doi:10.1016/j.jmb.2011.05.011 J. Mol. Biol. (2011) 410, 280–293

Contents lists available at www.sciencedirect.com

Journal of Molecular Biologyj ourna l homepage: ht tp : / /ees .e lsev ie r.com. jmb

Mitochondrial Aminoacyl-tRNA SynthetaseSingle-Nucleotide Polymorphisms That Lead to Defectsin Refolding but Not Aminoacylation

Rajat Banerjee1⁎, Noah M. Reynolds1, Srujana S. Yadavalli1,Cory Rice1, Hervé Roy1, Papri Banerjee2, Rebecca W. Alexander2

and Michael Ibba1,3,4

1Department of Microbiology, Ohio State University, Columbus, OH 43210-1292, USA2Department of Chemistry, Wake Forest University, Winston-Salem, NC 27109-7486, USA3Ohio State Biochemistry Program, Ohio State University, Columbus, OH 43210-1292, USA4Center for RNA Biology, Ohio State University, Columbus, OH 43210-1292, USA

Received 20 February 2011;received in revised form5 May 2011;accepted 6 May 2011Available online13 May 2011

Edited by A. Pyle

Keywords:aminoacyl-tRNA synthetase;folding;mitochondria;translation;tRNA

*Corresponding author. DepartmenBiotechnology, University of Calcuttrbbcgc@gmail.com.Abbreviations used: aaRS, amino

hmtPheRS, human mitochondrial phuman mitochondrial leucyl-tRNA

0022-2836/$ - see front matter © 2011 E

Defects in organellar translation are the underlying cause of a number ofmitochondrial diseases, including diabetes, deafness, encephalopathy, andother mitochondrial myopathies. Themost common causes of these diseasesare mutations in mitochondria-encoded tRNAs. It has recently becomeapparent that mutations in nuclear-encoded components of the mitochon-drial translation machinery, such as aminoacyl-tRNA synthetases (aaRSs),can also lead to disease. In some cases, mutations can be directly linked tolosses in enzymatic activity; however, for many, their effect is unknown. Toinvestigate how aaRS mutations impact function without changingenzymatic activity, we chose nonsynonymous single-nucleotide polymor-phisms (nsSNPs) that encode residues distal from the active site of humanmitochondrial phenylalanyl-tRNA synthetase. The phenylalanyl-tRNAsynthetase variants S57C and N280S both displayed wild-type aminoacyla-tion activity and stability with respect to their free energies of unfolding, butwere less stable at low pH. Mitochondrial proteins undergo partialunfolding/refolding during import, and both S57C and N280S variantsretained less activity than wild type after refolding, consistent with theirreduced stability at low pH. To examine possible defects in protein folding inother aaRS nsSNPs, we compared the refolding of the human mitochondrialleucyl-tRNA synthetase variant H324Q to that of wild type. The H324Qvariant had normal activity prior to unfolding, but displayed a refoldingdefect resulting in reduced aminoacylation compared to wild type afterrenaturation. These data show that nsSNPs can impact mitochondrialtranslation by changing a biophysical property of a protein (in this caserefolding) without affecting the corresponding enzymatic activity.

© 2011 Elsevier Ltd. All rights reserved.

t of Biotechnology and Dr. B. C. Guha Center for Genetic Engineering anda, 35 Ballygunge Circular Road, Kolkata 700 019, West Bengal, India. E-mail address:

acyl-tRNA synthetase; nsSNP, nonsynonymous single-nucleotide polymorphism;henylalanyl-tRNA synthetase; ANS, 1,8-anilino naphthyl sulfonate; hmtLeuRS,synthetase; PDB, Protein Data Bank; RMSF, root-mean-square fluctuation.

lsevier Ltd. All rights reserved.

281SNPs That Lead to Defects in Refolding

Introduction

In eukaryotes, ATP synthesis by oxidative phos-phorylation occurs in mitochondria. Mitochondriamaintain a small genome separate from that foundin the nucleus, as exemplified by the humanmitochondrial genome that encodes 13 proteins, 2rRNAs, and 22 tRNAs. Polypeptides encoded by thehuman mitochondrial genome are subunits ofrespiratory chain complexes and are essential formitochondrial function. To maintain viability, mito-chondria must import numerous components oftheir translational machinery from the cytosol,including proteins and some RNAs.1–3 Numerousproteins are imported into mitochondria, includingnuclear-encoded organelle-specific aminoacyl-tRNA synthetases (aaRSs4), whose role is to correct-ly pair amino acids with their cognate tRNAs duringtranslation. Mitochondrial aaRSs vary in structure;some correspond to splice variants of their cytosoliccounterparts,5 while others are of bacterial ancestryand are only distantly related to the correspondingcytosolic enzymes.6,7

Mitochondria are responsible for ATP synthesisduring aerobic respiration, and different mutationsthat compromise this essential function have beenlinked to a wide range of human diseases, includingMELAS (mitochondrial myopathy, encephalopathy,lactic acidosis, and stroke-like episodes), MERRF(myoclonic epilepsywith ragged red fibers), diabetes,and deafness.8 The comparative ease with which thehumanmitochondrial genome can be sequenced hasallowed the identification and characterization ofnumerous mutations that can be directly linked toparticular diseases. The most commonly reportedclass of mitochondrial genome mutations associatedwith diseases is that found in genes encodingtRNAs.9 Mutations in tRNAs can compromisemitochondrial translation in a number of ways,10,11

either by directly impairing protein synthesis8,12–15

or by disrupting biogenesis or folding of tRNA.16–19

Recent reports have shown that mitochondrialaminoacylation of tRNA may also be impaired insome diseases as a result of mutations in nucleargenes that encode mitochondrially targeted aaRSs.For mitochondrial aspartyl-tRNA synthetase(encoded by DARS2) and arginyl-tRNA synthetase(RARS2), loss of exons, premature termination, andmissense mutations lead to aaRS variants withsevere reductions in aminoacylation activity, leadingto diseases such as leukoencephalopathy20 andhypoplasia.21 Another class of DARS2 pathogenicmutations linked to leukoencephalopathy, where asingle amino acid substitution does not impairaminoacylation but instead prevents mitochondrialimport of aspartyl-tRNA synthetase, was alsorecently reported.22 In other instances, the molecularbasis by which changes to an organelle aaRS lead to

mitochondrial dysfunction are unclear as, for exam-ple, in the case of a mutation in the LARS2 gene(encoding leucyl-tRNA synthetase), which may playa role in susceptibility to type 2 diabetes.23,24

Most mitochondrial matrix proteins, such as theaaRSs, are synthesized in the cytosol with an N-terminal targeting sequence that facilitates import.During their subsequent import across the mito-chondrial membranes, cytosolically synthesizedmatrix proteins are either unfolded or partiallyunfolded, so that they adopt a molten-globule-likeconformation.25–29 Following import, mitochondrialproteins such as aaRSs must then be refoldedcorrectly within the organelle matrix to ensureproper function. Defects in the refolding of matrixproteins could be expected to lead to mitochondrialdysfunction, perhaps by triggering an unfoldedprotein response or by limiting levels of stablefolded active protein.30 One example of the latterhas been reported for a common mutant ofmitochondrial medium-chain acyl-CoA dehydroge-nase found in patients deficient in mitochondrialmedium-chain acyl-CoA dehydrogenase, where asingle amino acid substitution decreases proteinfolding and assembly in mitochondria.31

Despite the growing number of disease-relatednonsynonymous single-nucleotide polymorphisms(nsSNPs) in both cytosolic and mitochondrial aaRS-encoding genes,32 in most cases, the molecular basisunderlying loss of activity is still unknown.33

Nonsynonymous changes account for approximate-ly half of the characterized genetic diseases inhumans and are documented by two databasescontaining disease-causing variants: Online Mende-lian Inheritance in Man34 and the Human GeneMutation Database.35 The human population isestimated to have 67,000–200,000 nsSNPs, and eachperson is thought to be heterozygous for 24,000–40,000 nsSNPs.36 nsSNPs that affect aaRS activity,alter splice sites, and limit subcellular localizationhave each been described, but whether there are alsovariants that specifically impact stability and foldingis unknown. To investigate the possible effects ofmutations on the folding of mitochondrial aaRSs, westudied nsSNPs of human mitochondrial phenylala-nyl-tRNA synthetase (hmtPheRS), a monomericenzyme for which a high-resolution crystal structureis available.37 Using the National Center for Biotech-nology Information single-nucleotide polymor-phism database, we selected two nsSNPs forFARS2 (encoding hmtPheRS), both of which corre-spond to amino acid changes distal to the catalyticand tRNA binding sites of the enzyme. ThehmtPheRS nsSNP variants and an nsSNP of LARS2all retained wild-type aminoacylation activity butdisplayed reduced stability and refolding defects,suggesting a mechanism by which aaRS mutationsthat do not directly impact enzymatic activity canstill lead to loss of function.

282 SNPs That Lead to Defects in Refolding

Results

Free energy of unfolding of two naturallyoccurring variants of hmtPheRS

While many disease-causing nsSNPs that lead tochanges in the active site, editing site, oligomerinterface, or subcellular localization of aaRSs havebeen described, it is not known if nsSNPs that do not

target these properties could also impact function.Several nsSNPs of hmtPheRS have been documen-ted (Fig. 1a): three correspond to amino acidreplacements in the mitochondrial targeting se-quence (G3S, S4P, and L6P), two are in the tRNAanticodon binding domain (H408Q and T410P), andone is in the catalytic domain (T246M), while twoothers are located outside any of these key func-tional regions of the protein (S57C and N280S).Neither S57C nor N280S would be expected toimpact mitochondrial targeting or enzymatic activ-ity, making them suitable candidates for investigat-ing the potential effect of nsSNPs that map outsideknown functional regions. While clinical associationwith these nsSNPs is still unknown, the softwareSIFT38 and Polyphen39 both predict that S57C mightbe deleterious to function, while N280S substitutionis expected to be tolerated. Several recent reportsindicated that multiple mutations, particularlydouble nsSNP mutants in a single gene, can causesevere disease phenotypes,40–43 prompting us toinvestigate the S57C/N280S double mutant inparallel with the corresponding single mutants.Ser57 and Asn280 map to positions away from thecatalytic center and the anticodon binding domainof hmtPheRS, and neither variant showed signifi-cant loss in secondary structure or aminoacylationactivity in vitro compared to wild type (Fig. 1). Ureadenaturation profiles were determined for eachvariant and for the corresponding double mutant,and then compared to wild-type hmtPheRS, toascertain if these amino acid replacements lead tochanges in stability. All the protein samples wereincubated for 18–24 h at room temperature to ensurea complete thermodynamic equilibrium. Thechanges in the emission maxima of tryptophanfluorescence progressively shifted towards longerwavelengths, indicating solvent exposure of trypto-phans due to unfolding of native protein structure(Fig. 2a). Calculation of the free energies of unfold-ing based on urea denaturation revealed onlymarginal differences for the nsSNP variants

Fig. 1. Structure and activity of hmtPheRS. (a) Crystalstructure of hmtPheRS.37 The phenylalanyl adenylatebound in the active site of the enzyme, and the aminoacid substitutions associated with human nsSNPs areshown. (b) Aminoacylation of tRNA by hmtPheRS. Thereactions were performed using 3 μM E. coli tRNAPhe

transcribed in vitro and 200 nM hmtPheRS variants. Datapoints are an average of three replicates, with error barsrepresenting 1 SD. (c) Far-UV CD spectra of hmtPheRSvariants. The protein concentration was 5 μM, preparedfreshly in 50 mM Tris–HCl buffer (pH 7.5) containing 5%glycerol. The data were taken from 200 nm to 250 nmwitha step size of 1 nm in a 1-mm path-length cuvette, with1 nm of bandwidth and 5 s of averaging time. Five scanswere taken and averaged for each sample. Protein-onlyspectra were obtained by subtracting the CD signal fromthe corresponding buffer.

283SNPs That Lead to Defects in Refolding

compared to wild type, while the double mutantwas significantly more stable (Table 1). Ureadenaturation profiles were also measured in thepresence of 1,8-anilino naphthyl sulfonate (ANS),

(a)

0 2 4 6 8 100

0.2

0.4

0.6

0.8

1

1.2

[UREA] M

Fra

ctio

n u

nfo

lded

N280SS57CWT

S57C/N280S

(b)

0 2 4 6 8 100

1

2

3

4

[Urea] M

No

rm A

NS

Flu

o (

482

nm

) WT S57C N280S S57C/N280S

(c)

260 280 300 320 340-5

-4

-3

-2

-1

0

1

2

Wavelength ( in nm)

θ θ (in

mill

ideg

ree)

0 M Urea3 M Urea9.9 M Urea

(d)

215 220 225 230 235 240 245 250

-15

-10

-5

0

Wavelength ( in nm)

θ θ (in

mill

ideg

ree)

0 M Urea3 M Urea9.9 M Urea

which exhibits enhanced fluorescence upon bindingto exposed hydrophobic patches, to investigate ifany stable intermediates accumulated during pro-tein folding/unfolding.44 A dramatic increase inANS fluorescence was observed at around 3 M ureaand remained high up to 5 M urea, and then leveledoff completely around 9M urea, indicating completeunfolding (Fig. 2b). The ANS fluorescence increasearound 3 M urea indicated the potential formationof a “molten-globule”-like intermediate. In additionto the observed retention of secondary structureduring unfolding, molten globule states are charac-terized by loss of tertiary interactions, as observedby near-UV CD (Fig. 2c). At 3 M urea, where far-UVprobing showed marginal changes in secondarystructure (Fig. 2d), the side-chain CD signal isreduced by at least 50%, indicating loss of tertiaryinteractions for large parts of hmtPheRS (Fig. 2c).The S57C, N280S, and S57C/N280S varianthmtPheRSs all displayed a behavior nearly identicalto that of wild type during unfolding (data notshown), indicating that wild-type hmtPheRS and itsvariants all form stable molten-globule-like inter-mediates at around 3 M urea.

hmtPheRS variants with folding defects atlow pH

Mitochondrial import requires proteins to crossthe acidic intermembrane space, which can result in

Fig. 2. Unfolding of hmtPheRS. (a) Equilibrium ureaunfolding of hmtPheRS variants. The tryptophan emissionmaxima data obtained after taking the first derivative oftryptophan emission spectra using a Fluorolog-3 spectro-fluorometer (Horiba Jobin Yvon) with an integration timeof 1 s and equipped with a constant-temperature cellholder. The excitation wavelength was 295 nm, andemission was recorded from 310 nm to 450 nm. Theexcitation and emission slit widths were both 5 nm. Theprotein concentration was 2.5 μM. All the data wererepeated three times, and standard deviations weremeasured. (b) ANS binding of hmtPheRS variants as afunction of urea concentration. The protein concentrationwas 2.5 μM. The protein samples were incubatedovernight, and then a 25 molar excess of ANS wasadded. The protein samples were kept at 25 °C in the darkfor 1 h before measurements were taken. Data are theaverage of three measurements. (c) Near-UV CD spectra ofhmtPheRS. The data were taken after the incubation of theprotein samples for 4–5 h at 25 °C. The proteinconcentration was 10 μM, and the data were recordedfrom 250 nm to 340 nm with an Aviv 62A DS spectro-polarimeter (Aviv). The protein samples were scannedthree times and averaged. (d) Far-UV CD spectra ofhmtPheRS. The data were taken after the incubation of theprotein samples for 24 h at 25 °C. The protein concentra-tion was 5 μM, and the data were recorded from 215 nm to250 nm with an Aviv 62A DS spectropolarimeter (Aviv).The protein samples were scanned three times andaveraged.

Table 1. Unfolding parameters of hmtPheRS and itsvariants

hmtPheRSvariants

ΔGU–N(kcal mol−1)

mU–N(kcal mol−1 M−1)

Transitionmidpoint Cm (M)

Wild type −3.77±0.1 −0.61±0.1 6.2±0.1S57C −3.34±0.1 −0.57±0.1 5.9±0.1N280S −3.26±0.1 −0.57±0.1 5.7±0.1S57C/N280S −6.80±0.2 −0.86±0.2 7.9±0.2

Urea-induced chemical unfolding was performed with increasingconcentrations of the denaturant. The tryptophan emissionmaximavalue was recorded as a function of urea concentration. The datawere normalized and analyzed according to Eqs. (1) and (2).

Fig. 3. Change in the conformation of hmtPheRS as afunction of pH. (a) pH-induced changes in tryptophanemission maxima. The data were recorded with aFluorolog-3 spectrofluorometer (Horiba Jobin Yvon)using the same procedure described in Fig. 2a. (b) ANSbinding as a function of pH. The procedure used is asdescribed in Fig. 2b. (c) Ellipticity change at 222 nm ofhmtPheRS as a function of pH. The protein concentrationwas 5 μM. The data were taken from 200 nm to 250 nm.Values at 222 nm were recorded and averaged from threeexperiments. The protein samples were incubated over-night at 25 °C. Protein-only spectra were obtained bysubtracting the CD signal from the corresponding buffer.The protein samples were scanned five times.

284 SNPs That Lead to Defects in Refolding

partial pH-dependent denaturation and unfoldingdue to protonation of acidic residues.25–27 Measure-ment of tryptophan emission maxima change andANS binding as functions of pH from pH 1.5 to pH 8(Fig. 3a) showed that wild-type human mitochon-drial phenylalanyl-tRNA synthetase and the S57C/N280S double mutant had remarkable stability evenat low pH. The tryptophan emission maximumexhibited a 4-nm to a 5-nm blueshift compared tophysiological pH, indicating that the tryptophansbecome less accessible at low pH. CorrespondingANS fluorescence indicated that wild-typehmtPheRS maintained a molten-globule-like struc-ture from pH 1.5 to pH 5. The S57C and N280Svariants both showed decreased ANS intensitycompared to wild type and the double mutant atpH 4 and below, and both ANS fluorescence andtryptophan emission maxima indicated an unfoldedconformation of the variants between pH 2.0 andpH 3.5 (Fig. 3a and b). These differences in stabilitybetween wild-type hmtPheRS and the nsSNP-encoded variants were supported by far-UV CDmeasurements. Changes in ellipticity at 222 nm as afunction of pH (Fig. 3c) showed very little pertur-bation of secondary structure for the wild type or thedouble mutant from pH 1.5 to pH 8, while thensSNP-encoded variants showed reduced second-ary structure with a predominantly random coilstructure between pH 2 and pH 3.5.

Human mitochondrial nsSNP variants withreduced refolding capability

Aminoacylation activity measurements (Fig. 1b)showed that hmtPheRS nsSNP variants had asimilar charging capacity for cognate tRNAPhe aswild type, indicative of native-like catalytic cores.The variants were marginally less structurally stablethan wild type under certain conditions (see the textabove), and their stability was further investigatedby measuring aminoacylation activity followingextended incubation at room temperature. Wild-type hmtPheRS retained over 90% activity followingincubation at room temperature for 24 h, whereasthe S57C and N280S variants retained less than 30%

Fig. 4. Effect of refolding on hmtPheRS activity. (a)Aminoacylation activity after incubation at room temper-ature. Active fractions of 10 μM of all the protein sampleswere kept in aminoacylation buffer. After 24 h, the proteinsamples were diluted to 100 nM in aminoacylation buffer,and aminoacylation assay was performed in triplicate. Theaverage values, along with standard deviations, werecalculated and plotted. (b) Refolding of hmtPheRSvariants. The protein samples were incubated for 5 h atroom temperature at a concentration of 10 μM activefraction in aminoacylation buffer. The protein sampleswere then diluted to 1 μM, and active-site titration wasperformed. Dilution with the same pH buffer, instead ofaminoacylation buffer, was used as control. The data weretaken from pH 1 to pH 6.5. However, below pH 5.5, noactivity was observed after refolding. All activities arebased on aminoacylation plateaus.

285SNPs That Lead to Defects in Refolding

of their initial activity (Fig. 4a). To investigate if thisloss in activity was indicative of a reduced ability torefold from a partially denatured state, we deter-mined the number of active sites after low pHdenaturation and refolding. Wild-type hmtPheRSretained ∼80–95% of the active sites present prior todenaturation, while the S57C and N280S variantsretained ∼40–65% and 30–60%, respectively, sug-gesting that both were defective in refoldingcompared to wild type (Fig. 4b).To determine if refolding defects also lead to

reduced activity in nsSNP-encoded variants of other

mitochondrial aaRSs,we tested the effects of refoldingfrom low pH on the activity of H324Q humanmitochondrial leucyl-tRNA synthetase (hmtLeuRS).This variant is of particular interest, as it has beenassociated with possible susceptibility to a form oftype 2 diabetes.23 Far-UV CD spectra, tryptophanemission maxima, and ANS binding as a function ofpH were all identical for wild-type and H324QhmtLeuRS, indicating that both were equally stableand formedmolten-globule-like structures (Fig. 5a–c).This is in contrast to the nsSNP-encoded changes inhmtPheRS investigated above,which have significanteffects on stability. While no changes in unfoldingwere seen following the H324Q replacement, thisnsSNP-encoded variant did show altered refoldingcompared to wild-type hmtLeuRS. Low-pH denatur-ation and subsequent refolding reduced the numberofH324QhmtLeuRS active sites by 10–30% comparedtowild type, similar to the loss in activity observed forthe nsSNP-encoded hmtPheRS variants (Fig. 5d).These specific changes in the refolding of H324Qsuggest that the side chain of His324 may play acrucial role in tertiary domain–domain interactionsthat go missing when replaced with glutamine.These data show that the H324Q replacement inhmtLeuRS specifically targets refolding, in contrastto S57C and N280S hmtPheRSs, which have broaderdefects in stability encompassing both unfolding andrefolding.

In vivo activity of the nsSNP variants ofhmtPheRS

To investigate the possible effects on the growth ofthe nsSNP-encoded hmtPheRS variants, we com-plemented a Saccharomyces cerevisiae strain lackingits endogenous mitochondrial PheRS (MSF1). ThehmtPheRS (encoded by FARS2), the nsSNP variants,and the corresponding double mutant were insertedinto the low-copy-number centromeric plasmidpFL36, resulting in the plasmids pFL36-FARS2,pFL36-S57C, pFL36-N280S, and pFL36-S57C/N280S, which were then used to complement S.cerevisiae msf1Δ cells. There was no significantdifference in growth between the FARS2 wild typeand either of the two single nsSNP mutants onfermentative or respiratory media (Fig. 6a). Whilethere was no difference in growth seen on fermen-tative media, a growth difference was observedbetween the wild type and the S57C/N280S doublemutant on respiratory media. To test the significanceof the growth difference between the FARS2 wildtype and the nsSNP variants, we monitored growthin liquid media (Fig. 6b). In liquid respiratory media,a modest but significant growth difference betweenthe wild type and the S57C/N280S double mutantwas observed (Fig. 6b). While the hmtPheRS S57C/N280S double mutant is able to support the growthof S. cerevisiae on respiratory media, the observed

286 SNPs That Lead to Defects in Refolding

negative impact on growth suggests that mitochon-dria in this strain have some deficit in their ability torespire.

Molecular dynamics simulations of hmtPheRS

Twelve-nanosecond molecular dynamics simula-tions were carried out on wild-type hmtPheRS andthe S57C and N280S variants using methodspreviously described for Escherichia coli MetRS.45

Crystal coordinates from Protein Data Bank (PDB)ID 3CMQ were used for the wild-type protein, andvariants were generated by in silico mutagenesis ofthese coordinates. The crystallized protein lackedthe mitochondrial targeting sequence, and nodensity was observed for the 10 N-terminal residues;thus, the simulations encompassed residues 47–451.α-Carbon root-mean-square fluctuations (RMSFs)were calculated by comparison with an orientedaverage protein structure derived from each trajec-tory. Regions of hmtPheRS with high RMSF values(Fig. 7) correspond to sequences within or adjacentto surface loops. Differences in RMSF between thewild type and the S57C or N280S variants are lessthan 2-fold throughout, and the observation thatboth amino acid replacements tend to result inhigher (rather than lower) RMSFs is consistent withless stably structured proteins. A 12-ns simulation ofthe double variant S57C/N280S was also per-formed, and the α-carbon RMSFs calculated werewithin 2-fold of the wild-type and single-variantRMSFs similarly determined (Fig. 7). Thus, compu-tational modeling does not suggest functionalcooperativity between these sites on hmtPheRS.

Discussion

Formation of stable molten globule states byhuman mitochondrial aaRSs

Nuclear-encoded proteins are imported into mi-tochondria either unfolded or in a partially unfoldedextended conformation such as a molten globulestate.46 Unfolding can be driven by different factors,including components of the protein import ma-chinery itself and the mitochondrial membranepotential.28,29,47 The data presented here show thatfor hmtLeuRS and hmtPheRS, partial unfolding toform a molten-globule-like intermediate can readilyoccur, and that this requires moderately acidic pHconditions similar to those associated with the

Fig. 5. Structure and activity of hmtLeuRS. (a) Far-UVCD spectra of hmtLeuRS. Measurements were performedas described for Fig. 1c, except that the protein concentra-tion was 2.5 μM. (b) pH inducedwith tryptophan emissionmaxima changes. Measurements were recorded as de-scribed in Fig. 3a. (c) ANS binding as a function of pH. Thedata were taken as described in Fig. 3b. (d) Refolding ofhmtLeuRS variants. Measurements were performed asdescribed in Fig. 4b.

Fig. 6. Growth of S. cerevisiae W303 msf1Δ strains complemented with FARS2 wild type, S57C, N280S, and S57C/N280S. (a) Serial dilution of complemented strains on glucose (YPDA) and glycerol plus ethanol (YPGA plus ethanol)media. Plates were incubated at 30 °C for 2–3 days. Data points are an average of three replicates, with error barsrepresenting 1 SD. (b) Growth curve of complemented strains in YPGA. Cultures were grown at 30 °C, with shaking at300 rpm.

287SNPs That Lead to Defects in Refolding

generation of local pH gradients across the mito-chondrial membrane.48 An acidic pH environmentat the mitochondrial outer membrane matrix hasbeen shown to help the steroidogenic acute regula-tory protein adopt a functional molten globulestructure,49 and several other examples where either

a lower pH or a specific membrane compositionallows a protein to adopt predominantly a moltenglobule form have also been described.26,50 Theability of both hmtLeuRS and hmtPheRS to formremarkably stable molten-globule-like intermedi-ates at an acidic pH suggests that import into the

Fig. 7. Mobility of hmtPheRSvariants as determined by molecu-lar dynamics simulations. For wildtype, S57C, and N280S, each trace isthe average α-carbon RMSF de-rived from two independent 12-nssimulations carried out under con-ditions of normal pressure andtemperature following systempreparation and equilibration. Theresults of a single 12-ns simulationof the S57C/N280S variant areshown for comparison (see the textfor details).

288 SNPs That Lead to Defects in Refolding

mitochondrial matrix of these and perhaps otheraaRSs might be facilitated by adopting such astructure. Our data also suggest that the presenceof the N-terminal mitochondrial targeting sequenceof hmtPheRS does not alter the protein's folding.hmtPheRS lacking a targeting sequence producedheterologously in E. coli retained full activity in vivo(data not shown) and also refolded into its nativeactive state in vitro without the aid of anychaperones.

nsSNPs of mitochondrial aaRSs encoderefolding-specific defects that reduce overallaminoacylation activity

In recent years, mutations in human mitochon-drial aaRS-encoding genes that lead to defects ingene expression, enzymatic activity, and subcellularlocalization have been identified.20–22,51,52 Beyondthe well-characterized examples that have beenlinked directly to specific diseases, considerablymore mutations in aaRS-encoding genes have beenuncovered through sequencing of nsSNPs. In someinstances, extensive knowledge on the structure andfunction of aaRSs allows reasonably confidentpredictions to be made as to the possible effects ofsuch mutations. In other examples, such as thensSNPs of FARS2 and LARS2 studied here, howamino acid substitutions might impact function isunclear, making it difficult to discern whether or nota particular mutation may be deleterious. ThensSNP variants of hmtPheRS showed cognatetRNA charging activity comparable to wild type,and molecular dynamics simulations did not indi-cate any unexpected changes in the native foldedstructure as a result of the amino acid replacements.However, both nsSNP variants of hmtPheRS andthe nsSNP variant of hmtLeuRS refolded lessefficiently than their respective wild type in vitro,resulting in a net reduction in aminoacylationactivity. Whether this change in aaRS folding andaminoacylation activity compared to wild type issignificant for mitochondrial biogenesis is unclear.Although more severe or long-term effects onhuman cells cannot be excluded, the ability of thensSNP variants to function in yeast neverthelessindicates that these mutants do not critically limitmitochondrial import or translation, as reported forother nsSNPs.22,53,54 The only exception was theS57C/N280S double mutant, which was less activein vivo than either of the respective single mutants,despite the fact that all three hmtPheRSs showedcomparable losses in aminoacylation activity invitro. One notable feature of the S57C/N280Sprotein is that it unfolds more slowly than eitherof the single mutants or wild type, potentiallyleading to defects in protein import and supportingthe assertion that unfolding is an important stepduring translocation of hmtPheRS.

These findings reveal the existence of a new classof nsSNPs in the aaRS family that have no obviouseffect on the maturation or activity of newlysynthesized proteins, but instead perturb optimalunfolding and refolding. While these mutations areinitially less detrimental than those that, forexample, target the active site, they neverthelesshave the potential to significantly limit aaRSactivity. Mitochondrial aaRSs are most vulnerableto such mutations, as their activity is dependent onproper refolding following translocation into theorganelle. What is still unclear is to what extentunfolding and refolding defects might also affectcytoplasmic aaRSs, perhaps by making them lessstable and reducing their half lives, or by actingsynergistically with other classes of mutations.Further studies are now warranted to investigatethe extent to which other uncharacterized nsSNPseffect both cytoplasmic and mitochondrial aaRSrefolding, and how this contributes to diseasesresulting from such mutations.

Materials and Methods

Protein expression, purification, and characterization

Point mutations were introduced by site-directedmutagenesis using the QuikChange procedure (Strata-gene). Primers were obtained from Sigma ChemicalCompany (St. Louis, MO). Wild-type hmtPheRS andvariants were purified according to previously publishedprocedures.55 E. coli strain BL21 (pArgU218)/pET21c-PheRS expressing C-terminal His6-tagged hmtPheRS wasa gift from Prof. Linda Spremulli (University of NorthCarolina, Chapel Hill, NC). Rosetta(DE3) cells containingpRARE plasmids encoding tRNAs for rare codons weretransformed with the mutant hmtPheRS plasmid con-struct and grown on a medium containing both ampicillinand chloramphenicol. Cells were grown at 37 °C to anoptical density of 0.6 and induced with 0.4 mM IPTG for3 h. Cells were harvested and sonicated, and thesupernatant was collected after centrifugation at 75,000gfor 1 h. The supernatant was applied to a TALON® metalaffinity resin column (Clontech), washed in a buffercontaining 25 mM imidazole, and eluted with 25 mMTris–HCl (pH 8.0), 300 mM NaCl, 250 mM imidazole, and10% glycerol. Fractions containing hmtPheRS werechecked for purity by SDS-PAGE, pooled, and dialyzedfor 48 h at 4 °C against 50 mM Tris–HCl (pH 7.5), 5 mMMgCl2, and 100 mM KCl, with four buffer changes toremove all traces of imidazole that may interfere withfluorescence measurements. The purified enzyme wasconcentrated and adjusted to 50% (vol/vol) glycerol, andaliquots were stored at −80 °C. For purification of S57C/N280S hmtPheRS, cells were grown on autoinductionmedia, as previously described.56 Briefly, cells were grownovernight in LB medium containing ampicillin andchloramphenicol, reinoculated into fresh media, andthen grown to an optical density of 0.4–0.6.Ten milliliters of this culture was transferred to 1 L ofautoinduction media and grown overnight at 25 °C. The

289SNPs That Lead to Defects in Refolding

cells were then harvested, and protein purification wasperformed as described above for wild-type hmtPheRS.Purification of wild-type and H324Q hmtLeuRS wasperformed according to previously published protocols.57

tRNA aminoacylation

Aminoacylation was performed at 37 °C in aminoacyla-tion buffer [100mMNa-Hepes (pH 7.2), 30 mMKCl, 2 mMATP, and 10 mM MgCl2] containing 25 μM L-[14C]Phe(214 cpm pmol−1; Perkin-Elmer Lifesciences), 3 μM nativeE. coli tRNAPhe, and 200 nM hmtPheRS. Nine-microliteraliquots were removed and spotted on 3MM filter disks(Whatman), washed three times in 10% trichloroaceticacid, and dried. The amount of radioactivity retainedwas determined by liquid scintillation counting. Oneunit of hmtPheRS activity corresponded to the amountof enzyme necessary to catalyze the formation of 1 nmolPhe-tRNAPhe min− 1 mg−1 protein at 37 °C.

Active-site titration

Active-site titration was performed in a 50-μL reactionmixture containing 100 mM Na-Hepes (pH 7.2), 30 mMKCl, 10 mM MgCl2, 2 mM ATP, 25 μM L-[14C]Phe(214 cpm pmol−1) or L-[U-14C]Leu (306 cpm pmol−1),and 5 mM β-mercaptoethanol. Pyrophosphatase was alsoadded to ensure a unidirectional formation of aminoacyladenylate. The reaction was initiated by the addition ofhmtPheRS or hmtLeuRS (1 μM determined spectrophoto-metrically) to a reaction mixture preincubated at 37 °C for5 min. After addition of the enzyme, the reaction wasperformed for 10 min at 37 °C and then filtered through anitrocellulose membrane (PROTRAN BA85; Whatman)prewashed with cold 0.5× aminoacylation buffer. Thefilters were then washed with 3 mL of cold 0.5×aminoacylation buffer and dried at 80 °C for 15 min. Theamount of radioactivity retained was quantified by liquidscintillation counting.

Fluorescence spectroscopy

Fluorescence was measured with a Fluorolog-3 spec-trofluorometer (Horiba Jobin Yvon) equipped with aconstant-temperature cell holder at an integration time of1 s. All measurements were carried out at least threetimes. Protein concentrations were 2.5 μM for allfluorescence measurements. Tryptophan was selectivelyexcited at 295 nm, and emission was recorded from310 nm to 450 nm. Both excitation and emission slitwidths were set at 5 nm for all experiments, unlessmentioned otherwise. The λ maxima values wereobtained by taking the first derivative of the correspond-ing emission spectra.

Urea-induced unfolding

Molecular-biology-grade urea was obtained from SigmaChemical Company. All buffers were filtered using a 0.22-μm filter device from Millipore. Urea-induced denatur-ation of wild-type hmtPheRS and variants was performedwith increasing concentrations of the denaturant. Protein

samples were incubated at a desired denaturant concen-tration for 18–24 h at 25 °C to attain thermodynamicequilibrium (no further change in fluorescence intensity).Thermodynamic equilibrium was generally reached after5 h of incubation. The final concentrations of the proteinand denaturant in each sample were determined byspectrophotometry and refractive index measurements,respectively. Data are expressed in terms of the fractionunfolded (Fun), calculated from the equation:

Fun = Fobs − FNð Þ = FN − FUð Þ ð1Þ

where Fobs is the observed value of the tryptophanemission maxima at a given denaturant concentration,and FN and FU are the values of native and unfoldedprotein, respectively. By assuming a simple two-statemodel, we fitted the transitions to the following equation:

Fun = exp − DG H2OU–N + m D½ �

� �=RT

= 1 + exp − DG H2OU–N + m D½ �

� �= RT

h ið2Þ

where ΔGU–NH2O is the free-energy difference in the absence

of denaturant,m is the cooperativity of the reaction,D is thedenaturant concentration, R is the universal gas constant,and T is the temperature (in Kelvin).58 Kyplot (32 bit,version 2.0 beta 15; Koichi Yoshioka, 1997–2001) was usedto obtain the parameters using the nonlinear square fitmethod.

Acid denaturation

Acid denaturation of hmtPheRS was investigated as afunction of pH using KCl–HCl (pH 0.5–1.5), Gly–HCl(pH 2.0–3.5), sodium acetate (pH 4.0–5.5), and sodiumphosphate (pH 6.0–8.0) buffers.59 Analytical-grade chemi-cals were used for buffer preparation (all 50 mM), filteredand stored at −20 °C, and thawed at room temperatureimmediately prior to use. Protein samples were preparedindividually at different pH values. Protein stock solutionwas added to the appropriate buffer to reach a finalconcentration of 2.5 μM, and the mixture was incubatedfor 18–24 h at 25 °C to ensure that thermodynamicequilibrium had been reached. The final pH and concen-tration of the protein in each sample were then remea-sured.

Refolding assay

hmtPheRS or hmtLeuRS (10 μM) was incubated inaminoacylation buffer either at a different pH (pH 1–6with intervals of 0.5) or until a complete loss of activity inactive-site titration assays had been attained (usually 5–6 h). The reactions were then diluted 10-fold in aminoa-cylation buffer, and activity was again followed by active-site titration. Two separate control experiments wereperformed: in one case, the dilution was performed only in100 mMNa-Hepes buffer (pH 7.2), and in the second case,dilutions were performed in aminoacylation buffer corre-sponding to the same pH (only Na-Hepes was replaced bya suitable buffer at the pH mentioned below). All theexperiments were repeated three times, and averages werecalculated.

290 SNPs That Lead to Defects in Refolding

CD spectropolarimetry

CD spectra were measured at 25 °C with an Aviv 62ADS spectropolarimeter (Aviv). The protein concentrationwas 5 μM in 50 mM Tris–HCl (pH 7.5) and 5% glycerol.To measure changes in the secondary structure of theprotein, we monitored the far-UV CD spectra in theregion between 200 nm and 250 nm with a proteinconcentration of 5 μM (five scans per sample) and a stepsize of 1 nm in a 1-mm path-length cuvette, with 1 nmof bandwidth and 5 s of averaging time. Changes in thetertiary structure were observed in a 10-mm path-lengthcuvette in the near-UV region between 250 nm and320 nm (three scans per sample) at a protein concentra-tion of 10 μM, with all the other parameters identicalwith those used for far-UV CD measurements. Protein-only spectra were obtained by subtracting the CD signalfrom that for the corresponding buffer. For pH-inducednear-UV CD measurement, a highly concentrated pro-tein solution was diluted at once, and the final pH wasmeasured, as overnight dialysis in the same buffer led toaggregation.

ANS binding assay

The extent of exposure of hydrophobic surfaces in theenzyme was measured by the ability to bind to thefluorescent dye ANS.44 A stock solution of ANS wasprepared in methanol, and the dye concentration wasdetermined using an extinction coefficient of 5000 M−1

cm−1 at 350 nm.60 Protein (2.5 μM) was incubated witha 25-fold molar excess of ANS for more than 30 min atroom temperature in the dark, and the ANS fluores-cence was then measured. The excitation wavelengthwas set to 420 nm to avoid inner filter effect, and theemission spectra were collected between 430 nm and550 nm. The intensities at 482 nm were recorded andplotted as a function of pH.

Yeast complementation

S. cerevisiae W303 msf1Δ was created through thereplacement of the MSF1 open reading frame with aKanMX4 cassette by homologous recombination in aW303 MSF1 homozygous diploid. W303 msf1∷KanMX4was then obtained by sporulation and dissection.hmtPheRS was inserted into the low-copy-numbercentromeric plasmid pFL36, resulting in the plasmidpFL36-FARS2. To ensure that the FARS2 protein wastargeted to the S. cerevisiae mitochondria, we removedthe human mitochondrial targeting signal sequence fromthe FARS2 gene and cloned the remaining fragment ofthe FARS2 coding sequence behind the MSF1 mitochon-drial targeting sequence and the start codon (66 bp). Toallow for native MSF1 regulation of the construct, wecloned the hybrid FARS2 coding plus MSF1 mitochon-drial targeting sequence behind an additional 414 bp ofsequence upstream of the MSF1 start codon and in frontof 151 bp of sequence downstream of the MSF1 stopcodon. The FRS2 S57C, N280S, and S57C/N280Smutations were introduced through site-directed muta-genesis. The haploid W303 msf1∷KanMX4 strain wasseparately transformed with the plasmid pFL36-FARS2,

pFL36-S57C, pFL36-N280S, or pFL36-S57C/N280S,crossed with W303 MSF1, sporulated, and dissectedonto YPDA (yeast, peptone, dextrose, adenine). Theresulting W303 msf1∷KanMX4 haploids carrying theappropriate plasmid were selected and utilized in all S.cerevisiae growth assays. Growth curve assays wereconducted in triplicate at 30 °C in 100 mL of YPGA(yeast, peptone, glycerol, adenine) in 500-mL flasksshaken at 300 rpm, or on plates containing eitherYPGA or glycerol plus ethanol (YPGA plus ethanol)medium.

Molecular dynamics simulations of hmtPheRS

Crystal structure coordinates of hmtPheRS (PDB ID:3CMQ) were used for nanosecond-scale molecular dy-namics simulations, as described previously for E. coliMetRS.45 In silico mutagenesis was carried out on the3CMQ coordinates to generate S57C and N280S asfollows: the atoms of each variant not held in commonwith the wild-type protein were removed, and the aminoacids S57 and N280 were reassigned to cysteine andserine, respectively. Missing atoms (including hydrogens)were added using the standard CHARMM package.61 Anenergy minimization step was first performed to allow thenew side chain to relax, reducing any steric overlaps andoptimizing the position of new atoms. All atoms, exceptfor the substituted residue, were constrained using aharmonic position restraint force constant of 60.0 kcalmol−1 Å−2 for 500 cycles of conjugate gradient minimi-zation. Then, to allow side chains to relax, we constrainedall backbone atoms at the same force constant for another500 cycles of conjugate gradient minimization. Theharmonic restraint force constant was reduced by20.0 kcal mol−1 Å−2 after every 1000 such cycles ofminimization (500 cycles constraining all but thesubstituted residue, followed by 500 cycles constrainingthe heavy atoms) until a constant of 10 kcal mol−1 Å−2

had been reached. Lastly, all heavy atoms were con-strained to allow hydrogens to relax, using a harmonicrestraint force constant of 30 kcal mol− 1 Å− 2 anddecreasing by 10.0 kcal mol−1 Å−2 after every 500 cyclesof conjugate gradient minimization (again until a constantof 10 kcal mol−1 Å−2 had been reached). A modifiedCHARMM22 parameter set62 was used throughout. Asolvent box of TIP3P63 water molecules was added usingthe solvate package of VMD,64 and the system wasneutralized to an ionic strength of 50 nM with addition ofsodium and chloride ions using VMD's autoionizefunction.System equilibration was carried out for 224 ps

under conditions of normal pressure and temperature,as described previously,45 using the NAMD package65

and modified CHARMM22 parameter set. A further12 ns of simulation trajectory was carried out underthe same conditions of normal pressure and temper-ature, after which coordinates were aligned toaccount for rotational drift and an oriented averageprotein structure was determined using all proteinatoms. The RMSF for each α-carbon was calculatedcompared to this average PDB coordinate set. Twoindependent simulations (including system equilibra-tion, production, and analysis) were performed oneach protein.

291SNPs That Lead to Defects in Refolding

Acknowledgements

We thank Dr. M. P. King (Department ofBiochemistry and Molecular Pharmacology, ThomasJefferson University, Philadelphia, PA) for provid-ing a plasmid encoding wild-type hmtLeuRS andDr. L. M. Hart (Department of Molecular CellBiology, Leiden University Medical Center, TheNetherlands) for providing a plasmid encodingH324Q hmtLeuRS. This work was supported bygrants from the Binational Science Foundation(2005209 to M.I.) and the National Science Foun-dation (744791 to M.I. and 448243 to R.W.A.).

References

1. Alfonzo, J. D. & Söll, D. (2009). Mitochondrial tRNAimport—the challenge to understand has just begun.Biol. Chem. 390, 717–722.

2. Frechin, M., Kern, D., Martin, R. P., Becker, H. D. &Senger, B. (2010). Arc1p: anchoring, routing, coordi-nating. FEBS Lett. 584, 427–433.

3. Bonnefond, L., Fender, A., Rudinger-Thirion, J., Giege,R., Florentz, C. & Sissler, M. (2005). Toward the full setof human mitochondrial aminoacyl-tRNA synthe-tases: characterization of AspRS and TyrRS. Biochem-istry, 44, 4805–4816.

4. Duchene, A. M., Giritch, A., Hoffmann, B., Cognat, V.,Lancelin, D., Peeters, N. M. et al. (2005). Dual targetingis the rule for organellar aminoacyl-tRNA synthetasesin Arabidopsis thaliana. Proc. Natl Acad. Sci. USA, 102,16484–16489.

5. Tolkunova, E., Park, H., Xia, J., King, M. P. &Davidson, E. (2000). The human lysyl-tRNA synthe-tase gene encodes both the cytoplasmic and mito-chondrial enzymes by means of an unusualalternative splicing of the primary transcript. J. Biol.Chem. 275, 35063–35069.

6. Edwards, H. & Schimmel, P. (1987). An E. coliaminoacyl-tRNA synthetase can substitute for yeastmitochondrial enzyme function in vivo. Cell, 51,643–649.

7. Houman, F., Rho, S. B., Zhang, J., Shen, X., Wang, C.C., Schimmel, P. &Martinis, S. A. (2000). A prokaryoteand human tRNA synthetase provide an essentialRNA splicing function in yeast mitochondria. Proc.Natl Acad. Sci. USA, 97, 13743–13748.

8. Montoya, J., Lopez-Gallardo, E., Diez-Sanchez, C.,Lopez-Perez, M. J. & Ruiz-Pesini, E. (2009). 20 years ofhuman mtDNA pathologic point mutations: carefullyreading the pathogenicity criteria. Biochim. Biophys.Acta, 1787, 476–483.

9. Putz, J., Dupuis, B., Sissler, M. & Florentz, C. (2007).Mamit-tRNA, a database of mammalian mitochondri-al tRNA primary and secondary structures. RNA, 13,1184–1190.

10. Florentz, C., Sohm, B., Tryoen-Toth, P., Putz, J. &Sissler, M. (2003). Human mitochondrial tRNAs inhealth and disease. Cell. Mol. Life Sci. 60, 1356–1375.

11. Florentz, C. & Sissler, M. (2001). Disease-related versuspolymorphic mutations in human mitochondrial

tRNAs. Where is the difference? EMBO Rep. 2,481–486.

12. Sissler, M., Helm, M., Frugier, M., Giege, R. &Florentz, C. (2004). Aminoacylation properties ofpathology-related human mitochondrial tRNA(Lys)variants. RNA, 10, 841–853.

13. Ling, J., Roy, H., Qin, D., Rubio, M. A., Alfonzo,J. D., Fredrick, K. & Ibba, M. (2007). Pathogenicmechanism of a human mitochondrial tRNAPhe

mutation associated with myoclonic epilepsy withragged red fibers syndrome. Proc. Natl Acad. Sci.USA, 104, 15299–15304.

14. Kirino, Y., Goto, Y., Campos, Y., Arenas, J. & Suzuki,T. (2005). Specific correlation between the wobblemodification deficiency in mutant tRNAs and theclinical features of a human mitochondrial disease.Proc. Natl Acad. Sci. USA, 102, 7127–7132.

15. Li, R. & Guan, M. X. (2010). Human mitochondrialleucyl-tRNA synthetase corrects mitochondrial dys-functions due to the tRNALeu(UUR) A3243Gmutation, associated with mitochondrial encephalo-myopathy, lactic acidosis, and stroke-like symptomsand diabetes. Mol. Cell. Biol. 30, 2147–2154.

16. Mollers, M., Maniura-Weber, K., Kiseljakovic, E., Bust,M., Hayrapetyan, A., Jaksch, M. et al. (2005). A newmechanism for mtDNA pathogenesis: impairment ofpost-transcriptional maturation leads to severe deple-tion of mitochondrial tRNASer(UCN) caused byT7512C and G7497A point mutations. Nucleic AcidsRes. 33, 5647–5658.

17. Maniura-Weber, K., Helm, M., Engemann, K.,Eckertz, S., Mollers, M., Schauen, M. et al. (2006).Molecular dysfunction associated with the humanmitochondrial 3302A→G mutation in the MTTL1(mt-tRNALeu(UUR)) gene. Nucleic Acids Res. 34,6404–6415.

18. Roy, M. D., Wittenhagen, L. M. & Kelley, S. O. (2005).Structural probing of a pathogenic tRNA dimer. RNA,11, 254–260.

19. Wittenhagen, L. M. & Kelley, S. O. (2003). Impact ofdisease-related mitochondrial mutations on tRNAstructure and function. Trends Biochem. Sci. 28,605–611.

20. Scheper, G. C., van der Klok, T., van Andel, R. J., vanBerkel, C. G., Sissler, M., Smet, J. et al. (2007).Mitochondrial aspartyl-tRNA synthetase deficiencycauses leukoencephalopathy with brain stem andspinal cord involvement and lactate elevation. Nat.Genet. 39, 534–539.

21. Edvardson, S., Shaag, A., Kolesnikova, O., Gomori,J. M., Tarassov, I., Einbinder, T. et al. (2007).Deleterious mutation in the mitochondrial arginyl-transfer RNA synthetase gene is associated withpontocerebellar hypoplasia. Am. J. Hum. Genet. 81,857–862.

22. Messmer, M., Florentz, C., Schwenzer, H., Scheper,G. C., van der Knaap, M. S., Marechal-Drouard, L. &Sissler, M. (2011). A human pathology-relatedmutation prevents import of an aminoacyl-tRNAsynthetase into mitochondria. Biochem. J. 433,441–446.

23. t Hart, L. M., Hansen, T., Rietveld, I., Dekker, J. M.,Nijpels, G., Janssen, G. M. et al. (2005). Evidence thatthe mitochondrial leucyl tRNA synthetase (LARS2)

292 SNPs That Lead to Defects in Refolding

gene represents a novel type 2 diabetes susceptibilitygene. Diabetes, 54, 1892–1895.

24. Reiling, E., Jafar-Mohammadi, B., van 't Riet, E.,Weedon, M. N., van Vliet-Ostaptchouk, J. V., Hansen,T. et al. (2010). Genetic association analysis of LARS2with type 2 diabetes. Diabetologia, 53, 103–110.

25. Bychkova, V. E., Pain, R. H. & Ptitsyn, O. B. (1988).The ‘molten globule’ state is involved in the translo-cation of proteins across membranes? FEBS Lett. 238,231–234.

26. van der Goot, F. G., Gonzalez-Manas, J. M., Lakey,J. H. & Pattus, F. (1991). A ‘molten-globule’membrane-insertion intermediate of the pore-formingdomain of colicin A. Nature, 354, 408–410.

27. van der Goot, F. G., Lakey, J. H. & Pattus, F. (1992).The molten globule intermediate for protein insertionor translocation through membranes. Trends Cell. Biol.2, 343–348.

28. Huang, S., Ratliff, K. S. & Matouschek, A. (2002).Protein unfolding by the mitochondrial membranepotential. Nat. Struct. Biol. 9, 301–307.

29. Wilcox, A. J., Choy, J., Bustamante, C. & Matouschek,A. (2005). Effect of protein structure on mitochondrialimport. Proc. Natl Acad. Sci. USA, 102, 15435–15440.

30. Longley, M. J., Humble, M. M., Sharief, F. S. &Copeland,W. C. (2010). Disease variants of the humanmitochondrial DNA helicase encoded by C10orf2differentially alter protein stability, nucleotide hydro-lysis, and helicase activity. J. Biol. Chem. 285,29690–29702.

31. Bross, P., Jespersen, C., Jensen, T. G., Andresen, B. S.,Kristensen, M. J., Winter, V. et al. (1995). Effects of twomutations detected in medium chain acyl-CoA dehy-drogenase (MCAD)-deficient patients on folding,oligomer assembly, and stability of MCAD enzyme.J. Biol. Chem. 270, 10284–10290.

32. Antonellis, A. & Green, E. D. (2008). The role ofaminoacyl-tRNA synthetases in genetic diseases.Annu. Rev. Genomics Hum. Genet. 9, 87–107.

33. Stum, M., McLaughlin, H. M., Kleinbrink, E. L., Miers,K. E., Ackerman, S. L., Seburn, K. L. et al. (2011). Anassessment of mechanisms underlying peripheralaxonal degeneration caused by aminoacyl-tRNAsynthetase mutations. Mol. Cell. Neurosci. 46, 432–443.

34. Hamosh, A., Scott, A. F., Amberger, J. S., Bocchini, C.A. & McKusick, V. A. (2005). Online MendelianInheritance in Man (OMIM), a knowledgebase ofhuman genes and genetic disorders. Nucleic Acids Res.33, D514–D517.

35. Stenson, P. D., Ball, E. V., Mort, M., Phillips, A. D.,Shiel, J. A., Thomas, N. S. et al. (2003). Human GeneMutation Database (HGMD): 2003 update. Hum.Mutat. 21, 577–581.

36. Cargill, M., Altshuler, D., Ireland, J., Sklar, P., Ardlie,K., Patil, N. et al. (1999). Characterization of single-nucleotide polymorphisms in coding regions ofhuman genes. Nat. Genet. 22, 231–238.

37. Klipcan, L., Levin, I., Kessler, N., Moor, N., Finarov, I.& Safro, M. (2008). The tRNA-induced conformationalactivation of human mitochondrial phenylalanyl-tRNA synthetase. Structure, 16, 1095–1104.

38. Ng, P. C. & Henikoff, S. (2003). SIFT: predicting aminoacid changes that affect protein function. Nucleic AcidsRes. 31, 3812–3814.

39. Ramensky, V., Bork, P. & Sunyaev, S. (2002). Humannon-synonymous SNPs: server and survey. NucleicAcids Res. 30, 3894–3900.

40. Leung, W. C., Hessel, S., Meplan, C., Flint, J.,Oberhauser, V., Tourniaire, F. et al. (2009). Twocommon single nucleotide polymorphisms in thegene encoding beta-carotene 15,15′-monoxygenasealter beta-carotene metabolism in female volunteers.FASEB J. 23, 1041–1053.

41. Thummer, R. P., Drenth-Diephuis, L. J., Carney, K. E.& Eggen, B. J. (2010). Functional characterization ofsingle-nucleotide polymorphisms in the human undif-ferentiated embryonic-cell transcription factor 1 gene.DNA Cell. Biol. 29, 241–248.

42. Tejedor, M. T., Cenarro, A., Tejedor, D., Stef, M.,Mateo-Gallego, R., de Castro, I. et al. (2010). Haplo-type analyses, mechanism and evolution of commondouble mutants in the human LDL receptor gene.Mol.Genet. Genomics, 283, 565–574.

43. Shetty, S., Bhave, M. &Ghosh, K. (2011). Challenges ofmultiple mutations in individual patients with hae-mophilia. Eur. J. Haematol. 86, 185–190.

44. Semisotnov, G. V., Rodionova, N. A., Razgulyaev,O. I., Uversky, V. N., Gripas, A. F. & Gilmanshin,R. I. (1991). Study of the “molten globule”intermediate state in protein folding by a hydro-phobic fluorescent probe. Biopolymers, 31, 119–128.

45. Budiman, M. E., Knaggs, M. H., Fetrow, J. S. &Alexander, R. W. (2007). Using molecular dynamics tomap interaction networks in an aminoacyl-tRNAsynthetase. Proteins, 68, 670–689.

46. Schwartz, M. P., Huang, S. & Matouschek, A. (1999).The structure of precursor proteins during import intomitochondria. J. Biol. Chem. 274, 12759–12764.

47. Schleiff, E. & Becker, T. (2011). Common ground forprotein translocation: access control for mitochondriaand chloroplasts. Nat. Rev. Mol. Cell Biol. 12, 48–59.

48. Khalifat, N., Puff, N., Bonneau, S., Fournier, J. B. &Angelova, M. I. (2008). Membrane deformation underlocal pH gradient: mimicking mitochondrial cristaedynamics. Biophys. J. 95, 4924–4933.

49. Bose, H. S., Whittal, R. M., Baldwin, M. A. & Miller,W. L. (1999). The active form of the steroidogenicacute regulatory protein, StAR, appears to be amolten globule. Proc. Natl Acad. Sci. USA, 96,7250–7255.

50. Shin, I., Kreimer, D., Silman, I. & Weiner, L. (1997).Membrane-promotedunfoldingof acetylcholinesterase:a possiblemechanism for insertion into the lipid bilayer.Proc. Natl Acad. Sci. USA, 94, 2848–2852.

51. Belostotsky, R., Ben-Shalom, E., Rinat, C., Becker-Cohen, R., Feinstein, S., Zeligson, S. et al. (2011).Mutations in the mitochondrial seryl-tRNA synthe-tase cause hyperuricemia, pulmonary hypertension,renal failure in infancy and alkalosis, HUPRAsyndrome. Am. J. Hum. Genet. 88, 193–200.

52. Riley, L. G., Cooper, S., Hickey, P., Rudinger-Thirion,J., McKenzie, M., Compton, A. et al. (2010). Mutationof the mitochondrial tyrosyl-tRNA synthetase gene,YARS2, causes myopathy, lactic acidosis, and side-roblastic anemia–MLASA syndrome. Am. J. Hum.Genet. 87, 52–59.

53. Ueki, I., Koga, Y., Povalko, N., Akita, Y., Nishioka, J.,Yatsuga, S. et al. (2006). Mitochondrial tRNA gene

293SNPs That Lead to Defects in Refolding

mutations in patients having mitochondrial diseasewith lactic acidosis. Mitochondrion, 6, 29–36.

54. Rotig, A. (2010). Genetic bases of mitochondrialrespiratory chain disorders. Diabetes Metab. 36,97–107.

55. Yadavalli, S. S., Klipcan, L., Zozulya, A., Banerjee, R.,Svergun, D., Safro, M. & Ibba, M. (2009). Large-scalemovement of functional domains facilitates aminoa-cylation by human mitochondrial phenylalanyl-tRNAsynthetase. FEBS Lett. 583, 3204–3208.

56. Tyler, R. C., Sreenath, H. K., Singh, S., Aceti, D. J.,Bingman, C. A., Markley, J. L. & Fox, B. G. (2005).Auto-induction medium for the production of[U-15N]- and [U-13C, U-15N]-labeled proteins forNMR screening and structure determination. ProteinExpression Purif. 40, 268–278.

57. Park, H., Davidson, E. & King, M. P. (2003). Thepathogenic A3243G mutation in human mitochondri-al tRNALeu(UUR) decreases the efficiency of aminoa-cylation. Biochemistry, 42, 958–964.

58. Finn, E. B., Chem, X., Jennings, P. A., Soalau-Bethell,S. M. & Mathews, C. R. (1992). In Protein Engineering(Rees, A. R., Stemberg, M. J. E. & Wetzel, R., eds),Oxford University Press, New York, NY.

59. Dubey, V. K. & Jagannadham, M. V. (2003). Differ-ences in the unfolding of procerain induced by pH,

guanidine hydrochloride, urea, and temperature.Biochemistry, 42, 12287–12297.

60. Khurana, R. & Udgaonkar, J. B. (1994). Equilibriumunfolding studies of barstar: evidence for an alterna-tive conformation which resembles a molten globule.Biochemistry, 33, 106–115.

61. Brooks, B. R., Bruccoleri, R. E., Olafson, B. D., States,D. J., Swaminathan, S. & Karplus, M. (1983).CHARMM—a program for macromolecular energy,minimization, and dynamics calculations. J. Comput.Chem. 4, 187–217.

62. MacKerell, A. D., Bashford, D., Bellott, M., Dunbrack,R. L., Evanseck, J. D., Field, M. J. et al. (1998). All-atomempirical potential for molecular modeling anddynamics studies of proteins. J. Phys. Chem. B. 102,3586–3616.

63. Jorgensen, W. L., Chandrasekhar, J., Madura, J. D.,Impey, R. W. & Klein, M. L. (1983). Comparison ofsimple potential functions for simulating liquid water.J. Chem. Phys. 79, 926–935.

64. Humphrey, W., Dalke, A. & Schulten, K. (1996). VMD:Visual Molecular Dynamics. J. Mol. Graphics, 14, 33.

65. Kale, L., Skeel, R., Bhandarkar, M., Brunner, R.,Gursoy, A., Krawetz, N. et al. (1999). NAMD2: greaterscalability for parallel molecular dynamics. J. Comput.Phys. 151, 283–312.