp) (5) to the ET framework. We used dopa-mine D2 receptor (D2R), a target of drugs for neurological andpsychiatric diseases (28), to test whether ET-MIp could elucidate theallosteric functional communications from amino acid covariationpatterns and resolve the evolutionary distance at which the allostericpathways of D2R homologs are sufficiently conserved to detect res-idue−residue coupling signatures. D2R is expressed in the centralnervous system and responds to dopamine, the major catecholamineneurotransmitter. Canonical D2R signaling is effected by Gi/o classG proteins, which regulate ion channels (29, 30), MAPK kinases(31), phospholipase C (32), and inhibition of adenylyl cyclase (33).D1 class receptors (D1R and D5R) have lower affinities fordopamine (34–36) and activate adenylyl cyclase through Gs class Gproteins. To characterize allosteric communication between

Significance

Characterizing relationships among protein structure, function,and evolution requires understanding the evolutionary con-straints on each constituent residue of a protein. Previousstudies have shown that structural information can be re-trieved from evolutionary residue covariation in protein fami-lies. However, whether the evolutionary history in proteinsequences informs on functional interactions between non-adjacent residues has been unclear. Here, we developed amethod that uses evolutionary amino acid covariation to inferfunctionally coupled residue pairs in the dopamine D2 receptor.We discovered functional coupling between residue pairs thathave coevolved mainly to control responses to dopamine andmaintain them at wild-type levels. Our findings demonstrate thepossibility of extracting the networks of intramolecular allostericcommunication from evolutionary residue covariation patterns.

Author contributions: Y.S., A.D.W., T.G.W., and O.L. designed research; Y.S. and A.D.W.performed research; Y.S., A.D.W., and G.J.R. contributed new reagents/analytic tools; Y.S.,A.D.W., T.G.W., and O.L. analyzed data; and Y.S., A.D.W., T.G.W., and O.L. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1To whom correspondence may be addressed. Email: lichtarge@bcm.edu or twensel@bcm.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1516579113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1516579113 PNAS | March 29, 2016 | vol. 113 | no. 13 | 3539–3544

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 28

, 202

0

covarying pairs of residues ranked as important by ET (ET residuepairs), we examined functional coupling for ligand binding affinitiesand downstream Gi activation induced by agonist-stimulated D2R.

ResultsET-MIp Identified Pairs of Residue Positions with EvolutionaryCovariation Patterns. We hypothesized that accounting for spe-cies divergence would improve the detection of functionallycoupled residues over covariation analyses that ignore phyloge-netic information. ET-MIp adds the mutual information metric(5) to ET to keep track of the phylogenetic distance at which apair of residues vary (Fig. 1A; see Materials and Methods andSupporting Information for details). In ∼2,500 Class A GPCRtransmembrane (TM) domain sequences, we found that residuepairs with high ET-MIp scores were more enriched for directcontacts in a reference structure (PDB 2RH1) compared withresults obtained with leading alternative methods (5, 37, 38) (Fig.1B), showing that GPCR phylogenetic information improves thecoupling signal. Preliminary analysis of other protein familiessuggests that this result may be fairly general. This opens thepossibility that ET-MIp also detects functionally relevant co-variation among structurally distant yet coupled residue pairs.

Residue and Sequence Selection. To test predicted couplings ex-perimentally, we selected 10 covarying ET residue pairs in D2Rin which one or both of the residues were involved in allostericpathways of D2R ligand responses (23), plus two more pairs.Most of these are predicted to be functionally important by ETand alter function upon mutation (23). These pairs cover a rangeof ET-MIp coupling scores and involve structurally distant resi-dues (except T205M5.54/L379F6.41; see Table S1) whose cou-plings, if any, would be allosteric. To probe the role of functionalcoupling in discriminating between dopamine and serotoninduring evolution, ET residues in D2R were mutated to thecorresponding residues in the closely related 5-HT2A serotoninreceptor (5-HT2AR), so only sites at which 5-HT2AR and D2Rdiffer were included. As single mutations, these substitutions stillallow for a functional receptor (23). In addition to mutationselection, we considered the choice of sequences used for cal-culating scores, because this can strongly impact ET-MIp cou-plings. For structural contacts, ET-MIp can be applied to Class AGPCR sequences because they are all structurally similar. For

functional allostery in D2R, tuned to a specific ligand and sig-naling bias, a more restricted alignment may be best. Accord-ingly, multiple alignments were tested (Class A, bioamine,dopamine, and D2Rs) and yielded distinct coupling scores(Table S2).

Functional Interactions Between Covarying ET Residue Pairs MaintainedDopamine Response at WT Level. To test whether the selected ETresidue pairs were functionally coupled, we first compared theeffects of single and double mutations on dopamine efficacy usinga fluorescence-based assay to study Gi activation induced byagonist-stimulated D2R (Fig. 2; see Materials and Methods andSupporting Information for details). For five pairs (V83L2.53/V91S2.61, M117F3.35/Y199F5.48, I48T1.46/F110W3.28, V152A4.42/L171P4.61, and N124H3.42/T205M5.54), activation of Gi in re-sponse to dopamine was unexpectedly decreased or restored to anear-wild-type (WT) level in the double mutants, even thoughone or both of the constituent single mutants showed signifi-cantly enhanced response (Fig. 2 A and C). These results in-dicate that covarying ET residue pairs help maintain dopamineresponses at the WT level. A trend was that functional couplingwas more apparent in pairs with high evolutionary coupling po-tential when calculated using sequences only from dopaminereceptors (Fig. 2C). For example, the loss-of-function mutationL379F6.41 is rescued by T205M5.54, with which it has a strongevolutionary coupling potential, but not by N124H3.42, with whichthe coupling potential is weaker (Fig. 2 B and C). To estimate theepistatic effect of double mutations, we used four standard models(product, logarithmic, minimal, and additive interaction models)(39, 40). Except for the minimal model, the five residue pairs thatwere functionally coupled to maintain WT dopamine responsesyielded higher epistasis scores, and the epistasis scores correlatedbetter with the evolutionary coupling scores when calculated usinginput sequences made up only of dopamine receptors (Fig. 2C andFigs. S1–S4). This suggests that the allosteric communication re-sponsible for dopamine response is a unique evolutionary signa-ture of dopamine receptors. Compared with direct-couplinganalysis (DCA), Frobenius norm (FN), and direct information(DI) algorithms (37) (Tables S3 and S4), these ET-MIp resultscorrelated better with experimental epistasis scores (Table S5).To assess the effect of functional coupling on dopamine po-

tency, dose–response curves were generated to derive EC50 val-ues for WT and mutant D2Rs (Fig. 2 D and E and Table S6). Weused relative ln(EC50) values, defined as the difference betweenln(EC50 mutant) and ln(EC50 WT), to approximate free energychanges, because potency reflects differences in ligand bindingaffinity and/or activation kinetics, either of which must be de-termined by a free energy term. To assess nonadditivity, wecompared the sum of relative ln(EC50) values of the single mu-tants with the relative ln(EC50) value of the correspondingdouble mutants (Fig. 2F), except when mutants showed almostno response to dopamine. Unlike the three residue pairs with thelowest evolutionary coupling potential, the six pairs with thehighest evolutionary coupling potential were nonadditive—afunctional coupling that presumably fine-tunes the sensitivity todopamine (Fig. 2F). Here again, deviation from additivity cor-related most highly with the evolutionary coupling potentialderived from the dopamine receptor sequence set (Fig. 2G andFig. S5), and, overall, ET-MIp correlated better with non-additivity than DCA–FN and DCA–DI (Table S7).

Rescue Effect on Serotonin Activation Was Observed for Some D2RDouble Mutants. Some of the studied ET residues were involvedin discriminating against G16 activation induced by serotoninstimulation of D2R (23). To test whether the covarying ETresidue pairs are functionally coupled with respect to regulatingagonist specificity, we compared the effects of single and doublemutations on serotonin responses. The single mutants I48T1.46,

48-110 205-379 T L T L M V M MM IM IM MM MM MM F

I F I F V L T Y T Y T W T W T W A F T W

DRD2DRD3DRD4ADA2CADA2BADA2ADRD1DRD55HT1A5HT2A

205 379

48

110

False positive rate

True

pos

itive

rate

1.0

0.8

0.6

0.4

0.2

0.00.2 0.4 0.6 0.8 1.0

nMIMIpDCA-DIDCA-FNET-MIp

0.0

Fig. 1. ET-MIp decodes evolutionary correlations between residue posi-tions. (A) Examples of residue positions displaying covariation patterns in thecontext of evolutionary trees. For simplicity, only residues from human se-quences are shown. Covarying residue pairs were mapped onto the D3Rstructure (PDB 3PBL). Some have structural contacts (blue spheres), whereasothers are distant in the structure (red spheres). These positions are hy-pothesized to be functionally coupled during evolution. (B) Receiver oper-ating characteristic curves comparing the performance of ET-MIp and othermutual information (MI)-based methods, MIp (5), normalized MI (nMI) (38),and DCA algorithms, FN and DI (37), in identifying residues in contact (within6 Å) in the structure of the β2 adrenergic receptor (PDB 2RH1). Each methodwas applied to aligned sequences of ∼2,500 Class A GPCRs. The areas underthe curves, which indicate the accuracy of the prediction, are 0.75 for ET-MIp, 0.68 for DCA–FN, 0.65 for DCA–DI, 0.65 for MIp, and 0.56 for nMI.

3540 | www.pnas.org/cgi/doi/10.1073/pnas.1516579113 Sung et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 28

, 202

0

Y213I5.62, T205M5.54, L387C6.49, and I105K3.23 had enhancedresponses to serotonin, indicating that these positions in D2Rparticipate in discriminating against Gi activation induced byserotonin (Fig. 3C). A rescue effect was observed at I48T1.46/F110W3.28 and N124H3.42/T205M5.54, even though F110W3.28

and N124H3.42 alone abolished activation by serotonin, sug-gesting functional coupling in controlling the specificity for D2Ractivation in these cases (Fig. 3 A−C). Given the small number offunctional coupling cases found, we infer that discriminatingagainst serotonin is not the main functional role of the covaryingET residue pairs studied here. Moreover, this finding suggeststhat the allosteric communication responsible for specificity ofreceptor activation may vary across a protein family, makingit difficult for ET-MIp to extract such a pattern from a broadsequence input.

Nonadditivity of Free Energy Changes upon Ligand Binding WasObserved at Some D2R Double Mutants. To investigate whethercovarying ET residue pairs interact to control ligand affinity, spi-perone competition binding experiments determined Ki for bothdopamine and serotonin and measured the energetic perturbationsby calculating the Gibbs free energy change (ΔΔG0) (Table S8). Incontrast to EC50, these measurements on whole cell membranesreflect low-affinity non-G protein-coupled binding. Significant dif-ferences between (ΔΔG0

A + ΔΔG0B) and ΔΔG0

AB observed forF202L5.51/Y213I5.62 and T205M5.54/L379F6.41 revealed nonadditiveeffects on both dopamine (Fig. 4 A and C) and serotonin (Fig. 4B and D) binding, indicating functional coupling in regulatingreceptor−ligand binding affinity. Strikingly, F202L5.51/Y213I5.62

and T205M5.54/L379F6.41 are outside the ligand binding pocket,and yet alter receptor−ligand interaction (Fig. 2H), indicating

D2R (WT)V152AL171PV152A/L171P

Act

ivat

ion

by d

opam

ine

Time (s)

A

Time (s)

Act

ivat

ion

by d

opam

ine

T205M/L379FN124H/L379FL379F

T205MN124HD2R(WT)B

high low

Evolutionary Coupling PotentialC

I105K3.23

L171P4.61

V191L5.40

S193G5.42

I195F5.44

Y199F5.48

L387C6.49

F202L5.51

T205M5.54

V152A4.42

Y213I5.62

S409N7.36

V91S2.61

F110W3.28

V83L2.53

M117F3.35

C385M6.47

I48T1.46

N124H3.42

L379F6.41

H

-13 -12 -11 -10 -9 -8 -7 -6 -5 -4 -30

20

40

60

80

100

120 D2R (WT)F202LY213IF202L/Y213I

-13 -12 -11 -10 -9 -8 -7 -6 -5 -4 -30

20

40

60

80

100

120 D2R (WT)I105KI195FI105K/I195F

D

Log [Dopamine], M

Perc

ent o

f max

imal

resp

onse

Perc

ent o

f max

imal

resp

onse

Log [Dopamine], M

E

(V83

L) +

(V91

S)V8

3L/V

91S

(M11

7F) +

(Y19

9F)

M11

7F/Y

199F

(I48T

) + (F

110W

)I4

8T/F

110W

(F20

2L) +

(Y21

3I)

F202

L/Y2

13I

(V15

2A) +

(L17

1P)

V152

A/L

171P

(N12

4H) +

(T20

5M)

N12

4H/T

205M

(M11

7F) +

(L38

7C)

M11

7F/L

387C

(I105

K) +

(I19

5F)

I105

K/I1

95F

(V19

1L) +

(S40

9N)

V191

L/S4

09N

-1

0

1

2

3

4

5

ln EC50EC50WT

mutant A( (

*

**

*

*

*

*

F

Dev

iatio

n fr

om a

dditi

vity

Evolutionary coupling potential

G

0 20 40 60 80 1000

2

4

6

8

Dopamine

0 20 40 60 80 1000

1

2

3

Dopamine

Rel

ativ

e ac

tivat

ion

by d

opam

ine

ln EC50EC50WT

mutant AB( (

+ ln EC50EC50WT

mutant B( (

V83L

V91S

V83L

/V91

SM

117F

Y199

FM

117F

/Y19

9FT2

05M

L379

FT2

05M

/L37

9FS1

93G

C38

5MS1

93G

/C38

5M I48T

F110

WI4

8T/F

110W

F202

LY2

13I

F202

L/Y2

13I

V152

AL1

71P

V152

A/L

171P

N12

4HT2

05M

N12

4H/T

205M

N12

4HL3

79F

N12

4H/L

379F

M11

7FL3

87C

M11

7F/L

387C

I105

KI1

95F

I105

K/I1

95F

V191

LS4

09N

V191

L/S4

09N

0

1

2

3

4

5

6

7

* * *

******

**

*** *

*

*

*

*

**

* **

***

*

*****

* * *

****

****

*

*

* *

*** *

**

20 40 60 80 100

0.0

0.5

1.0

1.5

2.0

M117F/L387C

F202L/Y213IV152A/L171P

I105K/I195F

V191L/S409N

N124H/T205M

M117F/Y199F

I48T/F110W

V83L/V91S

R = 0.762

ln (o

bser

ved

or a

dditi

ve ra

tio)

Fig. 2. Functional coupling between covarying ET residue pairs was observed at Gi activation by dopamine. (A and B) Examples of D2R activation curves forcovarying ET residue pairs. Membrane potential changes induced by Gi activation in response to dopamine stimulation of D2Rs were detected with themembrane potential assay. HEK293 cells stably expressing TRPC4β were transiently transfected with negative control [pcDNA3.1(+)], WT, or mutant D2Rplasmids, loaded with potential-sensing dye, and stimulated with 10 μM dopamine after baseline fluorescence was read for 30 s. Each trace was normalized byreceptor surface expression after subtraction of the signal of negative control cells. (C) Maximal activation of mutant D2Rs, normalized to WT (bars indicatemean ± SEM, n = 3–8). Mutants were compared withWT using one-sample t test against WT ≡ 1 (asterisks directly above bars), and mutants within a covaryinggroup were compared with each other using one-way ANOVA followed by Bonferroni’s multiple comparison test (asterisks with brackets) (**P < 0.001; *P <0.05). Bars are color-coded according to the evolutionary coupling potential predicted using amino acid sequences sharing >35% identity with D2R; see TableS2 for values. (D and E) Dopamine dose–response curves for Gi activation were generated with the membrane potential assay as in A. Examples of nonadditiveand additive effects of double mutations on the potency of dopamine are shown in D and E, respectively. (F) Dopamine EC50 values for WT and mutant D2Rswere determined by dose–response curves (details in Table S6) and used for the log-additive analysis. Results represent mean ± SEM (n = 3–7; **P < 0.001; *P <0.05; independent two-tailed Student’s t tests). The color scale is as in C. (G) Evolutionary coupling potential, predicted as described for C, is plotted againstdeviation from additivity jlnðEC50mutant A=EC50 WT Þ+ lnðEC50mutant B=EC50 WT Þ− lnðEC50mutant AB=EC50 WT Þj using EC50 data shown in F. The two were highly cor-related (Pearson’s R = 0.762, P = 0.017). (H) Positions of the covarying ET residues, shown as spheres (Cα atoms), mapped onto the structure of D3R (PDB 3PBL).Different colors indicate different groups of covarying ET residue pairs.

Sung et al. PNAS | March 29, 2016 | vol. 113 | no. 13 | 3541

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 28

, 202

0

their involvement in allosteric pathways linking ligand bindingand receptor conformational change. Further, the free energychange demonstrated that ET-MIp was able to detect energeticcoupling for ligand binding at both proximal (3.6 Å) and long-distance (13.2 Å) residue pairs, even when located far away fromthe ligand binding site. Because the double mutations F202L5.51/Y213I5.62 and T205M5.54/L379F6.41 rendered both dopamine andserotonin binding energetically more favorable (Fig. 4 C and D),we did not find evidence for functional coupling responsible forthe specificity of ligand binding.

G Protein-Specific Effects. Our new assay for Gi activation let uscompare mutational effects on D2R activation of Gi vs. G16 (23),and revealed several striking contrasts. For serotonin activation,N124H3.42 greatly decreased maximal activation of Gi (Fig. 3C)but significantly increased maximal activation of G16. For maximaldopamine activation, T205M5.54, F110W3.28, I48T1.46, and V91S2.61

all displayed activation of G16 close to that of WT (23), but acti-vated the physiological partner of D2R, Gi, at levels ranging from2.5-fold to nearly fourfold higher than WT (Fig. 2C), implyingsomewhat different coupling mechanisms for these two G proteins.

DiscussionET-MIp predicts functionally coupled residue pairs by weighingmutual variations by the phylogenetic depth of the associatedevolutionary divergences. Experiments in D2R suggest themethod can reveal allosteric communication between distant andevolutionarily important residue positions.

The most striking observations are that covarying ET residuepairs work together to modulate efficacy and potency of dopa-mine in D2R, and that the main function encoded by the allo-steric communication is to maintain the dopamine response atthe WT level and fine-tune sensitivity to dopamine. ET-MIpidentified functional couplings that regulate receptor−ligand in-teractions through residue pairs outside the ligand binding pocket,underscoring the allosteric nature of the pathways extending fromthe ligand binding site. Covarying ET residue pairs control do-pamine responsiveness tightly, possibly reflecting an evolutionaryneed to maintain distinct dopamine affinities among dopaminereceptors (D1, D2, D3, D4, and D5) (34–36), each activating di-verse downstream signaling pathways. Coupling scores calculatedat varying evolutionary depth indicate that coevolution coincideswith functional separation of the subfamilies fairly late in evolu-tion. The results of cosubstitution suggest that deviation in eitherdirection from this highly tuned response conferred selectiveevolutionary disadvantage.There have been few studies of the functional relationships

between covarying residues (14, 16–18, 41). Our results revealfunctional coupling of the covarying ET residues in D2R, and thesynergistic or antagonistic features of the coupling. For example,F202L5.51 and Y213I5.62 have a synergistic effect on dopamineefficacy, whereas the effects on dopamine potency are antago-nistic (Fig. 2 C and F), suggesting that different interactionsgovern efficacy and potency. The observation that covarying ETresidue pairs mediate ligand-specific functional interactionswithout contacting the ligand directly supports the previousproposal of a conformational filter in D2R (23).Previous studies have proposed models of molecular switches

associated with receptor activation, such as the ionic lock (D/E)RY(42–47), transmission switch (48), and tyrosine toggle switch(NPxxY) (44–47, 49). The discovery of functionally coupledcovarying ET residue pairs provides insight into the allostericpathways connecting ligand binding, molecular switches, and Gprotein coupling in D2R. Based on the positions of the covaryingET residue pairs mapped onto the structure of D3R (50), we canclassify the coupling mechanisms into three categories. (i) Thecovarying ET residues are far from each other (Table S1), withone at or near the ligand binding site and the other close to themolecular switches, e.g., I48T1.46/F110W3.28. F1103.28 is atthe orthosteric binding pocket (50, 51), and I481.46 is in directcontact with D802.50, which interacts with the transmission switchand NPxxY motif through water molecules in molecular dy-namics simulations of β1 and β2 adrenergic receptors (52), sug-gesting a role for this pair in coupling ligand binding to switchresidues. (ii) Both of the covarying ET residues are at or near theligand binding site, with one close to the 3–7 lock switch(D1143.32–Y4167.43), which forms a link between TM3 and TM7that breaks upon receptor activation in rhodopsin (47, 50, 53).These include V83L2.53/V91S2.61 and M117F3.35/Y199F5.48. Aside-chain rotamer of V83L2.53 can interact with Y4167.43

(Fig. S6A) and an inward-facing side-chain rotamer of M117F3.35

could interact with D1143.32, if TM2 and TM3 move apart fromeach other (Fig. S6B). Their coupling may contribute to propa-gation of the dopamine binding signal to the 3–7 lock. (iii) Bothof the covarying ET residues are far from the ligand bindingpocket but close to the molecular switches or G protein-binding region. These include F202L5.51/Y213I5.62, N124H3.42/T205M5.54, and T205M5.54/L379F6.41. F2025.51 is part of thetransmission switch, and the side chain of the mutant F202L5.51 cancontact the side chain of I1223.40, which is part of the transmissionswitch (48) (Fig. S6C). Y2135.62 interacts with V2155.64, which wasfound to participate in receptor−G protein interactions (44, 46)(Fig. S6C). N124H3.42, T205M5.54, and L379F6.41 may interact withthe residues making up the hydrophobic barrier in TM2, TM3, andTM6 (47, 54). In addition, T2055.54 and L3796.41 are in direct contactwith P2015.50 and F3826.44, respectively, of the transmission switch

Rel

ativ

e ac

tivat

ion

by s

erot

onin

high low

Evolutionary Coupling Potential

D2R (WT)I48TF110WI48T/F110W

B

C

Act

ivat

ion

by s

erot

onin

Time (s) Time (s)

Act

ivat

ion

by s

erot

onin D2R (WT)

N124HT205MN124H/T205M

0 20 40 60 80 100

0

2

4

6

8

10

Serotonin

A

0 20 40 60 80 100

0

2

4

6

8

10

Serotonin

V83L

V91S

V83L

/V91

SM

117F

Y199

FM

117F

/Y19

9FT2

05M

L379

FT2

05M

/L37

9FS1

93G

C38

5MS1

93G

/C38

5M I48T

F110

WI4

8T/F

110W

F202

LY2

13I

F202

L/Y2

13I

V152

AL1

71P

V152

A/L

171P

N12

4HT2

05M

N12

4H/T

205M

N12

4HL3

79F

N12

4H/L

379F

M11

7FL3

87C

M11

7F/L

387C

I105

KI1

95F

I105

K/I1

95F

V191

LS4

09N

V191

L/S4

09N

0

1

2

3

4

5

6

7

8

***

*

*

* *

**

*

****

**

**** *

*

*

* * **

*

*

**

****

*** * **

**

**

******

**

**

*

* *

****

**

Fig. 3. A rescue effect on serotonin responses was observed in some of thecovarying ET residue pairs. (A and B) Membrane potential changes inducedby Gi activation in response to serotonin stimulation of D2Rs were measuredand analyzed as in Fig. 2 A and B. (C) Maximal activation by serotonin ofmutant D2Rs was normalized to that of WT, which was defined as 1. Bars,statistics, and color-coding are as described for Fig. 2C.

3542 | www.pnas.org/cgi/doi/10.1073/pnas.1516579113 Sung et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 28

, 202

0

(48) (Fig. S6D). A backbone hydrogen bond links L379F6.41 andthe G protein interacting residue L3756.37 (V3316.37 in D3R) (46)(Fig. S6E). Thus, T205M5.54 and L379F6.41 form the lynchpin in a

continuous network of contacts leading from the transmissionswitch to the G protein. The observation that T205M5.54/L379F6.41

are coupled in ligand binding is consistent with bidirectional com-munication between ligand binding and G protein activation. Theidea that coupled residues are links in a chain of allosteric inter-actions is supported by the observation that intervening resi-dues contacting them tend to have high coupling scores (Fig. S7and Table S9). However, these residues are identical in D2R and5-HT2AR and could not be tested in our paradigm.The side chain of M117F3.35 points toward the putative choles-

terol binding site (55, 56) and may enhance cholesterol interac-tions; Y199F5.48 points toward the putative receptor−dimerizationinterface (57, 58). V1524.42 contacts cholesterol-contacting positionI1564.46 (56), and L1714.61 points toward the dimerization interface.The side chains of both F202L5.51 and Y213I5.62 may lie at thesame interface. The functional coupling observed may reflect ef-fects on modulation by cholesterol (59) and GPCR dimerization(57, 60). The equivalent residue to Y2135.62 in activated β2 ad-renergic receptor (46) stabilizes the “outward” movement ofTM6 through a van der Waals contact, consistent with effects ofY213I5.62 on G protein coupling.Overall, the results support the idea that covariation patterns

are signatures preserved in protein sequences during evolutionand reflect functional interactions important for fitness-confer-ring properties. They open the possibility of improving structure-based drug design by accounting for intramolecular allostericcommunication. The involvement of covarying ET residue pairsin allosteric pathways linking ligand binding, molecular switches,and G protein coupling offers the potential to reengineer allo-steric pathways of receptor activation. The divergent effects onGi and G16 (23) activation observed in some mutants suggest thatcovarying ET residue pairs play a role in governing receptorpreference for downstream effectors. Thus, this work also servesas a starting point for studies on bias in activation of effectorscontrolled by covarying ET residue pairs, and on the interpretationof genome variations.

Materials and MethodsThe key methods are briefly described here. For full details, please seeSupporting Information.

Evolutionary Trace and MI Analysis. To identify evolutionarily coupled residuepairs, we integrated MI into the ET framework as follows:

ETðMIpði, jÞÞ=XN

n=1

1n

Xn

g=1

MIpgði, jÞ.

The alignment is broken up into subalignments g according to the phylo-genetic tree with N nodes. Our measure of mutual information, MIp (5), iscomputed for all possible residue pairs i and j for each subalignment selectedby the phylogenetic tree. The performance of ET-MIp in contact predictionwas compared with other methods as described in Supporting Information,using an alignment of 2,500 Class A GPCRs to predict interresidue contacts inthe structure of the β2 adrenergic receptor (PDB 2RH1). To identify covaryingresidues in D2R, BLAST (Basic Local Alignment Search Tool) analysis of D2Rwas first performed against the Uniref90 sequence database (61). To identifyhomologs, protein sequences were filtered by protein length and sequenceidentity (>35%, >42%, >50%), where each alignment was respectively madeup of all dopamine receptors, dopamine D2 and D3 receptors, and only D2receptors. The bioamine and Class A GPCR alignments were described pre-viously (23).

Membrane Potential Assay. Gi activation induced by agonist-stimulated D2Rtriggers the opening of TRPC4β channels in HEK293 cells, leading to mem-brane potential changes (62). Details are given in Supporting Information.

ACKNOWLEDGMENTS. We thank Melina A. Agosto and Rhonald Lua forconstructive suggestions, andMichael X. Zhu for providing the TRPC4β-expressingHEK293 cells. This work was supported by NIH Grants R01-GM066099, R01-EY011900, R01-EY007981, R01-GM079656, and T90-DK070109; by National Sci-ence Foundation Grant DBI-1356569; and by Welch Foundation Grant Q-0035.

T205ML379FT205M/L379F

D2R (WT)A

-7 -6 -5 -4 -3 -2 -1

020406080

100120

B

% [3 H

] spi

pero

ne b

indi

ng

Log [Serotonin], M

T205ML379FT205M/L379F

D2R (WT)

-8 -7 -6 -5 -4 -3 -2 -1

020406080

100120

% [3 H

] spi

pero

ne b

indi

ng

Log [Dopamine], M

Cha

nge

in G

ibbs

free

ene

rgy

upon

dopa

min

e bi

ndin

g (k

J/ m

ol)

high low

Evolutionary Coupling PotentialC ∆∆G0mutant A + ∆∆G0

mutant B∆∆G0

mutant AB

Cha

nge

in G

ibbs

free

ene

rgy

upon

sero

toni

n bi

ndin

g (k

J/ m

ol)

D

(M11

7F) +

(Y19

9F)

M11

7F/Y

199F

(T20

5M) +

(L37

9F)

T205

M/L

379F

(I48T

) + (F

110W

)I4

8T/F

110W

(F20

2L) +

(Y21

3I)

F202

L/Y2

13I

(V15

2A) +

(L17

1P)

V152

A/L

171P

(N12

4H) +

(T20

5M)

N12

4H/T

205M

(N12

4H) +

(L37

9F)

N12

4H/L

379F

(M11

7F) +

(L38

7C)

M11

7F/L

387C

(I105

K) +

(I19

5F)

I105

K/I1

95F

(V19

1L) +

(S40

9N)

V191

L/S4

09N

-8

-6

-4

-2

0

2

4

6

(M11

7F) +

(Y19

9F)

M11

7F/Y

199F

(T20

5M) +

(L37

9F)

T205

M/L

379F

(I48T

) + (F

110W

)I4

8T/F

110W

(F20

2L) +

(Y21

3I)

F202

L/Y2

13I

(V15

2A) +

(L17

1P)

V152

A/L

171P

(N12

4H) +

(T20

5M)

N12

4H/T

205M

(N12

4H) +

(L37

9F)

N12

4H/L

379F

(M11

7F) +

(L38

7C)

M11

7F/L

387C

(I105

K) +

(I19

5F)

I105

K/I1

95F

(V19

1L) +

(S40

9N)

V191

L/S4

09N

-4

-2

0

2

4

6

8

10

*

*

*

**

Fig. 4. Nonadditive effects on Gibbs free energy change upon ligand binding.(A and B) The [3H]spiperone binding to WT or mutant D2Rs was detected in thepresence of various concentrations of competing dopamine or serotonin. Ex-amples of competition binding curves for dopamine and serotonin are shown inA and B, respectively. Nonlinear regression analysis was performed with theone-site model to determine the IC50, which was then used in the Ki. (C and D)Values for free energy change (ΔG0) upon dopamine or serotonin binding werederived from ΔG0 = RT lnKi, where R = 8.314 J K−1·mol−1 and T = 298.15 K.The mutational effects on free energy change upon binding to dopamine(C) or serotonin (D) were examined by calculating ΔΔG0 of each D2R mutant asΔΔG0

mutant = ΔG0mutant − ΔG0

WT. Results of the log-additive analysis areexpressed as mean ± SEM (n = 3–6; **P < 0.001; *P < 0.05; independent two-tailed Student’s t tests). Color-coding of the bars is as described for Fig. 2C.

Sung et al. PNAS | March 29, 2016 | vol. 113 | no. 13 | 3543

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 28

, 202

0

1. Ortlund EA, Bridgham JT, Redinbo MR, Thornton JW (2007) Crystal structure of anancient protein: Evolution by conformational epistasis. Science 317(5844):1544–1548.

2. Jones DT, Buchan DW, Cozzetto D, Pontil M (2012) PSICOV: Precise structural contactprediction using sparse inverse covariance estimation on large multiple sequencealignments. Bioinformatics 28(2):184–190.

3. Nemoto W, Imai T, Takahashi T, Kikuchi T, Fujita N (2004) Detection of pairwiseresidue proximity by covariation analysis for 3D-structure prediction of G-protein-coupled receptors. Protein J 23(6):427–435.

4. Fuchs A, et al. (2007) Co-evolving residues in membrane proteins. Bioinformatics23(24):3312–3319.

5. Dunn SD, Wahl LM, Gloor GB (2008) Mutual information without the influenceof phylogeny or entropy dramatically improves residue contact prediction.Bioinformatics 24(3):333–340.

6. Burger L, van Nimwegen E (2010) Disentangling direct from indirect co-evolution ofresidues in protein alignments. PLOS Comput Biol 6(1):e1000633.

7. Morcos F, et al. (2011) Direct-coupling analysis of residue coevolution captures nativecontacts across many protein families. Proc Natl Acad Sci USA 108(49):E1293–E1301.

8. Marks DS, et al. (2011) Protein 3D structure computed from evolutionary sequencevariation. PLoS One 6(12):e28766.

9. Hopf TA, et al. (2012) Three-dimensional structures of membrane proteins from ge-nomic sequencing. Cell 149(7):1607–1621.

10. Nugent T, Jones DT (2012) Accurate de novo structure prediction of large trans-membrane protein domains using fragment-assembly and correlated mutationanalysis. Proc Natl Acad Sci USA 109(24):E1540–E1547.

11. Sułkowska JI, Morcos F, Weigt M, Hwa T, Onuchic JN (2012) Genomics-aided structureprediction. Proc Natl Acad Sci USA 109(26):10340–10345.

12. Weigt M, White RA, Szurmant H, Hoch JA, Hwa T (2009) Identification of direct res-idue contacts in protein−protein interaction by message passing. Proc Natl Acad SciUSA 106(1):67–72.

13. Livesay DR, Kreth KE, Fodor AA (2012) A critical evaluation of correlated mutationalgorithms and coevolution within allosteric mechanisms. Methods Mol Biol 796:385–398.

14. Lockless SW, Ranganathan R (1999) Evolutionarily conserved pathways of energeticconnectivity in protein families. Science 286(5438):295–299.

15. Gloor GB, Martin LC, Wahl LM, Dunn SD (2005) Mutual information in protein mul-tiple sequence alignments reveals two classes of coevolving positions. Biochemistry44(19):7156–7165.

16. Chakrabarti S, Panchenko AR (2009) Coevolution in defining the functional specificity.Proteins 75(1):231–240.

17. Süel GM, Lockless SW, Wall MA, Ranganathan R (2003) Evolutionarily conservednetworks of residues mediate allosteric communication in proteins. Nat Struct Biol10(1):59–69.

18. Halabi N, Rivoire O, Leibler S, Ranganathan R (2009) Protein sectors: Evolutionaryunits of three-dimensional structure. Cell 138(4):774–786.

19. Chi CN, et al. (2008) Reassessing a sparse energetic network within a single proteindomain. Proc Natl Acad Sci USA 105(12):4679–4684.

20. Tesileanu T, Colwell LJ, Leibler S (2015) Protein sectors: Statistical coupling analysisversus conservation. PLOS Comput Biol 11(2):e1004091.

21. Madabushi S, et al. (2004) Evolutionary trace of G protein-coupled receptors revealsclusters of residues that determine global and class-specific functions. J Biol Chem279(9):8126–8132.

22. Shenoy SK, et al. (2006) beta-arrestin-dependent, G protein-independent ERK1/2 ac-tivation by the beta2 adrenergic receptor. J Biol Chem 281(2):1261–1273.

23. Rodriguez GJ, Yao R, Lichtarge O, Wensel TG (2010) Evolution-guided discovery andrecoding of allosteric pathway specificity determinants in psychoactive bioamine re-ceptors. Proc Natl Acad Sci USA 107(17):7787–7792.

24. Kang HJ, Wilkins AD, Lichtarge O, Wensel TG (2015) Determinants of endogenousligand specificity divergence among metabotropic glutamate receptors. J Biol Chem290(5):2870–2878.

25. Lichtarge O, Bourne HR, Cohen FE (1996) An evolutionary trace method definesbinding surfaces common to protein families. J Mol Biol 257(2):342–358.

26. Katsonis P, Lichtarge O (2014) A formal perturbation equation between genotypeand phenotype determines the Evolutionary Action of protein-coding variations onfitness. Genome Res 24(12):2050–2058.

27. Wilkins AD, et al. (2013) Accounting for epistatic interactions improves the functionalanalysis of protein structures. Bioinformatics 29(21):2714–2721.

28. Bonci A, Hopf FW (2005) The dopamine D2 receptor: New surprises from an oldfriend. Neuron 47(3):335–338.

29. Yan Z, Song WJ, Surmeier J (1997) D2 dopamine receptors reduce N-type Ca2+ cur-rents in rat neostriatal cholinergic interneurons through a membrane-delimited,protein-kinase-C-insensitive pathway. J Neurophysiol 77(2):1003–1015.

30. Einhorn LC, Oxford GS (1993) Guanine nucleotide binding proteins mediate D2 do-pamine receptor activation of a potassium channel in rat lactotrophs. J Physiol 462:563–578.

31. Luo Y, Kokkonen GC, Wang X, Neve KA, Roth GS (1998) D2 dopamine receptorsstimulate mitogenesis through pertussis toxin-sensitive G proteins and Ras-involvedERK and SAP/JNK pathways in rat C6-D2L glioma cells. J Neurochem 71(3):980–990.

32. Hernandez-Lopez S, et al. (2000) D2 dopamine receptors in striatal medium spinyneurons reduce L-type Ca2+ currents and excitability via a novel PLC[beta]1-IP3-calcineurin-signaling cascade. J Neurosci 20(24):8987–8995.

33. Enjalbert A, Bockaert J (1983) Pharmacological characterization of the D2 dopaminereceptor negatively coupled with adenylate cyclase in rat anterior pituitary. MolPharmacol 23(3):576–584.

34. Sunahara RK, et al. (1991) Cloning of the gene for a human dopamine D5 receptorwith higher affinity for dopamine than D1. Nature 350(6319):614–619.

35. Van Tol HH, et al. (1991) Cloning of the gene for a human dopamine D4 receptor withhigh affinity for the antipsychotic clozapine. Nature 350(6319):610–614.

36. Sokoloff P, Giros B, Martres MP, Bouthenet ML, Schwartz JC (1990) Molecular cloningand characterization of a novel dopamine receptor (D3) as a target for neuroleptics.Nature 347(6289):146–151.

37. Baldassi C, et al. (2014) Fast and accurate multivariate Gaussian modeling of proteinfamilies: Predicting residue contacts and protein-interaction partners. PLoS One 9(3):e92721.

38. Shannon CE (1997) The mathematical theory of communication. 1963. MD Comput14(4):306–317.

39. Mani R, St Onge RP, Hartman JL, 4th, Giaever G, Roth FP (2008) Defining geneticinteraction. Proc Natl Acad Sci USA 105(9):3461–3466.

40. Araya CL, et al. (2012) A fundamental protein property, thermodynamic stability,revealed solely from large-scale measurements of protein function. Proc Natl Acad SciUSA 109(42):16858–16863.

41. Oliveira L, Paiva AC, Vriend G (2002) Correlated mutation analyses on very large se-quence families. ChemBioChem 3(10):1010–1017.

42. Palczewski K, et al. (2000) Crystal structure of rhodopsin: A G protein-coupled re-ceptor. Science 289(5480):739–745.

43. Ballesteros JA, et al. (2001) Activation of the beta 2-adrenergic receptor involvesdisruption of an ionic lock between the cytoplasmic ends of transmembrane seg-ments 3 and 6. J Biol Chem 276(31):29171–29177.

44. Choe HW, et al. (2011) Crystal structure of metarhodopsin II. Nature 471(7340):651–655.

45. Rasmussen SG, et al. (2011) Structure of a nanobody-stabilized active state of the β(2)adrenoceptor. Nature 469(7329):175–180.

46. Rasmussen SG, et al. (2011) Crystal structure of the β2 adrenergic receptor-Gs proteincomplex. Nature 477(7366):549–555.

47. Standfuss J, et al. (2011) The structural basis of agonist-induced activation in consti-tutively active rhodopsin. Nature 471(7340):656–660.

48. Deupi X, Standfuss J (2011) Structural insights into agonist-induced activation ofG-protein-coupled receptors. Curr Opin Struct Biol 21(4):541–551.

49. Fritze O, et al. (2003) Role of the conserved NPxxY(x)5,6F motif in the rhodopsinground state and during activation. Proc Natl Acad Sci USA 100(5):2290–2295.

50. Chien EY, et al. (2010) Structure of the human dopamine D3 receptor in complex witha D2/D3 selective antagonist. Science 330(6007):1091–1095.

51. Wang C, et al. (2013) Structural basis for molecular recognition at serotonin receptors.Science 340(6132):610–614.

52. Vanni S, Neri M, Tavernelli I, Rothlisberger U (2009) Observation of “ionic lock” for-mation in molecular dynamics simulations of wild-type beta 1 and beta 2 adrenergicreceptors. Biochemistry 48(22):4789–4797.

53. Trzaskowski B, et al. (2012) Action of molecular switches in GPCRs—Theoretical andexperimental studies. Curr Med Chem 19(8):1090–1109.

54. Li J, Edwards PC, Burghammer M, Villa C, Schertler GF (2004) Structure of bovinerhodopsin in a trigonal crystal form. J Mol Biol 343(5):1409–1438.

55. Cherezov V, et al. (2007) High-resolution crystal structure of an engineered humanbeta2-adrenergic G protein-coupled receptor. Science 318(5854):1258–1265.

56. Hanson MA, et al. (2008) A specific cholesterol binding site is established by the 2.8 Astructure of the human β2-adrenergic receptor. Structure 16(6):897–905.

57. Guo W, Shi L, Filizola M, Weinstein H, Javitch JA (2005) Crosstalk in G protein-coupledreceptors: Changes at the transmembrane homodimer interface determine activa-tion. Proc Natl Acad Sci USA 102(48):17495–17500.

58. Guo W, Shi L, Javitch JA (2003) The fourth transmembrane segment forms the in-terface of the dopamine D2 receptor homodimer. J Biol Chem 278(7):4385–4388.

59. Zocher M, Zhang C, Rasmussen SG, Kobilka BK, Müller DJ (2012) Cholesterol increaseskinetic, energetic, and mechanical stability of the human β2-adrenergic receptor. ProcNatl Acad Sci USA 109(50):E3463–E3472.

60. Han Y, Moreira IS, Urizar E, Weinstein H, Javitch JA (2009) Allosteric communicationbetween protomers of dopamine class A GPCR dimers modulates activation. NatChem Biol 5(9):688–695.

61. Suzek BE, Huang H, McGarvey P, Mazumder R, Wu CH (2007) UniRef: Comprehensiveand non-redundant UniProt reference clusters. Bioinformatics 23(10):1282–1288.

62. Kang HJ, et al. (2014) Selectivity and evolutionary divergence of metabotropic glu-tamate receptors for endogenous ligands and G proteins coupled to phospholipase Cor TRP channels. J Biol Chem 289(43):29961–29974.

3544 | www.pnas.org/cgi/doi/10.1073/pnas.1516579113 Sung et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 28

, 202

0