Supporting Information - CaltechAUTHORSangles in protein structures. J Mol Biol 231:1049–1067. 11....

Supporting InformationSobreira et al. 10.1073/pnas.1011223108SI Materials and MethodsALDH Sequences. The following genomes were examined for alde-hyde dehydrogenase (ALDH) sequences: Aedes aegypti, Anophelesgambiae, Bos taurus, Branchiostoma floridae, Caenorhabditis ele-gans, Canis familiaris, Ciona intestinalis, Ciona savignyi, Drosophilamelanogaster, Gallus gallus, Gasterosteus aculeatus, Homo sapiens,Macacamulatta,Monodelphis domestica,Musmusculus,Nematostellavectensis, Ornithorhyncus anatinus, Oryzias latipes, Pan troglodytes,Phanerochaete chrysosporium, Phycomyces blakesleeanus, Populustrichocarpa, Rattus norvegicus, Saccharomyces cerevisiae, Strong-ylocentrotus purpuratus, Takifugu rubripes, Tetraodon nigroviridis,and Xenopus tropicalis. To these sequences, we added selectedsequences fromSaccoglossus kowalevskiiESTand trace archives aswell as ALDH sequences from various other animals (Apis melli-fera, Bombyx mori, Drosophila pseudoobscura, Macaca fasciculata,Mesocricetus auratus, Oryctolagus cuniculus, Ovis aries, Pongopygmaeus, Taeniopygia guttata, Tribolium castaneum, Xenopuslaevis), plants (Arabidopsis thaliana, Nicotiana tabacum, Oryzasativa, Populus trichocarpa, Secale cereale, Sorghum bicolor, Zeamays), and fungi (Chaetomium globosum, Cordyceps bassiana,Emericella nidulans, Magnaporthe grisea, Phaeosphaeria nodorum,Phanerochaete chrysosporium, Phycomyces blakesleeanus). To fa-cilitate structural comparisons with published work, the number-ing of ALDH amino acid residues was based on the classicalnumbering of residues in the mature human ALDH2 enzyme,which places the catalytic Cys at amino acid position 302 (1).

Large-Scale Phylogenetic Analyses.ALDH sequences were obtainedfrom www.aldh.org, whole genome data, EST databases, andtrace archives by using both signature (InterPro IPR002086) (2)and BLAST searches (3). ALDH amino acid sequences werealigned by using MUSCLE (4) followed by manual refinement.Bayesian inference (BI) was carried out by using MrBayes 3.1 (5)with a WAG+I+Γ4 model predicted by ProtTest (6). Two runs of5 million generations were computed for each tree. Convergencewas verified, and the burn-in period was determined by plottinglog likelihood versus time. Consensus trees and posterior prob-abilities were calculated by using the 50% majority rule. Amaximum likelihood (ML) analysis was performed by usingRAxML-VI-HPC with the PROTMIXWAG parameter (7) andwith 100 bootstrap pseudoreplicates to assess node support. Thephylogenetic analyses carried out with BI and ML resulted inessentially identical tree topologies.

ALDH Signatures. ALDH1 and ALDH2 sequence signatures wereobtained by aligning human ALDH1A2 and human ALDH2 andconfirmed inanexhaustivealignmentcontainingvertebrateALDH1andALDH2sequencesbyusingMultAlin (multalin.toulouse.inra.fr/multalin). Individual amino acid frequencies were obtainedfrom weblogo.berkeley.edu.

Ancestral Sequence Reconstruction. Ancestral protein sequences atinternal nodes of the ALDH1/2 phylogeny were reconstructed byusing the program PAML 3.15 (8), assuming a WAG+Γ4 modelfor a data matrix of 447 amino acid residues. To evaluate and limitthe influence of fast evolving sequences on the ancestral sequencereconstruction, five separate datasets were analyzed (one in-cluding all available ALDH1/2 sequences, one excluding all se-quences with long branches, one excluding all protostomesequences, one excluding all lophotrochozoan sequences, and oneexcluding all tunicate sequences). For each analysis, a corre-sponding ML tree was calculated by using RAxML-VI-HPC with

the PROTMIXWAG parameter (7) with 100 bootstrap replicatesto assess node support, which served as input tree for the ancestralsequence calculation. These control calculations, based on vary-ing taxonomic sampling, successfully tested the robustness of theancestral sequence reconstruction approach as well as the sub-sequent ancestral channel modeling and ancestral channel vol-ume calculation.

3D Modeling. ALDH1 and ALDH2 structures were modeled byhomology using the 3D structure obtained from the Protein DataBank (PDB) of sheep ALDH1A1 (PDB ID code 1BXS) or ofhuman ALDH2 in complex with the dipsogenic inhibitor daidzin(PDB ID code 1OF7). Each target sequence was globally alignedwith the template by using ClustalW. This alignment was used asinput data for the modeling program Nest in Jackal (9). Pa-rameters were set for refinement in all loops and secondarystructure regions. The stereochemical quality of the 3D modelswas evaluated by using Procheck (10). Retinal binding to thesubstrate access channel was carried out by anchoring the alde-hyde structure in a position equivalent to that of daidzin. The 3Dstructure of retinal was obtained from MSD Ligand Chemistry(www.ebi.ac.uk/msd-srv/msdchem). Graphical analyses wereperformed with VMD (Visual Molecular Dynamics) (11).

Cluster Analysis. We have compiled data on 202 ALDH1/2 vol-umes derived from 3D structure studies. TwoStep Cluster algo-rithm as implemented in SPSS 13.0 (SPSS for Windows, Rel.13.0.2004) was used both to define the best number of clustersthat accommodate the empirical data and to classify individualALDH1/2s into their respective volume clusters. The best numberof clusters was chosen as a function of the smallest SchwartzBayesian Criterion (BIC).

Retinal Binding. Retinal binding to the substrate access channelwas carried out by anchoring the aldehyde structure in a positionequivalent to daidzin. The retinal 3D structure was obtained fromMSD Ligand Chemistry (www.ebi.ac.uk/msd-srv/msdchem). Theretinal atoms O1, C15, C14, C13, C12, C11, and C10 were su-perposed on the daidzin atoms O34, C31, C30, C29, C28, C26,and C25, respectively.

Substrate Access Channel Volume Calculations. ALDH models wereindividually placed inside a grid box, in which x, y, and z coor-dinates were spaced by 0.8 Å. Void regions inside this box weredetermined by sequentially moving a probe molecule with a 1.4Å radius through all grid points. At each point, the volume oc-cupied by the probe plus an average atomic radius of 1.6 Å wasscanned. If no protein atoms were detected, the point was con-sidered to be in a void region. Multiple cavities were detected,and to isolate the substrate access channel from other spaces,the structure of human ALDH2 in complex with the dipsogenicinhibitor daidzin (PDB ID code 1OF7) was superposed on theboxed model. The void region in the boxed ALDH modelmatching the localization of daidzin inside the ALDH2 channelwas isolated for volume calculation. The total cavity volume wasdetermined as a sum of all distinct volume elements within theisolated void region. Each volume element was a cube defined byeight points enclosing a volume of 0.512 Å3. Modeled cavitieswere manually inspected to remove subsidiary spaces not asso-ciated with the substrate access channel. The van der Waalsvolumes for all amino acids have been described (12).

Sobreira et al. www.pnas.org/cgi/content/short/1011223108 1 of 14

http://www.aldh.orghttp://multalin.toulouse.inra.fr/multalinhttp://multalin.toulouse.inra.fr/multalinhttp://weblogo.berkeley.eduhttp://www.ebi.ac.uk/msd-srv/msdchemhttp://www.ebi.ac.uk/msd-srv/msdchemwww.pnas.org/cgi/content/short/1011223108

Statistical Analyses. Substrate access channel volumes were ana-lyzed in unpaired sets of eukaryote ALDH1/2 and ALDH1Lenzymes as well as in paired sets of vertebrate ALDH1/2s(ALDH1As, ALDH1B1s, and ALDH2s) by nonparametricANOVA using the Kruskal–Wallis test and Dunn’s multiplecomparisons test.

Gene Expression Studies. Sexually mature adults of the Floridaamphioxus (B. floridae) were collected by shovel and sieve inTampa Bay, Florida. Gametes were obtained by electric stimu-lation of adults. After fertilization, embryos and larvae wereraised in the laboratory and fixed at different developmentalstages. Five cDNA libraries were used for cloning ALDH1 andALDH2 genes from amphioxus. The cDNA libraries were madefrom RNA of unfertilized eggs, gastrulae, neurulae, early larvae,and mature adults (13). In situ hybridization experiments werecarried out as described (14) by using increased hybridizationtemperatures (65–70 °C) and nonconserved regions of the dif-ferent genes as templates for antisense riboprobe synthesis toensure probe specificity. Control experiments were carried outwith sense riboprobes to verify the specificity of the expressionpatterns. After in situ hybridization, embryos were mounted on

glass slides, analyzed under the microscope, and photographedas whole mounts.C. intestinalis adults were obtained from M-Rep. Embryos and

sperm were surgically removed and kept separate until in vitrofertilization. Fertilized eggs were dechorionated as described (15).Embryos were raised at 18 °C on gelatin/formaldehyde-coateddishes in artificial seawater. Embryos were collected and fixed in4% paraformaldehyde at various developmental stages (5–8 h,10–12 h). Template DNA plasmids were retrieved from theC. intestinalisGene Collection Release 1 library (16). Digoxigenin-labeled antisense RNA probes for ALDH1a, ALDH1b, ALDH1d,and ALDH2 were synthesized from clones GC38i10, GC38i10,GC30b02, and GC07d10, respectively. The ALDH1c probe wasgenerated by using PCR amplification of the last exon from C.intestinalis genomic DNA. Whole-mount in situ hybridization wasessentially performed as described (17) by using an increased hy-bridization temperature of 65 °C. Control experiments were car-ried out with sense riboprobes. Embryos were mounted inPermount on glass slides and expression was analyzed by usinga Leica DMR microscope.

1. Moore SA, et al. (1998) Sheep liver cytosolic aldehyde dehydrogenase: The structurereveals the basis for the retinal specificity of class 1 aldehyde dehydrogenases.Structure 6:1541–1551.

2. Mulder NJ, et al. (2003) The InterPro Database, 2003 brings increased coverage andnew features. Nucleic Acids Res 31:315–318.

3. Altschul SF, et al. (1997) Gapped BLAST and PSI-BLAST: A new generation of proteindatabase search programs. Nucleic Acids Res 25:3389–3402.

4. Edgar RC (2004) MUSCLE: Multiple sequence alignment with high accuracy and highthroughput. Nucleic Acids Res 32:1792–1797.

5. Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic inference undermixed models. Bioinformatics 19:1572–1574.

6. Abascal F, Zardoya R, Posada D (2005) ProtTest: Selection of best-fit models of proteinevolution. Bioinformatics 21:2104–2105.

7. Stamatakis A (2006) RAxML-VI-HPC: Maximum likelihood-based phylogeneticanalyses with thousands of taxa and mixed models. Bioinformatics 22:2688–2690.

8. Yang Z (1997) PAML: A program package for phylogenetic analysis by maximumlikelihood. Comput Appl Biosci 13:555–556.

9. Petrey D, et al. (2003) Using multiple structure alignments, fast model building, andenergetic analysis in fold recognition and homology modeling. Proteins 53(Suppl):430–435.

10. Laskowski RA, Moss DS, Thornton JM (1993) Main-chain bond lengths and bondangles in protein structures. J Mol Biol 231:1049–1067.

11. Humphrey W, Dalke A, Schulten K (1996) VMD: Visual molecular dynamics. J MolGraph 14:27–28, 33–38.

12. Richards FM (1974) The interpretation of protein structures: Total volume, groupvolume distributions and packing density. J Mol Biol 82:1–14.

13. Yu J-K, et al. (2007) Axial patterning in cephalochordates and the evolution of theorganizer. Nature 445:613–617.

14. Holland LZ, Holland PWH, Holland ND (1996) Revealing homologies between bodyparts of distantly related animals by in situ hybridization to developmental stages:Amphioxus versus vertebrates. In Molecular Zoology: Advances, Strategies, andProtocols, eds Ferraris JD, Palumbi SR (Wiley-Liss, New York), pp 267–282 and473–483.

15. Mita-Miyazawa I, Ikegami S, Satoh N (1985) Histospecific acetylcholinesterasedevelopment in the presumptive muscle cells isolated from 16-cell-stage ascidianembryos with respect to the number of DNA replications. J Embryol Exp Morphol 87:1–12.

16. Satou Y, et al. (2002) A cDNA resource from the basal chordate Ciona intestinalis.Genesis 33:153–154.

17. Christiaen L, et al. (2008) The transcription/migration interface in heart precursors ofCiona intestinalis. Science 320:1349–1352.

18. Klyosov AA (1996) Kinetics and specificity of human liver aldehyde dehydrogenasestoward aliphatic, aromatic, and fused polycyclic aldehydes. Biochemistry 35:4457–4467.


www.pnas.org/cgi/content/short/1011223108

ALDH 5

ALDH 7

ALDH 6ALDH 4

ALDH 3

ALDH 8

ALDH 9

ALDH 16

MetazoanALDH 1

MetazoanALDH 2

FungalALDH 1/2

EukaryoticALDH 1/2

MetazoanALDH 1L

0.3

1

0.790.

58

0.98

0.75

0.82

0.97

1

0.63

1

0.79

0.99

0.96

0.97

1

1

0.84

0.98

1

1

0.82

1

0.75

0.56

1

1

0.95

1

1

0.9

1

1

0.91

1

0.8

0.89

0.61

0.98

0.73

0.95

0.5

0.9

1

1

28.0

0.84

1

0.97

1

47.0

1

0.930.53

1

0.58

0.92

1

0.8

18.0

1

0.99

1

1

1

0.96

0.62

1

0.95

0.93

1

1

0.96

1

0.54

0.99

0.61

0.79

1

1

0.88

0.54

1

0.830.

84

0.98

1

1

1

0.99

1

0.99

0.84

1

0.64

1

0.62

0.63

0.64

0.99

11

1

0.86

1

0.66

10.51

0.51

0.51

1

0.81

O16

648_

ALD

H4F

1_S

c

O94788_ALDH1A2_RALDH2_Hs

07835_Ci

170078_Nv

2006

99_N

v

XP_786833_Sp

XP_7

8607

7_Sp

O74

766_

ALD

H4D

1_S

c

11750_Ci

CG96

29PA

_Dm

09381_Ci

AK

0958

27_A

LDH

1L2_

TFD

H2_

Hs

CG

3309

2PA

_Dm

89632_Nv

261794_Bf

ALDH

16A1

_Hs07939_C

i

P43353_ALDH3B1_Hs

P380

67_A

LDH5

C1_S

c

2850

95_B

f

XP_7

8915

1_Sp

XP_78

8850_S

p

XP_793

662_Sp

206892_Bf

2314

13_N

v

P00352_ALDH1A1_RALDH1_Hs

XP

_793

827_

Sp

Q02252_A

LDH

6A1_H

s

P465

62_A

LDH7

A2_C

eO44555_ALDH3C2_Ce

00550_Ci

Q9H2A2_ALDH8A1_Hs

P54115_A

LDH

1F2_S

c

Q9P7K9_ALDH15B1_Sc

192075

_Nv

106815_Bf

P49189

_ALDH

9A1_TM

ABAD

H_Hs

150665_Nv

P30837_ALDH1B1_ALDHXmito_Hs

P40047_A

LDH

1D1_S

c

AK112485_Ci

XP_7

9315

6x_S

p

113974_Bf

177644_Nv

P07

275_

ALD

H4B

1_S

c

Q19

428_

ALD

H1L

1_C

e

P494

19_A

LDH7

A1_A

TQ1_

Hs

09379_Ci

P46367_A

LDH

1D2_S

c102986_Nv

07300_Ci

O02266_ALDH5B1_Ce

Q9V4M1_ALDH3G1_Dm

XP

_784

777_

Sp

09471_Ci

127028_B

f

280571_Bf

XP_782957_Sp

86762_Bf

282069_Bf

162370

_Nv

O14

293_

ALD

H1S

1_S

c14829_Ci

XP_790286_Sp

XP

_783603_Sp

mD_1

E4H

DLA_4

XN

V9Q

XP

_782

169_

Sp

Q9VLC5_ALDH1A10_Dm

P30838_A

LDH

3A1_H

s

234970_B

f

CG

6661PA

_Dm

P05091_ALDH2_Hs

181421_Nv

O16518_ALDH3C1_Ce

P30

038_

ALD

H4A

1_H

s

P48448_ALDH3B2_Hs

Q9U

RW

9_A

LDH

1R1_

Sc

Q18822_ALDH8B1_Ce

XP_788528_Sp

07853_Ci

230530_Bf

Q9V

B96_A

LDH

1M1_D

m

1545

09_N

v

Q9UT

M8_A

LDH5

D1_S

c

2353

02_N

v

O75

891_

FTH

FD_A

LDH

1L1_

Hs

124720_Bf

Q9VBP6_ALDH4G1_Dm

09472_Ci

2145

34_B

f

XP_786787_Sp

Q20352_AL

DH9B1_Ce

0751

7_Ci

P51648_A

LDH

3A2_H

s

P47895_ALDH1A3_RALDH3_Hs05778_C

i

1149

17_B

f

cS_2

G1H

DLA_17774

P

1039

08_N

v

245626_Nv

XP_7

9781

5_Sp

P51649_ALD

H5A1_Hs

BW

049365_Ci

O59808_ALDH10C1_Sc

2695

54_B

f

Q04458_ALDH14_Sc

0807

6_C

i

179476_Nv

O46056_A

LDD

H6A

2_Dm

Q9TXM

0_ALDH1J2_Ce

Q20780_ALDH9B2_Ce

118947_Bf

P38694_ALDH15A1_Sc

XP_7892

39_Sp

Q9V

IC9_

ALD

H1L

1_D

m

1865

47_N

v1752

87_N

v

0265

8_C

i

0647

2_Ci

P52713_A

LDH

6A3_C

e

Fig. S1. Phylogram of a comprehensive phylogenetic analysis of the ALDH superfamily. Posterior probabilities are indicated at each node. Each individualALDH family is supported by significant posterior probabilities (>0.95). Bf, Branchiostoma floridae; Ce, Caenorhabditis elegans; Ci, Ciona intestinalis; Dm,Drosophila melanogaster; Hs, Homo sapiens; Nv, Nematostella vectensis; Sc, Saccharomyces cerevisiae; Sp, Strongylocentrotus purpuratus.



Metazo

an ALD

H 8

FungalA

LDH

1/2M

etazoan A

LDH

1LP

lantA

LDH

1/2M

etazoan A

LDH

1

RALDH 1

RALDH 2

RALDH 3

Metazo

an ALD

H 2

RALDH 4

1

0,99

1

0,65

0,6

1

1

1

0,86

1

1

0,95

1

0,95

1

0,51

1

0,92

1

1

0,82

0,76

1

1

0,98

0,52

1

0,64

1

1

0,58

1

1

1

1

1

0,9

1

1

1

1

1

0,83

1

0,99

1

0,98

1

1

0,53

1

0,64

1

1

1

1

0,84

0,68

1

1

1

1

1

1

1

1

0,91

0,721

0,76

0,74

1

0,77

1

0,92

1

1

1

1

1

1

0,97

1

1

1

1

0,56

1

1

0,68

11

0,5

1

0,76

0,54

0,77

0,98

1

0,72

0,72

0,64

0,67

0,98

0,86

1

0,98

0,65

1

133924_ALDH1/2e_Pc

Q18822_ALDH8B1_Ce

XP_786787_ALDH2b_Sp

XP_001061932_ALDH8A1_Rn

177987_ALDH2_Lg

Q9H2A2_ALDH8A1_ALLDH12_Hs

193864_ALDH8_Lg

Q19428_ALDH1L1_Ce

140859_ALDH1/2g_Pc

Q8S528_ALDH2B7_At

Q9FRX7_ALDH2B1_Os

39661_ALDH1/2b_Pb

192810_ALDH1b_Lg

09161_ALDH8_Cs

P40047_ALDH1D1_Sc

159056_ALDH1b_Ct

07853_ALDH1c_Ci

NP_082546_ALDH1B1_Mm

218707_ALDH1a_Lg

P47771_ALDH1G2_ScO14293_ALDH1S1_Ssp

Q9VLC5_ALDH1A10_Dm

P46367_ALDH1D2_Sc

Q9VIC9_ALDH1L1_Dm

133289_ALDH1/2_Pc

XP_793827_ALDH1La_Sp

NM_009656_ALDH2_Mm

ALDH2_Sk

ALDH1L_Sk

AAL99611_ALDH2D1_Zm

NM_053896_ALDH1A2_Rn

07835_ALDH1b_Ci

Q20780_ALDH9B2_Ce

Q9SU63_ALDH2B4_At

124720_ALDH1d_Bf

Q9LRI6_ALDH2B5_Os

823362_ALDH2B1_Pt

35266_ALDH1/2a_Pb

138693_ALDH1/2c_Pc

01705_ALDH2_Cs

XP_790286_ALDH8b_Sp

830473_ALDH2B3_Pt

ALDH1Ae_Sk

O94788_ALDH1A2_RALDH2_Hs

181421_ALDH1a_Nv


Q9LRE9_ALDH2C1_Os


AK095827_ALDH1L2_Hs

BW049365_ALDH1d_Ci

NM_053080_ALDH1A3_Mm

179476_ALDH2_Nv

137014_ALDH1/2a_Pc

Q43274_ALDH2B1_Zm

ALDH1Ab_Sk

08076_ALDH1L_Ci

NP_705771_ALDH1L2_Mm

07939_ALDH1a_Ci

Q9VB96_ALDH1M1_Dm

109073_ALDH8b_Ct

200699_ALDH1L_Nv

237306_ALDH1L_Lg

809215_ALDH1A1_Pt

269554_ALDH1L_Bf

XP_786833_ALDH2a_Sp

AK112485_ALDH2_Ci


245626_ALDH1b_Nv


ALDH1Ac_Sk

151890_ALDH1a_Ct

Q9TXM0_ALDH1J2_Ce

228352_ALDH8a_Ct

228199_ALDH1L_Ct

118947_ALDH1b_Bf


30289_ALDH1/2c_Pb

106815_ALDH2_Bf

AAL99609_ALDH2C1_Zm

XP_788528_ALDH8a_Sp

P54115_ALDH1F2_Sc

89632_ALDH8_Nv

140689_ALDH1/2f_Pc

XP_784777_ALDH1Lb_Sp

86762_ALDH1c_Bf

P05091_ALDH2_Hs

Q8BH00_ALDH8A1_Mm

AAM27004_ALDH1A1_At

AAH46315_ALDH1A7_MmNM_017272_ALDH1A4_Rn

P30837_ALDH1B1_ALDHXmito_Hs

ALDH1Aa_Sk

AAL99613_ALDH2B5_Zm

145966_ALDHc_Ct

14829_Ci

O75891_ALDH1L1_Hs

NP_081682_ALDH1L1_Mm


207312_ALDH1c_Lg

1350_ALDH1/2b_Pc

ALDH1Ad_Sk

01325_ALDH1b_Cs

04405_ALDH1L_Cs

261794_ALDH1a_Bf

113974_ALDH1e_Bf

282069_ALDH8_Bf

Q9URW9_ALDH1R1_Ssp

NM_032416_ALDH2_Rn

P93344_ALDH2B2_Nt

666446_ALDH2B2_Pt

NP_071992_ALDH1L_Rn

NM_001011975_ALDH1B1_Rn

206892_ALDH1f_Bf

03997_ALDH1a_Cs

183731_ALDH2_Ct

0.2

Fig. S2. Phylogram of a phylogenetic analysis of the ALDH1/2, ALDH1L, and ALDH8 families with the latter as outgroup. Posterior probabilities are indicatedat each node. Each individual ALDH family is supported by significant posterior probabilities (>0.95). At, Arabidopsis thaliana; Bf, Branchiostoma floridae;Ct, Capitella teleta; Ce, Caenorhabditis elegans; Ci, Ciona intestinalis; Cs, Ciona savignyi; Dm, Drosophila melanogaster; Hs, Homo sapiens; Lg, Lottia gi-gantea; Mm, Mus musculus; Nv, Nematostella vectensis; Nt, Nicotiana tabacum; Os, Oryza sativa; Pc, Phanerochaete chrysosporium; Pb, Phycomyces bla-kesleeanus; Pt, Populus trichocarpa; Rn, Rattus norvegicus; Sc, Saccharomyces cerevisiae; Sk, Saccoglossus kowalevskii; Sp, Strongylocentrotus purpuratus;Ssp, Schizosaccharomyces pombe; Zm, Zea mays.



Distribution clusters of ALDH1/2 volumes

Volumes (Å³)

Fig. S3. ALDH1/2 clusters according to substrate access channel volume distribution. Average and SD values for the three clusters are 592.29 ± 42.07 for largechannels, 453.33 ± 27.39 for medium channels, and 354.79 ± 32.50 for small channels.

Gly Cys

CysSer

Ile Cys

Gly ThrPhe

Phe

Met

Leu

Leu CysPhe

ALDH1aPos 124Pos 459Pos 303Vol (Å³)

GlyPheThr640

ALDH1dPos 124Pos 459Pos 303Vol (Å³)

ALDH1cPos 124Pos 459Pos 303Vol (Å³)

ALDH1bPos 124Pos 459Pos 303Vol (Å³)

SerPheCys514

IleMetCys467

GlyLeuCys371

ALDH2Pos 124Pos 459Pos 303Vol (Å³)

LeuPheCys331

5-8 hours 10-12 hoursPosition124Position

459Position

303Position 124+ β-ionone

BA C

Fig. S4. Ciona intestinalis ALDH1/2 duplicates. Phylogeny (A), channel structure (B), and developmental expression (C). Amino acid signatures of the substrateentry channel at positions 124 (the mouth), 459 (the neck), and 303 (the bottom) are indicated. For the expression analyses, early embryos (5–8 h) and tailbudstage embryos (10–12 h) are shown. (Scale bars: 50 μm.)



Position 124 Position 459 Position 303Position 124+ β-ionone

Ala CysMet

AlaMetCys561

ALDH1bPos 124Pos 459Pos 303Vol (Å³)

Leu CysPhe

ALDH2LeuPheCys321

Pos 124Pos 459Pos 303Vol (Å³)

Gly ThrPhe

ALDH1aGlyPheThr546


Fig. S5. Phylogenetic and structural analysis of ALDH1/2s from the ascidian tunicate Ciona savignyi. The two ALDH1 duplicates from C. savignyi display het-erogeneous properties at their substrate entry channels. Although ALDH1a and ALDH1b both display bulky amino acids at position 459, corresponding to thechannel neck, ALDH1a displays the smallest amino acid (Gly) at position 124, which corresponds to the channel mouth. In contrast, in ALDH1b, this residue issubstituted by the slightly larger Ala, which creates an obstacle to accommodate the large β-ionone moiety of retinaldehyde. Moreover, in ALDH1a, the aminoacid at position 303 is Thr, which is typical for ALDH1s, whereas in ALDH1b, this amino acid is Cys, which is typical for ALDH2s. This pattern suggests that in C.savignyi, ALDH1a is best adapted for retinoic acid synthesis and that ALDH1b incorporated structural features reminiscent of ALDH2s, which process smaller, toxicaldehydes.



Branchiostoma floridae

Scaffold_560: 425 283 bp


Scaffold_21: 3 715 217 bp

Scaffold_31: 3 505 904 bp

Scaffold_118: 1 970 888 bp

Scaffold_155: 1 674 870 bp

Scaffold_444: 671 902 bp120 kb

ALDH2

ALDH1a

ALDH1b

ALDH1c

ALDH1d

ALDH1e

ALDH1f

ALDH2

ALDH1

Ciona intestinalis




ALDH1

ALDH2

ALDH1a Ciona intestinalis

ALDH1a Ciona savignyi

ALDH1b Ciona intestinalis

ALDH1b Ciona savignyi

ALDH1c Ciona intestinalis

ALDH1d Ciona intestinalis

ALDH2 Ciona intestinalis

ALDH2 Ciona savignyi

20 kb

A

B

Fig. S6. Genomic linkage of ALDH1 and ALDH2 genes in the cephalochordate amphioxus and in the ascidian tunicate Ciona intestinalis. The position ofpredicted ALDH1 and ALDH2 sequences on assembled scaffolds of the amphioxus (A) and the C. intestinalis (B) genome are shown. (A) The phylogeneticrelationship between the six lineage-specific amphioxus ALDH1 duplicates and the single amphioxus ALDH2 is indicated. For ALDH2, the analysis identified twocopies in the amphioxus genome, one on scaffold 550 and one on scaffold 118, representing a single copy of amphioxus ALDH2 on each of the two ho-mologous chromosomes. For the amphioxus ALDH1 duplicates, the analysis suggests that at least some genes, like ALDH1c and ALDH1d (on scaffold 155),ALDH1a and e (on scaffold 560) or ALDH1a, ALDH1e, and ALDH1f (on scaffold 31) are located on the same scaffold and might, hence, be linked on thechromosome. In sum, the phylogeny and synteny data suggest that the six amphioxus-specific ALDH1 duplicates evolved by tandem duplications from anALDH1a-like ancestor. (B) The phylogenetic relationship between C. intestinalis and C. savignyi ALDH1s and ALDH2s is indicated. In the C. intestinalis genome,ALDH2 and ALDH1a are found on individual scaffolds (on scaffold 184 and 18, respectively), whereas the other three C. intestinalis ALDH1 duplicates, ALDH1b,ALDH1c, and ALDH1d, are clustered on a single scaffold (on scaffold 112), suggesting that these three genes are linked on the same chromosome and mightthus have originated by tandem duplications.



Met Phe Cys

Glu Val Cys

Gly

Leu Phe

Val

Cys

Cys

Position 124 Position 459 Position 303Position 124+ β-ionone

GluValCys453

ALDH1B1H. sapiensPos 124Pos 459Pos 303Vol (Å³)

ALDH2aS. purpuratus

LeuPheCys356


ALDH2bS. purpuratus

GlyValCys487


ALDH2H. sapiens

MetPheCys377


Fig. S7. Plasticity of the substrate access channel in deuterostomes. Vertebrate and echinoderm ALDH2s are examples for the evolutionary plasticity ofALDH1/2 channels. In vertebrates, there are two members of the ALDH2 clade: the ALDH2s and the ALDH1B1s. Vertebrate ALDH2s have conserved the aminoacid signatures (Met124, Phe459) for shaping a narrow entrance (the mouth) and a tight-fitting proximal third (the neck) of the substrate access channel. TheALDH1B1s incorporated smaller Glu124 and Val459, which offer less resistance for accommodating the β-ionone moiety of retinaldehyde and alleviate con-striction at the channel neck, hence increasing the overall channel volumes (362 ± 30 Å3 versus 451 ± 30 Å3 for ALDH2 and ALDH1B1, respectively, P < 0.05).A similar trend is observed for sea urchin (Strongylocentrotus purpuratus) ALDH2s. S. purpuratus ALDH2a displays the typical ALDH2 pattern with bulky Leu124and Phe459 at the channel mouth and neck, respectively. In contrast, S. purpuratus ALDH2b incorporated small amino acids (Gly124 and Val459), which leavethe channel mouth wide open to accommodate the β-ionone moiety of retinaldehyde and alleviate constriction at the channel neck thus increasing the volumefrom 356 Å3 in ALDH2a to 487 Å3 in ALDH2b.



124

459

303

124

459

303

Large channelWide entrance

Small channelNarrow entrance

ALDH1 ALDH2

Functionalshifts

Fig. S8. Gene duplication and functional evolution in the ALDH1/2 family. Metazoan ALDH1s are generally characterized by a large substrate entry channel(SEC) with a wide channel entrance, whereas metazoan ALDH2s have a small SEC with a narrow channel entrance (the three channel signatures important fordefining the ALDH1 and ALDH2 SECs are indicated: position 124 located at the channel mouth, position 459 at the neck of the channel, and position 303 at thechannel bottom). After duplication, ALDH1/2 duplicates accumulated mutations in their SECs that led to smaller channel sizes in ALDH1 duplicates (e.g., am-phioxus and ascidian tunicate ALDH1s) and to bigger channel volumes in ALDH2 duplicates (e.g., vertebrate ALDH1B1 and sea urchin ALDH2). Hence, duplicatedALDH1/2s experienced functional shifts of their SECs, which are indicative of changes in specificity toward aldehydes of different sizes.



Table S1. Sequence signatures of vertebrate ALDH1 and ALDH2 enzymes

Amino acidsubstitution

Vertebratesignature

Cephalochordatesignature

Tunicatesignature Type of substitution

Structural locationof the signature

Thr-104-Ala No No Polar to small aliphatic apolar Exposed surface (α helix-B)

Gly-110-Asn No Yes Small aliphatic apolar to polar Buried residue (α helix-B)

Asn-116-Ile No Yes Polar to aliphatic apolar Buried residue (α helix-C)

Gly-123-Asp No No Polar to negatively charged Channel (α helix-C)

Gly-124-Met No No Small aliphatic apolar to largealiphatic

Channel mouth (α helix-C)

Thr-128-Cys Yes Yes Polar to polar thiol-containing Buried residue (α helix-C)

Phe-175-Gln Yes Yes Bulky aromatic to polar Buried residue (α helix-D)

Ser-184-Ala No No Polar to small aliphatic apolar Buried residue (α helix-D)

Cys-185-Thr Yes Yes Thiol-containing to polar Buried residue (α helix-D)

Thr-188-Val Yes Yes Polar to aliphatic apolar Buried residue (β sheet-8)

Val-191-Met Yes Yes Aliphatic apolar to large aliphaticapolar

Buried residue (β sheet-8)

Pro-193-Val Yes Yes Apolar cyclic to aliphatic apolar Buried residue (β sheet-8)

Lys-251-Arg/His No No Positively charged to largerpositively charged imidazole

Channel (α helix-G)

Glu-255-Val No No Negatively charged to aliphaticapolar

Oligomerization surface(α helix-G)

Ala-279-Ser No Yes Small aliphatic apolar to polar Exposed surface (connectingloop between β sheet-12

and α helix-H)

Asn-285-Trp No No Polar to large aromatic Exposed surface (α helix-H)

Gln-292-Phe Yes Yes Polar to aromatic Exposed surface (α helix-H)



Table S1. Cont.


Vertebratesignature




Gly-293-Ala No No Small aliphatic apolar to smalleraliphatic apolar

Buried residue (α helix-H)

Ile-303-Cys No No Aliphatic apolar tothiol-containing

Channel bottom (connectingloop between α helix-H and β

sheet-13)

Glu-311-Gln No No Negatively charged to polar Exposed surface (β sheet-13)

Ser-313-Asp No No Polar to negatively charged Exposed surface (connectingloop between β sheet-13

and α helix-I)

Arg-320-Glu No No Positively charged to negativelycharged

Exposed surface (α helix-I)

Asp-355-Gly No No Negatively charged to smallaliphatic apolar

Exposed surface (α helix-J)

Leu-356-Tyr No No Aliphatic apolar to large aromatic Exposed surface (α helix-J)

Glu-368-Leu No Yes Negatively charged to largealiphatic apolar

Buried residue (β sheet-16)

Ser-387-Gly No Yes Polar to small aliphatic apolar Exposed residue (β sheet-14)

Arg-394-Thr No No Positively charged to polar Exposed residue (α helix-K)

Gln-405-Met No No Polar to large aliphatic apolar Buried residue (β sheet-16)

Ile/Val/Leu-440-Asp No No Aliphatic apolar to negativelycharged

Oligomerization surface(α helix-M)

Thr-441-Tyr Yes No Polar to large aromatic Exposed surface (α helix-M)

Ser-444-Gln No No Polar to larger polar Exposed surface (α helix-M)



Table S1. Cont.


Vertebratesignature




Ser-457-Asp No No Polar to negatively charged Channel (connecting loop betweenβ sheet-18 and α helix-N)

Val-459-Phe Yes No Aliphatic apolar to aromatic Channel neck (connecting loopbetween β sheet-18 and α helix-N)

Ser-460-Gly No No Polar to small aliphatic apolar Exposed surface (connecting loopbetween β sheet-18 and α helix-N)

Signatures are displayed as motifs with 4–10 amino acids. Motifs are shown as pictograms with relative frequencies symbolized by amino acid height asdetermined in an alignment of 24 vertebrate ALDH1s and 12 vertebrate ALDH2s. Represented here are only amino acids that effectively distinguish vertebrateALDH1s from ALDH2s. Species-specific and vertebrate ALDH1A paralog-specific signatures are not included. The amino acids that are not conserved betweenALDH1s and ALDH2s are surrounded by a red box, and their localization along with the nature of the substitutions between ALDH1s and ALDH2s is displayed.The ability of each vertebrate signature to distinguish ALDH1s from ALDH2s in invertebrate chordates (amphioxus and ascidian tunicates) is also shown.

Table S2. Reconstruction of ancestral sequences in the ALDH1/2 family

Channel size

Node Support* Sequence description P† Signature‡ Volume, Å3 Category

A 1/75 Eukaryote ALDH1/2 ancestor 0.915 Met,Phe,Cys 378.4 SmallB 0.98/59 Fungal ALDH1/2 ancestor 0.918 Met,Phe,Cys 336.5 SmallC 1/94 Plant/Metazoan ALDH1/2 ancestor 0.937 Met,Phe,Cys 287.4 SmallD 1/100 Plant ALDH1/2 ancestor 0.912 Met,Phe,Cys 286.0 SmallE 1/100 Plant ALDH1-like ancestor 0.914 Ala,Phe,Val 559.2 BigF 1/100 Plant ALDH2-like ancestor 0.917 Met,Phe,Cys 333.8 SmallG 1/65 Metazoan ALDH1/2 ancestor 0.948 Met,Phe,Cys 368.8 SmallH 1/69 Metazoan ALDH2 ancestor 0.966 Met,Phe,Cys 444.9 MediumI 1/92 Vertebrate ALDH1 ancestor 0.956 Gly,Leu,Thr 604.3 Large

*Phylogenetic support for the node, where the ancestral sequence was reconstructed (posterior probability/bootstrap value).†Mean probability of the reconstructed ancestral sequence.‡The order of the amino acids corresponds to the mouth, neck, and bottom signatures, respectively.

Dataset S1. Sequences used for the comprehensive phylogenetic analysis of the ALDH superfamily

Dataset S1 (XLS)

Accession numbers and database information are given for each sequence.

Dataset S2. Sequences used for the phylogenetic analysis of the ALDH1/2, ALDH1L, and ALDH8 families

Dataset S2 (XLS)

Accession numbers and database information are given for each sequence. In addition, channel volumes and channel size categories are indicated for ALDHenzymes with sufficient sequence information and structural compatibility with the crystal templates human ALDH2 and sheep ALDH1A1 (Protein Data BankID codes 1OF7 and 1BXS, respectively).


http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1011223108/-/DCSupplemental/sd01.xlshttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1011223108/-/DCSupplemental/sd02.xlswww.pnas.org/cgi/content/short/1011223108

Movie S1. Six-nanosecond simulation of substrate behavior within the ALDH2 substrate entry channel. Acetaldehyde (shown in orange), a natural substrateof ALDH2, was guided into catalysis position, 3.45 Å from the γ-sulfur (shown in yellow) of the catalytic Cys302 and 4.77 Å from the oxidized NAD (ball and stickconfiguration, bottom left) by using the crystal coordinates for ALDH2 plus crotonal (Protein Data Bank ID codes 1OF7 and 10O1). Amino acid signatures at theALDH2 channel mouth, neck, and bottom are, respectively, represented in blue (Leu124), green (Phe459), and purple (Cys303) in van der Walls configurations.Phe465 (shown in white) is conserved in metazoan ALDHs and keeps acetaldehyde in position throughout the simulation. Phe465 is stabilized by Phe459, whichis the vertebrate ALDH2 neck signature. Phe459 is latched in position by Cys303, the vertebrate ALDH2 bottom signature. Thus, concerted action of vertebrateALDH2 signatures at channel bottom and neck, together with a conserved ALDH channel amino acid, keeps acetaldehyde close enough to Cys302 and the NADacceptor to guarantee efficient catalysis. The dashed line highlighting the distance between the γ-sulfur of Cys302 and the C1 of acetaldehyde is obscured bythe close proximity of these two atoms.

Movie S1

Movie S2. Six-nanosecond simulation of substrate behavior within the ALDH1 substrate entry channel. Acetaldehyde (shown in orange) was guided intoposition, 3.45 Å from the γ-sulfur (shown in yellow) of the catalytic Cys302 and 4.77 Å from the oxidized NAD (ball and stick configuration, bottom left) byusing the ALDH1A1 (Protein Data Bank ID code 1BXS) and ALDH2 plus crotonal (Protein Data Bank ID codes 1OF7 and 10O1) crystal coordinates. Amino acidsignatures at the ALDH1 channel mouth, neck, and bottom are, respectively, represented in blue (Gly124), green (Val459), and purple (Ile303) in van der Wallsconfigurations. Phe465, conserved in metazoan ALDHs, overlies acetaldehyde and is represented in white. In ALDH1, acetaldehyde moves away from itsposition close to the catalytic Cys302 and the NAD acceptor. The acetaldehyde escape is preceded by a pendulum-like Phe465 swing toward the channel mouth,which releases the constraints on the underlying acetaldehyde. This swing is preceded by quick disengagement of the bottom and neck amino acids Ile303 andVal459, respectively, which are not bulky enough to reconstitute the interaction surfaces maintained by the typical ALDH2 amino acids Phe459 and Cys303. Theconcerted latching of Phe465 by bottom and neck channel amino acids is thus lost in ALDH1s, which fail to maintain substrate position stable enough forefficient catalysis, which is consistent with their large Km values for acetaldehyde. The dashed line highlights the distance between the γ-sulfur of Cys302 andthe C1 of acetaldehyde.

Movie S2


http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1011223108/-/DCSupplemental/sm01.avihttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1011223108/-/DCSupplemental/sm02.aviwww.pnas.org/cgi/content/short/1011223108

Movie S3. Six-nanosecond simulation of formaldehyde, the smallest (C1) aldehyde (shown in orange) inside the ALDH1 substrate entry channel. Formal-dehyde was guided into position, 3.45 Å from the γ-sulfur (shown in yellow) of the catalytic Cys302 and 4.77 Å from the oxidized NAD (ball and stick con-figuration, bottom left) by using the ALDH1A1 (Protein Data Bank ID code 1BXS) and ALDH2 plus crotonal (Protein Data Bank ID codes 1OF7 and 10O1) crystalcoordinates. Amino acid signatures at the ALDH1 channel mouth, neck, and bottom are, respectively, represented in blue (Gly124), green (Val459), and purple(Ile303) in van der Walls configurations. Phe465, conserved in metazoan ALDHs, overlies the formaldehyde and is represented in white. Note that formal-dehyde cannot be kept inside the ALDH1 channel and, thus, migrates toward the channel mouth, which is consistent with the inability of ALDH1s to processformaldehyde. The dashed line highlights the distance between the γ-sulfur of Cys302 and the C1 of formaldehyde.

Movie S3

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1011223108/-/DCSupplemental/sm03.aviwww.pnas.org/cgi/content/short/1011223108

Date post:	11-Feb-2021
Category:	Documents
Upload:	others
View:	0 times
Download:	0 times

Supporting Information - CaltechAUTHORSangles in protein structures. J Mol Biol 231:1049–1067. 11....

Documents