Article

Epistasis Constrains Mutational Pathways of HemoglobinAdaptation in High-Altitude PikasDanielle M Tuftsy1 Chandrasekhar Natarajany1 Inge G Revsbech2 Joana Projecto-Garcia1

Federico G Hoffmann34 Roy E Weber2 Angela Fago2 Hideaki Moriyama1 and Jay F Storz1

1School of Biological Sciences University of Nebraska Lincoln2Department of Bioscience Zoophysiology Aarhus University Aarhus Denmark3Department of Biochemistry Molecular Biology Entomology and Plant Pathology Mississippi State University4Institute for Genomics Biocomputing and Biotechnology Mississippi State University

yThese authors contributed equally to this work

Corresponding author E-mail jstorz2unledu

Associate editor David Irwin

Abstract

A fundamental question in evolutionary genetics concerns the roles of mutational pleiotropy and epistasis in shapingtrajectories of protein evolution This question can be addressed most directly by using site-directed mutagenesis toexplore the mutational landscape of protein function in experimentally defined regions of sequence space Here weevaluate how pleiotropic trade-offs and epistatic interactions influence the accessibility of alternative mutational path-ways during the adaptive evolution of hemoglobin (Hb) function in high-altitude pikas (Mammalia Lagomorpha) Bycombining ancestral protein resurrection with a combinatorial protein-engineering approach we examined the func-tional effects of sequential mutational steps in all possible pathways that produced an increased HbndashO2 affinity Theseexperiments revealed that the effects of mutations on HbndashO2 affinity are highly dependent on the temporal order inwhich they occur Each of three -chain substitutions produced a significant increase in HbndashO2 affinity on the ancestralgenetic background but two of these substitutions produced opposite effects when they occurred as later steps in thepathway The experiments revealed pervasive epistasis for HbndashO2 affinity but affinity-altering mutations produced nosignificant pleiotropic trade-offs These results provide insights into the properties of adaptive substitutions in naturallyevolved proteins and suggest that the accessibility of alternative mutational pathways may be more strongly constrainedby sign epistasis for positively selected biochemical phenotypes than by antagonistic pleiotropy

Key words adaptation epistasis hemoglobin high altitude molecular evolution protein evolution

IntroductionThe adaptive evolution of protein function may typically in-volve the sequential fixation of individual amino acid muta-tions with each substitution producing an incrementalimprovement in the selected biochemical phenotypeHowever there are two interrelated factors that can poten-tially affect fixation probabilities and constrain the number ofselectively accessible pathways to high-fitness genotypesMutational pleiotropy (where a given mutation perturbs mul-tiple phenotypes) and epistasis (nonadditive interactions be-tween mutations) As amino acid mutations often affectmultiple aspects of protein biochemistry (Taverna andGoldstein 2002 Wang et al 2002 DePristo et al 2005Bloom et al 2006 Weinreich et al 2006 Tokuriki et al2008 Tokuriki and Tawfik 2009) mutations that improveone aspect of protein function may simultaneously compro-mise other functions Pleiotropic trade-offs can also give riseto a context-dependence of mutational effects as the phe-notypic effect of a given mutation may be contingent on theexistence of compensatory mutations at other sites in thesame protein

During the adaptive evolution of a particular proteinfunction epistasis can affect the fixation probabilities of func-tion-altering mutations and influence trajectories of proteinevolution in two ways 1) The sign or magnitude of the mu-tationrsquos effect on the selected phenotype may be conditionalon genetic background and 2) the sign or magnitude of themutationrsquos pleiotropic effects may be conditional on geneticbackground Both forms of epistatic interaction can influencewhether evolution is more likely to follow some pathwaysrather than others (Weinreich et al 2006 DePristo et al2007 Ortlund et al 2007 Poelwijk et al 2007 Bridghamet al 2009 Lozovsky et al 2009 Carneiro and Hartl 2010 daSilva et al 2010 Kryazhimskiy et al 2011 2014 Kvitek andSherlock 2011 Rokyta et al 2011 Salverda et al 2011 Ostmanet al 2012 Weinreich and Knies 2013 de Visser and Krug2014 Harms and Thornton 2014)

A compelling system for investigating the pleiotropic andepistatic effects of function-altering mutations is provided bythe evolution of increased hemoglobin (Hb)-O2 affinities invertebrate species that have adapted to environmental hyp-oxia Among vertebrates elevated HbndashO2 affinities have

The Author 2014 Published by Oxford University Press on behalf of the Society for Molecular Biology and EvolutionThis is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License(httpcreativecommonsorglicensesby-nc40) which permits non-commercial re-use distribution and reproduction in anymedium provided the original work is properly cited For commercial re-use please contact journalspermissionsoupcom Open AccessMol Biol Evol doi101093molbevmsu311 Advance Access publication November 18 2014 1

MBE Advance Access published December 3 2014 at U

niversity of Nebraka-L

incoln Libraries on D

ecember 8 2014

httpmbeoxfordjournalsorg

Dow

nloaded from

evolved in multiple species that have adapted to chronic O2

deprivation at high altitude (Storz 2007 Weber 2007 Storzand Moriyama 2008 Storz et al 2009 Storz Runck et al 2010Natarajan et al 2013 Projecto-Garcia et al 2013 Revsbechet al 2013) Under severe hypoxia an increased HbndashO2 affin-ity can help preserve an adequate level of tissue oxygenationby enhancing pulmonary O2 uptake while simultaneouslymaintaining the pressure gradient that drives O2 diffusionfrom capillary blood to the tissue mitochondria (reviewedby Storz Scott et al 2010 Mairbeuroaurl and Weber 2012)Even slight increases in HbndashO2 affinity can efficiently safe-guard arterial O2 saturation under hypoxia thereby comple-menting numerous other physiological adjustments in thecardiopulmonary system (Turek and Kreuzer 1976 TurekKreuzer Ringnalda 1978 Turek Kreuzer Turek-Maischeideret al 1978 Mairbeuroaurl 1994)

Mammalian Hb is a heterotetramer consisting of two-type subunits and two -type subunits that each containa heme groupmdasha porphyrin ring with a ferrous (Fe2+) ironatom capable of reversibly binding a single dioxygen moleculeThe Hb tetramer undergoes an oxygenation-linked transitionin quaternary structure whereby the two semirigid 11 and22 dimers rotate around each other by 15 during thetransition between the deoxy (T-state) and oxy (R-state) con-formations (Perutz 1972 1989 Baldwin and Chothia 1979)This oxygenation-linked shift in quaternary structure betweenthe T- and R-states is central to the allosteric function of Hb asan O2-transport molecule and underpins the subunit coop-erativity of O2-binding as well as the sensitivity of O2-affinityto allosteric effectors such as protons Cl ions and organicphosphates as well as the sensitivity of O2-affinity to allostericeffectors such as protons Cl- ions and organic phosphatesAllosteric effectors have the effect of reducing HbndashO2 affinitybecause they preferentially bind to deoxyHb thereby stabiliz-ing the low-affinity T structure through the formation of ad-ditional hydrogen-bonds and salt bridges within and betweensubunits (Bettati et al 1983 Perutz 1989)

There are numerous possible mutational changes that canalter HbndashO2 affinity but many or most such changes havedeleterious pleiotropic effects on other structural or func-tional properties (Bellelli et al 2006 Bonaventura et al2013 Varnado et al 2013) For example active site mutationsthat alter the polarity or hydrophobicity of the distal hemepocket can produce significant changes in ligand affinity butsuch mutations can reduce structural stability andor increaserates of spontaneous heme oxidation (autoxidation) whichrenders the oxidized Fe3+ ldquometHbrdquo incapable of O2-binding(Varnado et al 2013) Similarly mutations that increase HbndashO2 affinity by suppressing sensitivity to allosteric effectorssacrifice an important mechanism of phenotypic plasticitybecause changes in intraerythrocytic concentrations of or-ganic phosphates and other effectors are rendered less effec-tive as a means of modulating bloodndashO2 affinity in responseto transient changes in metabolic demand or environmentalO2 availability Thus mutations with identical main effects onHbndashO2 affinity may still be unequal in the eyes of naturalselection if they vary in the magnitude of antagonistic pleiot-ropy (Otto 2004 Streisfeld and Rausher 2011)

To assess the roles of pleiotropy and epistasis in the evo-lution of protein function we used a combinatorial proteinengineering approach to dissect the mechanistic basis ofevolved differences in HbndashO2 affinity between two sister spe-cies of North American pikas (Mammalia Lagomorpha) thathave contrasting altitudinal range limits Our comparison in-cluded the American pika (Ochotona princeps) an alpine spe-cialist and the closely related collared pika (O collaris) whichhas a predominantly lowland distribution in Alaska andnorthwestern Canada Ochotona princeps inhabits the highestsummits in the contiguous United States at elevations ofapproximately 4300 m where the partial pressure of O2

(PO2) is 60 of the sea level value (~96 vs ~160 torr) Pikasoriginated and diversified in the Palearctic (Erbajeva 1994)and O princeps and O collaris represent the only two extantpika species in North America The two species diverged ap-proximately 4ndash8 Ma (Lanier and Olson 2009) AlthoughO princeps is mainly found above timberline in high-altitudetalus habitats (Smith and Weston 1990 Hafner 1994) fossilsand biogeographic evidence indicate that the species alsoinhabited lower elevations during Quaternary glacial cycles(Mead 1987 Hafner 1993 Galbreath et al 2009 2010 Lanierand Olson 2013) Because of the dramatic difference in eleva-tional range limits between O princeps and O collaris com-parisons between these two closely related species provide anopportunity to identify the molecular basis of evolved differ-ences in Hb function that may have adaptive significance

Using ancestral protein resurrection and site-directed mu-tagenesis we measured the structural and functional effectsof sequential mutations in all possible pathways connectingthe resurrected ancestral (low-affinity) Hb and the derived(high-affinity) Hb of the high-altitude species In addition tomeasuring oxygenation properties of each engineered recom-binant Hb (rHb) mutant we also measured pleiotropic effectson functional properties that can potentially trade-off withHbndashO2 affinity

The experiments were designed to answer the followingquestions 1) What fraction of the evolved change in HbndashO2

affinity is attributable to substitutions at each step in thepathway 2) Are the functional effects of mutations condi-tional on genetic background For example do the functionaleffects of individual mutations depend on the sequentialorder in which they occur And 3) Do affinity-enhancingmutations have deleterious pleiotropic effects on other struc-tural or functional properties The experimental results re-vealed that the sign and magnitude of mutational effects onHbndashO2 affinity were highly dependent on the genetic back-ground in which they occurred but affinity-altering muta-tions produced no significant pleiotropic effects on otheraspects of Hb function

Results and DiscussionFor the experimental analysis of Hb function we collectedspecimens of O princeps from a high-altitude locality in theSouthern Rockies and specimens of O collaris from lowlandlocalities in interior Alaska (see Materials and Methods) Weexperimentally confirmed that both species express a singletetrameric 22 Hb isoform during adult life consistent with

2

Tufts et al doi101093molbevmsu311 MBE at U

niversity of Nebraka-L

incoln Libraries on D

ecember 8 2014

httpmbeoxfordjournalsorg

Dow

nloaded from

genomic evidence that O princeps possesses single copies ofpostnatally expressed - and -globin genes (HBA and HBBrespectively fig 1A) We sequenced both HBA alleles andboth HBB alleles from multiple individuals per species

The sequence data revealed that the Hbs of the two speciesare distinguished by five amino acid substitutions in the-chain subunits (sites 5 58 62 123 and 126fig 1B) Using blood samples from individuals with known

RabbitO princepsO collarisO hyperboreaO pallasiO curzoniaeO dauricaO rufescens

RabbitO princepsO collarisO hyperboreaO pallasiO curzoniaeO dauricaO rufescens

HBA

RabbitO princepsO collarisO hyperboreaO pallasiO curzoniaeO dauricaO rufescens

RabbitO princepsO collarisO hyperboreaO pallasiO curzoniaeO dauricaO rufescens

HBB

β-globin gene cluster

α-globin gene cluster

HBZ HBA HBQ

HBE HBG HBD HBB

0 kb 10 kb

A

B

FIG 1 Pika globin genes (A) Genomic structure of the - and -globin gene clusters of Ochotona princeps (GenBank accession numbers NT_167265and AC234505 respectively) Isoelectric focusing experiments revealed that adult O princeps and O collaris both express a single tetrameric 22 Hbisoform that incorporates the -chain products of the HBA gene and -chain products of the HBB gene Products of the remaining - and -like globingenes are not incorporated into functional Hb tetramers in definitive erythrocytes Based on data from other eutherian mammals products of three ofthe remaining globin genes are expected to encode -type (HBZ [-globin]) or -type (HBE [-globin] and HBG [-globin]) subunits that areincorporated into structurally distinct Hb isoforms that are expressed at different stages of prenatal development (B) Alignment of HBA and HBBamino acid sequences from seven pika species including the two North American sister species O princeps and O collaris Orthologous sequences fromthe European rabbit (Oryctolagus cuniculus) are provided for comparison In addition to the five -chain substitutions at sites 5 58 62 123 and 126 thatdistinguish the Hbs of O princeps and O collaris (denoted by boxes) there is one nearly fixed difference at site 78 in the -chain

3

Epistasis Constrains Pathways of Hemoglobin Evolution doi101093molbevmsu311 MBE at U

niversity of Nebraka-L

incoln Libraries on D

ecember 8 2014

httpmbeoxfordjournalsorg

Dow

nloaded from

HBA and HBB genotypes we tested for differences in theoxygenation properties of purified Hbs from highland O prin-ceps (n = 4) and lowland O collaris (n = 4) that were distin-guished by the five -chain substitutions To gain insights intofunctional mechanisms that may be responsible for evolveddifferences in O2-binding properties we measured O2-affini-ties of purified Hbs under four treatments In the absence ofallosteric effectors (stripped) in the presence of Cl ions(added as 01 M KCl) in the presence of 23-diphosphoglyce-rate (DPG at 2-fold molar excess over tetrameric Hb) and inthe simultaneous presence of KCl and DPG The (KCl+DPG)treatment most closely approximates in vivo conditions inmammalian red blood cells (Mairbeuroaurl and Weber 2012) sofrom this point forward we focus on comparisons of HbndashO2

affinity in the presence of anionic effectorsThe experiments revealed that the Hb of O princeps ex-

hibits a uniformly higher O2-affinity than that of O collaris asindicated by the lower values of P50 (the partial pressure of O2

at which Hb is 50 saturated table 1 and fig 2A) The Hb of Oprinceps exhibited a significantly higher intrinsic O2 affinitythan that of O collaris as indicated by the lower P50 forstripped Hb (P50(stripped)) and this difference persisted inthe presence of Cl ions (P50(KCl)) in the presence of DPG(P50(DPG)) and in the simultaneous presence of both allostericeffectors (P50(KCl+DPG) table 1 and fig 2A) Measures of allo-steric regulatory capacity (fig 2B) revealed that the speciesdifferences in P50(KCl+DPG) were exclusively attributable to dif-ferences in intrinsic O2 affinity there were no appreciabledifferences in the responsiveness to allosteric effectors Thiswas confirmed by dosendashresponse curves for DPG which dem-onstrated that HbndashO2 affinities of both species had identicalassociation constants for DPG binding (fig 3) Hbs of bothspecies were also characterized by highly similar cooperativitycoefficients (table 1)

Sequence data from additional species of pikas enabled usto reconstruct the -chain sequence of the princepscollariscommon ancestor (Anc1 fig 4A) For each of the five aminoacid sites posterior probabilities of estimated ancestral stateswere as follows 5G (1000) 58A (0984) 62A (0987)123T (0997) and 126A (0947) The ancestral sequencereconstruction revealed that three of the five -chain substi-tutions that distinguish the Hbs of the two species occurredon the branch leading to the high-altitude O princeps5GlyAla (G5A) 62AlaThr (A62T) and126AlaVal (A126V) and the remaining two substitu-tions occurred on the branch leading to the low-altitude Ocollaris 58AlaPro (A58P) and 123ThrSer (T123S)(fig 4A) As a first step in the protein-engineering experi-ments we synthesized purified and functionally testedrHbs representing the wildtype Hbs of O princeps andO collaris and the reconstructed Hb of their common ances-tor Anc1 (fig 4A) A triangulated comparison involving thereconstructed Anc1 and its modern-day descendants re-vealed that the elevated HbndashO2 affinity of O princeps is aderived character state P50(KCl+DPG) for the highland O prin-ceps rHb was 218 lower (ie O2 affinity was higher) relativeto Anc1 (supplementary table S1 Supplementary Material

P50

(tor

r)

0

5

10

15High (O princeps)Low (O collaris)

00

01

02

03

04

Δlo

g-P

50

stripped +KCl KCl+DPG+DPG KCl-str DPG-str (KCl+DPG)-str

A B

FIG 2 Hbs of the high-altitude Ochotona princeps exhibit uniformly higher O2-affinities than those of the low-altitude O collaris (A) P50 values (mean SEM) for purified Hbs of O princeps (n = 4 individuals) and O collaris (n = 4 individuals) measured at pH 74 and 37 C in the absence (stripped) andpresence of allosteric effectors ([Cl] 01 M [HEPES] 01 M DPGHb tetramer ratio 20 [heme] 03 mM) (B) Log-transformed differences in P50 valuesof Hbs from O princeps and O collaris in the presence and absence of allosteric effectors The log-P50 values measure the extent to which HbndashO2

affinity is reduced in the presence of a given allosteric effector (Cl DPG or both anions together) See table 1 for raw data

Table 1 O2-Affinities (P50 torr) and Cooperativity Coefficients (n50)(mean SEM) of Purified Hbs from High- and Low-Altitude PikaSpecies (Ochotona princeps and O collaris respectively)

O princeps (high altitude) O collaris (low altitude)(n = 4) (n = 4)

P50 (torr)

Stripped 520 029 623 016

+KCl 1007 065 1166 132

+DPG 1051 063 1291 031

KCl + DPG 1124 069 1324 033

n50

Stripped 199 021 235 005

+KCl 253 020 254 006

+DPG 241 009 256 022

KCl + DPG 251 016 258 005

NOTEmdashO2-equilibrium curves were measured at 03 mM heme in 01 M HEPES bufferat pH 74 37 C in the absence (stripped) and presence of added KCl (01 M)andor DPG (DPGHb tetramer ratio 20) as indicated Sample sizes refer to thenumber of individuals analyzed per species

4

Tufts et al doi101093molbevmsu311 MBE at U

niversity of Nebraka-L

incoln Libraries on D

ecember 8 2014

httpmbeoxfordjournalsorg

Dow

nloaded from

online) Consistent with our experiments on the native HbsP50(KCl+DPG) for the highland O princeps rHb was 144 lowerrelative to the lowland O collaris rHb In conjunction with theancestral sequence reconstruction these results indicate thatthe derived increase in O2 affinity of O princeps Hb was at-tributable to the independent or joint effects of the threesubstitutions at sites 5 62 and 126 (fig 4A)

To complement experimental measures of the functionaleffects of each amino acid substitution we used a codon-based maximum-likelihood approach to test for evidence ofpositive selection in an alignment of HBA and HBB sequencesfrom all examined pika species (see Materials and Methods)

In the case of HBB likelihood ratio tests (LRTs) revealedsignificant variation among sites in the ratio of nonsynon-ymous-to-synonymous substitution rates (= dNdS) and abranch-sites test (Zhang et al 2005 Yang and dos Reis 2011)implicated positive selection in driving substitutions at both62 and 126 in O princeps (supplementary table S2Supplementary Material online)

We used site-directed mutagenesis to synthesize the in-ferred ancestral genotype (Anc1) the derived triple-mutantgenotype of O princeps and each of the six possible muta-tional intermediates that connect them (fig 4B) This combi-natorially complete set of three-site genotypes enabled us tomeasure the additive and nonadditive effects of princeps-specific substitutions on each of eight ( = 23) possible geneticbackgrounds Of the six (=3) possible forward trajectoriesthat lead from Anc1 to the derived triple-mutant -globingenotype of O princeps our experiments revealed that only asingle pathway yielded an incremental increase in HbndashO2

affinity at each successive step G5A (first) A62T (second)and A126V (third fig 5) Each of the three possible firstmutational steps produced a significant increase in HbndashO2

affinity as did one of six possible second steps (A126V on theldquoAAArdquo background) and one of three possible final steps(A126V on the ldquoATArdquo background fig 5) However eachmutationrsquos effect on HbndashO2 affinity was highly dependent onthe genetic background in which it occurred as one of sixpossible second steps (G5A) and two of three possible finalsteps (G5A and A62T) yielded significant ldquoreductionsrdquo inHbndashO2 affinity The G5A and A62T substitutions exhibitedsign epistasis for HbndashO2 affinity Both substitutions signifi-cantly increased affinity on the Anc1 background but theyboth produced opposite effects when they occurred on back-grounds in which the other substitutions had already oc-curred (fig 5 supplementary table S1 Supplementary

FIG 4 Inferred history of substitution at five sites that distinguish the HBB genes of the high-altitude Ochotona princeps and the low-altitude O collaris(A) Of the five amino acid substitutions that distinguish the Hbs of the two sister species ancestral sequence reconstruction revealed that threeoccurred on the branch leading to O princeps and the remaining two occurred on the branch leading to O collaris ldquoAnc1rdquo represents the reconstructedHBB sequence of the princepscollaris common ancestor The phylogeny of pika species is taken from (Lanier and Olson 2009) (B) Graphicalrepresentation of sequence space where each vertex of the cube represents a discrete three-site 5ndash62ndash126 genotype and each edge connectsgenotypes separated by a single mutational step The low-affinity ancestral Anc1 and the high-affinity Hb of O princeps are connected by six (=3)possible forward trajectories through sequence space

[DPG] mM

00 05 10 15 20

log

P 50

06

08

10

12

14

O princepsO collaris

FIG 3 Variation in HbndashO2 affinity (as indexed by P50) as a function ofDPG concentration for purified Hbs of Ochotona princeps and O collarisCurves were fit using a double hyperbolic function The Hbs of the twopika species exhibit no discernible difference in DPG binding constantswith an apparent dissociation equilibrium constant for the high-affinityDPG site of approximately 002 mM

5

Epistasis Constrains Pathways of Hemoglobin Evolution doi101093molbevmsu311 MBE at U

niversity of Nebraka-L

incoln Libraries on D

ecember 8 2014

httpmbeoxfordjournalsorg

Dow

nloaded from

Material online) Similar sign-epistatic effects of affinity-alter-ing mutations have been documented in protein engineeringstudies of other vertebrate Hbs (Natarajan et al 2013Projecto-Garcia et al 2013)

In contrast to the sign-epistatic effects of G5A andA62T the A126V substitution had an affinity-enhancingeffect on each of the four possible genetic backgrounds(fig 5) although the magnitude of the affinity-enhancingeffect was highly context-dependent A126V produced a322 reduction in P50(KCl+DPG) when it occurred as the firstsubstitution (on the Anc1 background) compared with a58 reduction when it occurred as the final substitution(on a background in which G5A and A62T had alreadyoccurred) Relative to the Anc1 reference genotype pairwiseepistasis was consistently negative for the free energy of HbndashO2 binding (G) indicating that the combined effects ofaffinity-altering mutations were always less than the sum oftheir individual effects on the Anc1 background (fig 6)

Structural Basis of Mutational Effects

Homology modeling and comparative data from human Hbmutants provided insights into the structural mechanismsresponsible for the individual functional effects of HBB sub-stitutions in the O princeps lineage The G5A substitution ispredicted to stabilize the A-helix of the -chain by reducingbackbone entropy and by reducing the solvent accessibility ofapolar surface (Lopez-Llano et al 2006) This stabilization of-helical conformation can propagate to the 12 and 21

intersubunit contacts that mediate oxygenation-linkedtransitions in quaternary structure thereby perturbing theallosteric equilibrium between the high-affinity (R-state)and low-affinity (T-state) conformations (Dumoulin et al1997 1998 Bellelli et al 2006)

The A62T substitution is predicted to increase the rigidityof the E-helix because the -oxygen atom of 62Thr forms a

polar contact with the 58Ala main-chain oxygen atom(fig 7) This localized stabilization of E-helix secondary struc-ture produces a subtle conformational change in the -chaindistal heme pocket thereby reducing steric hindrance of heme ligand-binding Similarly a nonpolarpolar A62Pmutation at the same site in human Hb (Hb Duarte) alsoproduces an increased O2 affinity while preserving normalcooperativity and DPG sensitivity (Beutler et al 1974Ceccarelli et al 2006) The structural mechanisms responsiblefor the observed interaction effects are not immediately clearas sites 5 62 and 126 are structurally remote from oneanother in the folded -chain subunit (fig 7) indicating thatepistatic interactions do not involve direct steric contact be-tween residue side chains or short-range steric interactionmediated by a third residue or bound ligand Instead theobserved epistatic effects must stem from indirect second-order perturbations possibly involving dynamics of the allo-steric transition in quaternary structure between differentoxygenation states of the Hb tetramer

Pleiotropic Trade-Offs

Analysis of the full set of rHb mutants revealed that HbndashO2

affinity in the presence of allosteric effectors (P50(KCl+DPG)) wasnot significantly correlated with allosteric regulatory capacityas measured by both Cl sensitivity (r =0263 P = 0530)and DPG sensitivity (r =0200 P = 0636) the cooperativityof O2-binding as measured by Hillrsquos cooperativity coefficientn50 (r = 0313 P = 0450) autoxidation rate (r = 0066P = 0877) or structural mobility as measured by a computa-tionally predicted index of energetic frustration (Jenik et al2012) (r =0451 P = 0262) Functional experiments on theengineered rHb mutants revealed sign and magnitude epis-tasis for HbndashO2 affinity but affinity-altering mutations

Expected ΔG (kcalmol) -16 -15 -14 -13 -12 -11 -10 -09

Obs

erve

dΔG

(kc

alm

ol)

-16

-15

-14

-13

-12

-11

-10

-09

single-mutantsdouble-mutantstriple-mutant

Additive

Anc1 (GAA)

AAA

GTA GAV

ATA

AAV GTV

ATV

FIG 6 Relative to the ancestral genotype (Anc1) pairwise epistasis forthe free energy of HbndashO2 binding G was consistently negative (indi-cating negative epistasis for HbndashO2 affinity) G values for the double-and triple mutant genotypes were measured relative to Anc1 and wereplotted against values expected under an additive model ObservedGrsquos for the full set of double- and triple-mutant genotypes fellbelow the y = x diagonal reflecting the fact that the combined effectsof affinity-altering mutations were consistently less than the sum oftheir individual effects on the Anc1 background Error bars denote95 confidence intervals

Mutational step

P50

(KC

l+D

PG

) (to

rr)

7

8

9

10

11

12

1st 2nd 3rd

Anc1 (GAA)

AAA

GTAGAV

ATA

GTVAAV

ATV

G5AA62TA126V

FIG 5 Trajectories of change in HbndashO2 affinity (indexed by P50(KCl+DPG))for each of six forward mutational pathways leading from the ancestralthree-site 5ndash62ndash126 ldquoAnc1rdquo genotype (GAA) to the derived triple-mutant genotype of O princeps (ATV) Only one of six possible forwardpathways yields a monotonic increase in HbndashO2 affinity G5A (1st)A62T (2nd) A126V (3rd) Error bars denote 95 confidence intervals

6

Tufts et al doi101093molbevmsu311 MBE at U

niversity of Nebraka-L

incoln Libraries on D

ecember 8 2014

httpmbeoxfordjournalsorg

Dow

nloaded from

produced no appreciable effects on other structural or func-tional properties that potentially trade-off with O2-affinity

The substitutions that we examined are substitutionsthat actually occurred in the ancestry of modern-day Oprinceps It may be that the pleiotropic and epistatic effectsof substitutions that actually did occur are qualitatively orquantitatively distinct from those of all possible affinity-al-tering substitutions that could have occurred (Draghi andPlotkin 2013 Gong and Bloom 2014) More general conclu-sions about how pleiotropy influences the fixation probabil-ities of mutations can be obtained by comparing thespectrum of substitutions that have actually contributedto evolutionary changes in protein function with the spec-trum of spontaneous mutations that affect protein function(Streisfeld and Rausher 2011)

Evolutionary Implications of Sign Epistasis betweenAffinity-Altering Mutations

Insights into the form and prevalence of intramolecular epis-tasis can illuminate the extent to which realized pathways ofprotein evolution represent deterministic or historically con-tingent outcomes Epistasis between mutant sites in the sameprotein can exert a deterministic influence on molecular evo-lution by constraining the number of selectively accessiblemutational pathways to high-fitness genotypes thereby

enhancing the repeatability (and hence predictability) ofadaptive walks through sequence space (Weinreich et al2005 2006 DePristo et al 2007 Lozovsky et al 2009 daSilva et al 2010 Rokyta et al 2011) However epistasis canalso exert a stochastic influence because if the phenotypiceffects of mutations are conditional on the genetic back-ground in which they occur then the range of selectivelyaccessible mutational pathways will be historically contingenton which particular mutations happen to occur first (Kvitekand Sherlock 2011 Salverda et al 2011 Harms and Thornton2014 Kryazhimskiy et al 2014)

If the derived increase in HbndashO2 affinity in the O princepslineage evolved under the influence of positive directionalselection (as suggested by results of the branch-sites testsupplementary table S2 Supplementary Material online)then our results indicate that any one of the three observed-chain substitutions could have occurred as the first step inthe adaptive walk However as the initial substitution cansignificantly alter the sign andor magnitude of the pheno-typic effect of each subsequent mutation that occurs on thenewly modified background the selection coefficient (andhence fixation probability) of each new mutation will behistorically contingent on which affinity-enhancing mutationhappened to be the first one to fix This role for contingencyalso holds in cases where the epistatically interacting sites aresimultaneously polymorphic

FIG 7 Homology model of pika Hb showing detailed view of the 1 subunit and the locations of three amino acid substitutions (G5A A62T andA126V) that occurred in the Ochotona princeps lineage (A) The proximal histidine 92 (which covalently binds the fifth coordination site of the hemeiron) and the distal histidine 63 (which stabilizes bound ligand at the sixth coordination site on the opposite side of the heme plane) are shown inblue Counting from the N-terminal valine of the -chain polypeptide 5 is the second residue of the A helix 62 is the sixth residue of the E-helix and126 is the fourth residue of the H helix Inset (top right) shows the full Hb tetramer with the focal 1 subunit depicted as a ribbon structure the 2subunit is shown in blue and the 1 and 2 subunits are shown in green (B) The A62T substitution increases the rigidity of the E-helix because the -oxygen atom of 62Thr forms a polar contact with the 58Ala main-chain oxygen atom (C) The methyl side-chain of the ancestral 62Ala forms nocontact with 58Ala resulting in a less stable -helical conformation

7

Epistasis Constrains Pathways of Hemoglobin Evolution doi101093molbevmsu311 MBE at U

niversity of Nebraka-L

incoln Libraries on D

ecember 8 2014

httpmbeoxfordjournalsorg

Dow

nloaded from

ConclusionsThe functional effects of affinity-altering mutations in pika Hbwere highly dependent on the sequential order in which theyoccurred Two of the historical -globin substitutions inO princeps increased or decreased HbndashO2 affinity dependingon whether they occurred as the first step or the last step inthe mutational pathway Regardless of the temporal order inwhich the three -chain substitutions actually occurredduring the evolution of the increased HbndashO2 affinity inO princeps our results revealed no evidence for ineluctablepleiotropic trade-offs that could constrain mutational path-ways of Hb evolution These results suggest that the accessi-bility of alternative mutational pathways may be morestrongly constrained by sign epistasis for positively selectedbiochemical phenotypes than by antagonistic pleiotropy Thecaveat is that this inference is only valid for the circumscribedregion of sequence space that we examined and for the par-ticular biochemical properties that we investigated In con-trast to the deleterious pleiotropic effects documented forspontaneous affinity-altering mutations in human Hb(Bellelli et al 2006 Storz and Moriyama 2008 Varnado et al2013) the mutations that have contributed to evolutionarychanges in HbndashO2 affinity in O princeps produced fine-tunedchanges in O2-affinity with minimal pleiotropic effects Thesefindings are consistent with theory suggesting that function-altering mutations with minimal pleiotropic effects will tendto make disproportionate contributions to phenotypic evo-lution (Otto 2004 Streisfeld and Rausher 2011)

Materials and Methods

Specimen Collection

For the survey of HBA and HBB sequence variation and thefunctional analyses of native Hbs we collected seven O prin-ceps near the summit of Mt Evans Clear Creek Co Colorado(391502400N 1061005400W 4262 m above sea level [asl])and a total of six O collaris from Eagle Summit Yukon-Koyukuk Co Alaska (65290800N 145240000W 949 m asl)and Rainbow Ridge Fairbanks North Star Co Alaska(661803200N 1454002100W 1119 m asl) Specimens ofO princeps were collected under permit 10TR2058 from theColorado Division of Wildlife and were deposited as voucherspecimens in the Harold W Manter Laboratory collection ofthe University of Nebraska State Museum (HWML catalognos NP1195 1201ndash1202) Specimens of O collaris were col-lected under permit 11-057 from the Alaska Department ofFish and Game and were deposited as voucher specimens inthe University of Alaska Museum of the North (UAMNUAM111535 111549 111591 111675 111698 and 111847)Pikas were euthanized in accordance with guidelines ap-proved by the University of Nebraska Institutional AnimalCare and Use Committee (IACUC no 519) Blood was col-lected from each animal through cardiac puncture and redcell fractions were snap-frozen in liquid nitrogen Liver tissuesamples were also collected from each individual as a sourceof genomic DNA Liver samples were immediately frozen inliquid nitrogen and stored at80 C prior to DNA extraction

To conduct a broader survey of intraspecific DNA poly-morphism in the HBA and HBB genes we augmented oursample of wild-caught pikas with frozen tissue samples fromadditional specimens of O princeps and O collaris from theUniversity of California Museum of Vertebrate Zoology (MVZ202371) and UAMN (UAM 100795 102429 102435) To se-quence orthologous HBA and HBB genes in other pika specieswe obtained tissue samples from field-collected specimens ofO daurica (n = 3) from Khan Taishir Mountain Mongolia(461402300N 962103000E 1965 m asl) and O pallasi (n = 1)from Jargalant Khairkhan Mountain Mongolia (474402100N922702400E 1724 m asl) Voucher specimens were exportedunder permit G1215 and were deposited in the HWML col-lection (NK223524ndash223526 223869) We obtained frozentissue samples of O rufescens (MVZ 191920) and O hyper-borea (UAM 84368) and we retrieved O curzoniae HBA andHBB sequences from GenBank (EF429202 and DQ839484respectively)

DNA Sequencing

For each pika specimen genomic DNA was extracted fromliver tissue using the DNeasy Blood and Tissue kit accordingto the manufacturerrsquos protocol (Qiagen Valencia CA) Wepolymerase chain reaction (PCR)-amplified the HBA and HBBgenes of each of the examined pika species using the ExpandHigh Fidelity PCR system (Roche Applied ScienceIndianapolis IN) For the survey of polymorphism in eachof the two focal species we sequenced both alleles of theHBA and HBB genes (for each gene n = 16 and 18 alleles forO princeps and O collaris respectively) Thermal cycler con-ditions were as follows An initial denaturing step for 2 minfollowed by 35 cycles of 94 C for 30 s 55ndash57 C for 30 s 72 Cfor 2 min and final extension of 72 C for 10 min Primers forthe PCR-amplification of the HBA and HBB genes are pro-vided in supplementary table S3 Supplementary Materialonline PCR products were purified and sequenced bidirec-tionally All sequences were deposited in GenBank under thefollowing accession numbers KC757706ndash757708 KC757676ndashKC757683 KC757705 KC757674 KC757675 KC757684KC757685 KC757646 KC757688 KC757689 KC757648ndash757653 KC757690 KC757654 KC757655 KC757691-757696KC757656ndash757661 KC757697ndash757704 KC757662ndash757673KC757686 KC757687 and KC757647

Molecular Evolution Analyses

To make inferences about patterns of molecular evolutionand to reconstruct ancestral sequences we estimated separatephylogenies for the HBA and HBB genes of pikas using therabbit orthologs as outgroup sequences We aligned the nu-cleotide sequences based on their conceptually translatedamino acid sequences using MAFFT (version 7182 Katohand Standley 2013) and we then performed phylogeny recon-structions using maximum likelihood as implemented inTreefinder ver March 2011 (Jobb et al 2004) We evaluatedsupport for the nodes with 1000 bootstrap pseudoreplicatesWe used the ldquopropose modelrdquo tool of Treefinder to select thebest-fitting models of nucleotide substitution for each codon

8

Tufts et al doi101093molbevmsu311 MBE at U

niversity of Nebraka-L

incoln Libraries on D

ecember 8 2014

httpmbeoxfordjournalsorg

Dow

nloaded from

position based on the Akaike information criterion with cor-rection for small sample size The estimated trees were usedfor all downstream analyses

To reconstruct the HBB amino acid sequence of the prin-cepscollaris common ancestor we used a maximum-likeli-hood approach (Yang et al 1995) as implemented in PAML48 (Yang 2007) To test for evidence of positive selection inpika HBA and HBB genes we quantified variation in (=dNdS the ratio of the rate of nonsynonymous substitution pernonsynonymous site [dN] to the rate of synonymous substi-tution per synonymous site [dS] site) using a maximum-like-lihood approach (Goldman and Yang 1994) implemented inthe CODEML program v 48 (Yang 2007) In all cases we usedLRTs to compare nested sets of models (Yang 1998) We firstcompared models that allow to vary among codons in thealignments of HBA and HBB coding sequence (M0 vs M3M1a vs M2a M7 vs M8) The latter two LRTs are tests ofpositive selection as the alternative models (M2a and M8)allow a subset of sites to have 4 1 in contrast to the nullmodels (M1a and M7) that only include site classes withlt 1 These analyses revealed statistically significant varia-tion in among sites in the HBB alignment (supplementarytable S2 Supplementary Material online) so we then imple-mented branch and branch-site analyses The branch-siteanalyses permit inferences about changes in the selectiveregime in a subset of codons between foreground and back-ground branches (Yang and Nielsen 2002 Yang et al 2005Yang and dos Reis 2011) In the alternative hypothesis (ModelA) a fraction of sites in foreground branches are permitted toshift to a regime of positive selection (indicated by 4 1)under the null model in contrast sites are only permitted toshift between neutral evolution and purifying selection Weimplemented three different branch models 1) A single-model where all branches in the tree had the same which corresponds to M0 from the sites analyses 2) a two- model involving a joint estimate of for branches sub-tending the two high-altitude species (O curzionae and Oprinceps) and a separate estimate of for the remainingldquobackgroundrdquo branches and 3) a three-model with separate estimates for O curzionae O princeps and all backgroundbranches We then conducted branch-site analyses of HBBsequences that focused specifically on O princeps We com-pared two sets of models one in which the branch subtend-ing the set of O princeps sequences was set as the foregroundbranch and one in which the entire set of O princeps se-quences (O princeps clade) was set as foreground We usedthe Bayes Empirical Bayes approach to calculate the posteriorprobability that a given site belongs to the site class with 4 1 (ie that the substitution at the site in questionwas driven by positive selection in the foreground branch)(Yang et al 2005)

Native Hb Purification

We used isoelectric focusing (PhastSystem GE HealthcareBio-Sciences Piscataway NJ) to characterize the Hb isoformcomposition of circulating red blood cells in both O princepsand O collaris Native pika Hbs were purified by means of

anion-exchange chromatography (HiTrap QHP GEHealthcare Piscataway NJ) as described previously (Storzet al 2012 Revsbech et al 2013) The column was pre-equil-ibrated with 20 mM Tris (pH 84) and the sample was elutedusing a linear gradient of 0ndash02 M NaCl Samples were de-salted by dialysis against 10 mM HEPES buffer (pH 74) andwere concentrated using Millipore centrifugal filter units (30-kDa cutoff) Hb concentrations were determined from absor-bance spectra using standard extinction coefficients at 577and 540 nm

Vector Construction and Site-Directed Mutagenesis

Nucleotide sequences of pika HBA and HBB genes were op-timized with respect to Escherichia coli codon preferencesand were then synthesized by Eurofins (Huntsville AL)Gene cassettes for the HBA and HBB genes and the methio-nine aminopeptidase (MAP) gene were tandemly cloned intothe custom expression vector (pGM plasmid) described byNatarajan et al (2011 2013) In order to maximize efficiencyin the posttranslational cleaving of N-terminal methionineresidues from the nascent - and -chain polypeptides anadditional copy of the MAP gene was cloned into a coex-pressed plasmid pCO-MAP with a kanamycin resistancegene Site-directed mutagenesis was performed using theQuikChange II XL Site-Directed Mutagenesis kit fromStratagene (La Jolla CA) Each engineered codon change inthe globin cassette was verified by plasmid sequencingSequences for the mutagenic primers are provided in supple-mentary table S3 Supplementary Material online

Expression and Purification of rHb

All rHb mutants were expressed in the JM109 (DE3) E colistrain as described previously (Natarajan et al 2011 2013Projecto-Garcia et al 2013 Cheviron et al 2014) Bacterialcells were subject to dual selection in an LB agar plate con-taining ampicillin and kanamycin which ensures that trans-formed cells receive both the pGM and pCO-MAP plasmidsHigh-level expression of MAP enzyme proved critical for syn-thesizing fully functional rHbs All pika rHb mutants wereexpressed in identical experimental conditions Large-scaleproduction was conducted in batches containing 15ndash20 lof TB medium Cells were grown at 37 C in an orbitalshaker at 200 rpm until the absorbance reached an opticaldensity of 06 at 600 nm Expression of the HBA HBB andMAP genes was induced with 02 mM IPTG and the culturewas supplemented with hemin (50mgml) and glucose (20 gl) The cells were then grown at 28 C for 16 h in an orbitalshaker at 200 rpm The bacterial culture was saturated withCO for 15 min and the cells were harvested by centrifugationand stored at 80 C

As a preparative step for the protein purification the cellswere resuspended in lysis buffer (50 mM Tris 1 mM ethylene-diaminetetraacetic acid [EDTA] 05 mM DTT pH 76) withlysozyme (1 mgg wet cells) and were incubated in an ice bathfor 30 min The bacterial cells were sonicated followed byaddition of polyethyleneimine solution (05ndash1) and centri-fugation at 13000 rpm (4 C) for 45 min The clarified

9

Epistasis Constrains Pathways of Hemoglobin Evolution doi101093molbevmsu311 MBE at U

niversity of Nebraka-L

incoln Libraries on D

ecember 8 2014

httpmbeoxfordjournalsorg

Dow

nloaded from

supernatant was then subjected to overnight dialysis in Trisbuffer (20 mM TrisndashHCl with 05 mM EDTA pH 83) TherHbs were then purified by two-step ion-exchange chroma-tography We used prepacked Q-Sepharose columns (HiTrapQHP 5 ml 17-1154-01 GE Healthcare) equilibrated with Tris-buffer (20 mM TrisndashHCl with 05 mM EDTA pH 83) Thesamples were passed through the column and the rHb solu-tions were eluted against a linear gradient of 0ndash05 M NaClThe eluted samples were desalted by overnight dialysis withphosphate buffer (10 mM sodium phosphate buffer 05 mMEDTA pH 68) Each dialyzed rHb sample was then passedthrough an SP-Sepharose column (HiTrap SPHP 1 ml 17-1151-01 GE Healthcare) with an elution buffer (20 mMsodium phosphate buffer 05 mM EDTA pH 83) Sampleswere concentrated and desalted by overnight dialysis against10 mM HEPES buffer (pH 74) and were stored at 80 Cprior to the measurement of O2-equilibrium curves and au-toxidation rates The purity of each rHb preparation was con-firmed by 20 sodium dodecyl sulfate-polyacrylamide gelelectrophoresis and measures of the absorbance spectra ofoxy deoxy and CO derivatives at 450ndash600 nm

Experimental Measurements of Hb Function

Measures of HbndashO2 affinity were conducted using a modifieddiffusion chamber where changes in absorbance were re-corded in conjunction with stepwise changes in the O2 ten-sion of equilibration gases as described previously (Weber1992 Runck et al 2010 Grispo et al 2012 Storz et al 2012Revsbech et al 2013) Measured O2-equilibrium curves werebased on 4ndash6 saturation steps Values of P50 (partial pressureof O2 [PO2] at half-saturation) and the cooperativity coeffi-cient n50 were calculated by using nonlinear regression to fitthe O2-saturation data to the Hill equation (Y = PO2

n[P50

n + PO2n] where Y is the fractional O2 saturation and n

is the cooperativity coefficient) O2-binding equilibria ofnative and rHb solutions (03 mM heme) were measured at37 C in 01 M HEPES buffer pH 74 in the absence (stripped)and presence of added allosteric effectors (01 M KCl andorDPG at 20-fold molar excess over tetrameric Hb) These arestandard experimental conditions (Imai 1982 Mairbeuroaurl andWeber 2012) that closely approximate intraerythrocytic effec-tor concentrations in vivo 100 mM Cl approximates thenaturally occurring concentration inside mammalian redblood cells (where K+ is the main counter ion as opposedto Na+ in the plasma) and a DPGHb tetramer ratio = 20 fallswithin the physiological range for most eutherian mammals(Bunn 1971 Duhm 1971 Mairbeuroaurl and Weber 2012 Tuftset al 2013) HbndashO2 equilibria at physiological concentrationsof Cl ions and DPG ensure that the in vitro measurementsare relevant to in vivo conditions (Mairbeuroaurl and Weber2012) Possible discrepancies in the measured P50 values fornative Hbs and rHb mutants could stem from the fact thatthe native Hb samples are characterized by some degree ofstructural heterogeneity due to heterozygosity at - andor-globin sites whereas rHbs have an invariant amino acidcomposition For this reason all inferences are based on com-parisons between native Hbs of the two species (table 1) or

comparisons among the rHb mutants (supplementary tableS1 Supplementary Material online)

To characterize the pleiotropic effects of affinity-alteringmutations we measured four properties that can potentiallytrade-off with HbndashO2 affinity Structural mobility cooperativ-ity of O2-binding allosteric regulatory capacity (the sensitivityof HbndashO2 affinity to the inhibitory effects of anionic effectors)and susceptibility to spontaneous heme oxidation (autoxida-tion rate supplementary table S1 Supplementary Materialonline) Although there are numerous potential pleiotropiceffects of amino acid mutations that could be measured wechose to focus on four quantifiable properties that are ofknown importance and physiological relevance (Olson andMaillett 2005 Bonaventura et al 2013 Varnado et al 2013)To predict structural mobility of tetrameric Hb we used thecrystal structure of European hare Hb (Lepus europaeus PDBID 3LQD) to build a structural model of pika Hb using SWISS-MODEL (Arnold et al 2006) derivatives were modeled usingMODELLER 9V11 (Sali and Blundell 1993) Molecular interac-tions were identified by PISA (Krissinel and Henrick 2007)and molecular graphics were generated using PyMol(Schreuroodinger LLC) Conformational stress and ligand bindingenergy were computed using the Frustratometer (Jenik et al2012) and AutoDock (Morris et al 2009) programs As ameasure of cooperativity we used Hillrsquos coefficient (n50) inthe presence of Cl and DPG As a measure of allostericregulatory capacity we measured the difference in log-trans-formed P50 values for stripped Hb in the presence and ab-sence of KCl and DPG Following standard protocols forvertebrate Hbs (Jensen 2001) we measured autoxidationrates of purified rHbs at 10mM concentration in 50 mMTris buffer pH 74 at 37 C using an Agilent 8453 UV-VisibleSpectrophotometer (Agilent Technologies Santa Clara CA)Absorbance was recorded every hour at 560 576 and 630 nmover a 14-h period For each experiment measurements ofautoxidation rates were derived from the average of absor-bance traces from the three wavelengths

Supplementary MaterialSupplementary tables S1ndashS3 are available at Molecular Biologyand Evolution online (httpwwwmbeoxfordjournalsorg)

Acknowledgments

The authors thank LE Olson AM Gunderson SL GardnerK Roccaforte ZA Cheviron and AM Runck for assistance inthe field K G McCracken L E Olson and A M Gundersonfor logistical support in Alaska and HC Lanier for helpfuldiscussions They are also grateful to four anonymous re-viewers for insightful comments on the manuscript This re-search was funded by grants from the National Institute ofHealth (HL087216) to JFS the American Society ofMammalogists to DMT the School of Biological SciencesUniversity of Nebraska to DMT the Carlsberg Foundation(2013-01-0178) to AF and the Faculty of Science andTechnology Aarhus University to AF and REW

10

Tufts et al doi101093molbevmsu311 MBE at U

niversity of Nebraka-L

incoln Libraries on D

ecember 8 2014

httpmbeoxfordjournalsorg

Dow

nloaded from

ReferencesArnold K Bordoli L Kopp J Schwede T 2006 The SWISS-MODEL work-

space a web-based environment for protein structure homologymodelling Bioinformatics 22195ndash201

Baldwin J Chothia C 1979 Haemoglobin the structural changes relatedto ligand binding and its allosteric mechanism J Mol Biol 129175ndash220

Bellelli A Brunori M Miele AE Panetta G Vallone B 2006 The allostericproperties of hemoglobin insights from natural and site directedmutants Curr Protein Pept Sci 717ndash45

Bettati S Mozzarelli A Perutz MF 1983 Allosteric mechanism ofhemoglobin rupture of salt bridges raises oxygen affinity of theT-structure J Mol Biol 281581ndash585

Beutler E Lang A Lehmann H 1974 Hemoglobin Duarte(22

62(E6)AlaPro) a new unstable hemoglobin with increasedoxygen affinity Blood 43527ndash535

Bloom JD Labthavikul ST Otey CR Arnold FH 2006 Protein stabilitypromotes evolvability Proc Natl Acad Sci U S A 1035869ndash5874

Bonaventura C Henkens R Alayash AI Banerjee S Crumbliss AL 2013Molecular controls of the oxygenation and redox reactions of he-moglobin Antioxid Redox Signal 182298ndash2313

Bridgham JT Ortlund EA Thornton JW 2009 An epistatic ratchet con-strains the direction of glucocorticoid receptor evolution Nature461515ndash519

Bunn HF 1971 Differences in interaction of 23-diphosphoglycerate withcertain mammalian hemoglobins Science 1721049ndash1050

Carneiro M Hartl DL 2010 Adaptive landscapes and protein evolutionProc Natl Acad Sci U S A 1071747ndash1751

Ceccarelli M Ruggerone P Anedda R Fais A Era B Sollaino MC CordaM Casu M 2006 Structure-function relationship in a variant he-moglobin a combined computational-experimental approachBiophys J 913529ndash3541

Cheviron ZA Natarajan C Projecto-Garcia J Eddy DK Jones J CarlingMD Witt CC Moriyama H Weber RE Fago A et al 2014Integrating evolutionary and functional tests of adaptive hypothe-ses a case study of altitudinal differentiation in hemoglobin functionin an Andean sparrow Zonotrichia capensis Mol Biol Evol 312948ndash2962

da Silva J Coetzer M Nedellec R Pastore C Mosier DE 2010 Fitnessepistasis and constraints on adaptation in a human immunodefi-ciency virus type 1 protein region Genetics 185293ndash303

de Visser JAGM Krug J 2014 Empirical fitness landscapes and the pre-dictability of evolution Nat Rev Genet 15480ndash490

DePristo MA Hartl DL Weinreich DM 2007 Mutational reversionsduring adaptive protein evolution Mol Biol Evol 241608ndash1610

DePristo MA Weinreich DM Hartl DL 2005 Missense meanderings insequence space a biophysical view of protein evolution Nat RevGenet 6678ndash687

Draghi JA Plotkin JB 2013 Selection biases the prevalence and type ofepistasis along adaptive trajectories Evolution 673120ndash3131

Duhm J 1971 Effects of 23-diphosphoglycerate and other organic phos-phate compounds on oxygen affinity and intracellular pH of humanerythrocytes Pflugers Arch 326341

Dumoulin A Manning LR Jenkins WT Winslow RM Manning JM 1997Exchange of subunit interfaces between recombinant adult and fetalhemoglobinsmdashevidence for a functional inter-relationship amongregions of the tetramer J Biol Chem 27231326ndash31332

Dumoulin A Padovan JC Manning LR Popowicz A Winslow RM ChaitBT Manning JM 1998 The N-terminal sequence affects distant helixinteractions in hemoglobinmdashimplications for mutant proteins fromstudies on recombinant hemoglobin felix J Biol Chem 27335032ndash35038

Erbajeva MA 1994 Phylogeny and evolution of Ochotonidae with em-phasis on Asian ochotonids In Tomida Y Li CK Setoguchi Teditors Rodent and lagomorph families of Asian origins and diver-sification Proceedings of Workshop WC-2 29th InternationalGeological Congress Kyoto Japan Tokyo National ScienceMuseum Monographs p 1ndash13

Galbreath KE Hafner DJ Zamudio KR 2009 When cold is better cli-mate-driven elevation shifts yield complex patterns of diversificationand demography in an alpine specialist (American pika Ochotonaprinceps) Evolution 632848ndash2863

Galbreath KE Hafner DJ Zamudio KR Agnew K 2010 Isolation andintrogression in the Intermountain West contrasting gene geneal-ogies reveal the complex biogeographic history of the American pika(Ochotona princeps) J Biogeogr 37344ndash362

Goldman N Yang ZH 1994 Codon-based model of nucleotide substi-tution for protein-coding DNA sequences Mol Biol Evol 11725ndash736

Gong LI Bloom JD 2014 Epistatically interacting substitutions are en-riched during adaptive protein evolution PLoS Genet 10e1004328

Grispo MT Natarajan C Projecto-Garcia J Moriyama H Weber RE StorzJF 2012 Gene duplication and the evolution of hemoglobin isoformdifferentiation in birds J Biol Chem 28712

Hafner DJ 1993 North American pika (Ochotona princeps) as a LateQuaternary biogeographic indicator species Quat Res 39373ndash380

Hafner DJ 1994 Pikas and permafrost post-Wisconsin historical zooge-ography of Ochotona in the Southern Rocky Mountains USA ArctAlp Res 26375ndash382

Harms MJ Thornton JW 2014 Historical contingency and itsbiophysical basis in glucocorticoid receptor evolution Nature 512203ndash207

Imai K 1982 Allosteric effects in haemoglobin Cambridge CambridgeUniversity Press

Jenik M Gonzalo Parra R Radusky LG Turjanski A Wolynes PG FerreiroDU 2012 Protein frustratometer a tool to localize energetic frus-tration in protein molecules Nucleic Acids Res 40W348ndashW351

Jensen FB 2001 Comparative analysis of autoxidation of haemoglobin JExp Biol 2042029ndash2033

Jobb G von Haeseler A Strimmer K 2004 TREEFINDER a powerfulgraphical analysis environment for molecular phylogenetics BMCEvol Biol 418

Katoh K Standley DM 2013 MAFFT multiple sequence alignment soft-ware version 7 improvements in performance and usability Mol BiolEvol 30772ndash780

Krissinel E Henrick K 2007 Inference of macromolecular assembliesfrom crystalline state J Mol Biol 372774ndash797

Kryazhimskiy S Dushoff J Bazykin GA Plotkin JB 2011 Prevalence ofepistasis in the evolution of influenza A surface proteins PLoS Genet7e1001301

Kryazhimskiy S Rice DP Jerison ER Desai MM 2014 Global epistasismakes adaptation predictable despite sequence-level stochasticityScience 3441519ndash1522

Kvitek DJ Sherlock G 2011 Reciprocal sign epistasis between frequentlyexperimentally evolved adaptive mutations causes a rugged fitnesslandscape PLoS Genet 7e1002056

Lanier HC Olson LE 2009 Inferring divergence times within pikas(Ochotona spp) using mtDNA and relaxed molecular dating tech-niques Mol Phylogenet Evol 531ndash12

Lanier HC Olson LE 2013 Deep barriers shallow divergences reducedphylogeographical structure in the collared pika (MammaliaLagomorpha Ochotona collaris) J Biogeogr 40466ndash478

Lopez-Llano J Campos LA Sancho J 2006 -helix stabilization by ala-nine relative to glycine roles of polar and apolar solvent exposuresand of backbone entropy Proteins 64769ndash778

Lozovsky ER Chookajorn T Brown KM Imwong M Shaw PJKamchonwongpaisan S Neafsey DE Weinreich DM Hartl DL2009 Stepwise acquisition of pyrimethamine resistance in the ma-laria parasite Proc Natl Acad Sci U S A 10612025ndash12030

Mairbeuroaurl H 1994 Red blood cell function in hypoxia at altitude andexercise Int J Sports Med 1551ndash63

Mairbeuroaurl H Weber RE 2012 Oxygen transport by hemoglobin ComprPhysiol 21463ndash1489

Mead JI 1987 Quaternary records of the pika Ochotona in NorthAmerica Boreas 16165ndash171

Morris GM Huey RB Lindstrom W Sanner MF Belew RK Goodsell DSOlson AJ 2009 Autodock4 and AutoDockTools4 automated dock-ing with selective receptor flexibility J Comput Chem 162785ndash2791

11

Epistasis Constrains Pathways of Hemoglobin Evolution doi101093molbevmsu311 MBE at U

niversity of Nebraka-L

incoln Libraries on D

ecember 8 2014

httpmbeoxfordjournalsorg

Dow

nloaded from

Natarajan C Inoguchi N Weber RE Fago A Moriyama H Storz JF 2013Epistasis among adaptive mutations in deer mouse hemoglobinScience 3401324ndash1327

Natarajan C Jiang X Fago A Weber RE Moriyama H Storz JF 2011Expression and purification of recombinant hemoglobin inEscherichia coli PLoS One 6e20176

Olson JS Maillett DH 2005 Designing recombinant hemoglobin for useas a blood substitute In Winslow RM editor Blood substitutes SanDiego (CA) Academic Press p 354ndash374

Ortlund EA Bridgham JT Redinbo MR Thornton JW 2007 Crystalstructure of an ancient protein evolution by conformational epis-tasis Science 3171544ndash1548

Ostman B Hintze A Adami C 2012 Impact of epistasis and pleiotropyon evolutionary adaptation Proc R Soc Lond B Biol Sci 279247ndash256

Otto SP 2004 Two steps forward one step back the pleiotropic effectsof favoured alleles Proc R Soc Lond B Biol Sci 271705ndash714

Perutz MF 1972 Nature of haem-haem interaction Nature 237495ndash499

Perutz MF 1989 Mechanisms of cooperative and allosteric regulation inproteins Q Rev Biophys 22139ndash237

Poelwijk FJ Kiviet DJ Weinreich DM Tans SJ 2007 Empirical fitnesslandscapes reveal accessible evolutionary paths Nature 445383ndash386

Projecto-Garcia J Natarajan C Moriyama H Weber RE Fago A ChevironZA Dudley R McGuire JA Witt CC Storz JF 2013 Repeated eleva-tional transitions in hemoglobin function during the evolution ofAndean hummingbirds Proc Natl Acad Sci U S A 11020669ndash20674

Revsbech IG Tufts DM Projecto-Garcia J Moriyama H Weber RE StorzJF Fago A 2013 Hemoglobin function and allosteric regulation insemi-fossorial rodents (family Sciuridae) with different altitudinalranges J Exp Biol 2164264ndash4271

Rokyta DR Joyce P Caudle SB Miller C Beisel CJ Wichman HA 2011Epistasis between beneficial mutations and the phenotype-to-fit-ness map for a ssDNA virus PLoS Genet 7e1002075

Runck AM Weber RE Fago A Storz JF 2010 Evolutionary and func-tional properties of a two-locus b-globin polymorphism in Indianhouse mice Genetics 1841121ndash1131

Sali A Blundell TL 1993 Comparative protein modeling by satisfactionof spatial restraints J Mol Biol 234779ndash815

Salverda MLM Dellus E Gorter FA Debets AJM van der Oost JHoekstra RF Tawfik DS de Visser JAGM 2011 Initial mutationsdirect alternative pathways of protein evolution PLoS Genet 7e1001321

Smith AT Weston ML 1990 Ochotona princeps Mamm Species 3521ndash8

Storz JF 2007 Hemoglobin function and physiological adaptation tohypoxia in high-altitude mammals J Mammal 8824ndash31

Storz JF Moriyama H 2008 Mechanisms of hemoglobin adaptation tohigh altitude hypoxia High Alt Med Biol 9148ndash157

Storz JF Runck AM Moriyama H Weber RE Fago A 2010 Geneticdifferences in hemoglobin function between highland and lowlanddeer mice J Exp Biol 2132565ndash2574

Storz JF Runck AM Sabatino SJ Kelly JK Ferrand N Moriyama H WeberRE Fago A 2009 Evolutionary and functional insights into themechanism underlying high-altitude adaptation of deer mouse he-moglobin Proc Natl Acad Sci U S A 10614450ndash14455

Storz JF Scott GR Cheviron ZA 2010 Phenotypic plasticity and geneticadaptation to high-altitude hypoxia in vertebrates J Exp Biol 2134125ndash4136

Storz JF Weber RE Fago A 2012 Oxygenation properties and oxidationrates of mouse hemoglobins that differ in reactive cysteine contentComp Biochem Physiol A Mol Integr Physiol 161265ndash270

Streisfeld MA Rausher MD 2011 Population genetics pleiotropy andthe preferential fixation of mutations during adaptive evolutionEvolution 65629ndash642

Taverna DM Goldstein RA 2002 Why are proteins marginally stableProteins 46105ndash109

Tokuriki N Stricher F Serrano L Tawfik DS 2008 How protein stabilityand new functions trade off PLoS Comput Biol 4e1000002

Tokuriki N Tawfik DS 2009 Stability effects of mutations and proteinevolvability Curr Opin Struct Biol 19596ndash604

Tufts DM Revsbech IG Cheviron ZA Weber RE Fago A Storz JF 2013Phenotypic plasticity in blood-oxygen transport in highland andlowland deer mice J Exp Biol 2161167ndash1173

Turek Z Kreuzer F 1976 Effect of a shift of the oxygen dissociation curveon myocardial oxygenation at hypoxia Adv Exp Med Biol 75657ndash662

Turek Z Kreuzer F Ringnalda BEM 1978 Blood gases at several levels ofoxygenation in rats with a left-shifted blood oxygen dissociationcurve Pfleurougers Arch 3767ndash13

Turek Z Kreuzer F Turek-Maischeider M Ringnalda BEM 1978 Bloodoxygen content cardiac output and flow to organs at several levelsof oxygenation in rats with a left-shifted blood oxygen dissociationcurve Pfleurougers Arch 376201ndash207

Varnado CL Mollan TL Birukou I Smith BJZ Henderson DP Olson JS2013 Development of recombinant hemoglobin-based oxygen car-riers Antioxid Redox Signal 182314ndash2328

Wang XJ Minasov G Shoichet BK 2002 Evolution of an antibioticresistance enzyme constrained by stability and activity trade-offs JMol Biol 32085ndash95

Weber RE 1992 Use of ionic and zwitterionic (tris bistris and HEPES)buffers in studies on hemoglobin function J Appl Physiol 721611ndash1615

Weber RE 2007 High-altitude adaptations in vertebrate hemoglobinsRespir Physiol Neurobiol 158132ndash142

Weinreich DM Delaney NF DePristo MA Hartl DL 2006 Darwinianevolution can follow only very few mutational paths to fitter pro-teins Science 312111ndash114

Weinreich DM Knies JL 2013 Fisherrsquos geometric model of adaptationmeets the functional synthesis data on pairwise epistasis for fitnessyields insights into the shape and size of phenotype space Evolution672957ndash2972

Weinreich DM Watson RA Chao L 2005 Sign epistasis andgenetic constraint on evolutionary trajectories Evolution 591165ndash1174

Yang Z 2007 PAML 4 Phylogenetic analysis by maximum likelihoodMol Biol Evol 241586ndash1591

Yang Z dos Reis M 2011 Statistical properties of the branch-site test ofpositive selection Mol Biol Evol 281217ndash1228

Yang ZH 1998 Likelihood ratio tests for detecting positive selection andapplication to primate lysozyme evolution Mol Biol Evol 15568ndash573

Yang ZH Kumar S Nei M 1995 A new method of inference of ancestralnucleotide and amino-acid sequences Genetics 1411641ndash1650

Yang ZH Nielsen R 2002 Codon-substitution models for detectingmolecular adaptation at individual sites along specific lineagesMol Biol Evol 19908ndash917

Yang ZH Wong WSW Nielsen R 2005 Bayes empirical Bayes inferenceof amino acid sites under positive selection Mol Biol Evol 221107ndash1118

Zhang JZ Nielsen R Yang ZH 2005 Evaluation of an improved branch-site likelihood method for detecting positive selection at the mo-lecular level Mol Biol Evol 222472ndash2479

12

Tufts et al doi101093molbevmsu311 MBE at U

niversity of Nebraka-L

incoln Libraries on D

ecember 8 2014

httpmbeoxfordjournalsorg

Dow

nloaded from