The mitochondrial transcription and packaging factor Tfam imposes

1290 VOLUME 18 NUMBER 11 NOVEMBER 2011 nature structural & molecul ar biology

a r t i c l e s

Among eukaryotic protein complexes, the mitochondrial oxidative

phosphorylation (OXPHOS) machinery is unique in having a

bigenomic origin. Most of the OXPHOS machinery is encoded by

the nuclear genome, but 13 essential subunits of respiratory chain

complexes I, III, IV and V are encoded by the 16-kb mitochondrial

genome. Tfam (also known as mtTFA), a DNA-binding protein in

mitochondria, is a central player in expression and maintenance of

mitochondrial DNA (mtDNA), and therefore is essential for ATP

production via OXPHOS

1,2

. The mammalian mitochondrial genome

contains three promoters—the light strand promoter (LSP), the

heavy strand promoter 1 (HSP1) and the heavy strand promoter 2

(HSP2)—that drive expression of mtDNA transcripts. Transcription

from LSP and HSP1 has been reconstituted in vitro, and normal

levels of transcription require Tfam

1–4

. Moreover, because truncated

RNA transcripts from LSP are used to prime DNA replication, Tfam

is secondarily essential for mtDNA replication. Mice lacking Tfam

therefore show impaired mtDNA transcription and loss of mtDNA,

leading to bioenergetic insufficiency and embryonic lethality

Upstream of both the LSP and HSP1 transcriptional start sites,

Tfam recognizes a binding site that has been defined by DNase I

footprinting experiments

3,4

. Tfam contains two HMG-box domains

followed by a short C-terminal tail

. HMG-box domains are DNA-

binding motifs that bind to the minor groove of DNA and, in some

cases, result in DNA bending

. Tfam belongs to the HMG-box sub-

group that contains tandem HMG-box domains

. Several proteins

in this subgroup, such as Tfam, have important structural roles in

DNA organization, but there is currently no information about how

two HMG-box domains can be spatially coordinated to affect DNA

structure. The C-terminal tail of Tfam is essential for transcriptional

activation

and also for its physical association with Tfb2m

, another

transcription factor required for mtDNA transcription. As a result,

it has been proposed that Tfam binding allows recruitment of Tfb2m

by the C-terminal tail.

In addition to its transcriptional function, Tfam is thought to have

an important role in mtDNA packaging

10,11

. Although Tfam functions

as a sequence-specific transcription factor, it also has high affinity

for nonspecific DNA. Unlike nuclear DNA, mtDNA is not associated

with histones. mtDNA genomes within the mitochondrial matrix are

organized into compact DNA–protein complexes called nucleoids

Tfam is one of the most abundant proteins associated with mtDNA

nucleoids

, and its levels have been estimated to be sufficient to coat

the entire mitochondrial genome

. The levels of Tfam correlate with

the levels of mtDNA

. The yeast ortholog of Tfam, ARS-binding

factor 2, mitochondrial (Abf2), has no role in transcription, and its

major function is thought to be in the organization of the mitochon-

drial genome

To understand how Tfam mediates these multiple functions on

mtDNA, we have solved the structure of human Tfam in complex with

the LSP binding site. The structure shows how Tfam coordinates its

two HMG-box domains to impose a dramatic U-turn on the DNA. To

bend DNA, Tfam uses structural principles analogous to those used

by the HU family of prokaryotic nucleoid proteins, which, like Tfam,

have architectural roles in genome organization. Moreover, we find

this DNA bending is more important for transcriptional activation

at LSP than HSP1.

RESULTS

Structure determination

We solved the 2.5-Å crystal structure of human Tfam bound to a

28-bp DNA fragment derived from LSP (Table 1 and Fig. 1a–d).

Division of Biology, California Institute of Technology, Pasadena, California, USA.

Division of Chemistry, California Institute of Technology, Pasadena, California, USA.

Howard Hughes Medical Institute, California Institute of Technology, Pasadena, California, USA. Correspondence should be addressed to D.C.C. ([email protected]).

Received 20 June; accepted 13 September; published online 30 October 2011; doi:10.1038/nsmb.2159

The mitochondrial transcription and packaging factor

Tfam imposes a U-turn on mitochondrial DNA

Huu B Ngo

, Jens T Kaiser

& David C Chan

1,3

Tfam (transcription factor A, mitochondrial), a DNA-binding protein with tandem high-mobility group (HMG)-box domains,

has a central role in the expression, maintenance and organization of the mitochondrial genome. It activates transcription from

mitochondrial promoters and organizes the mitochondrial genome into nucleoids. Using X-ray crystallography, we show that

human Tfam forces promoter DNA to undergo a U-turn, reversing the direction of the DNA helix. Each HMG-box domain

wedges into the DNA minor groove to generate two kinks on one face of the DNA. On the opposite face, a positively

charged a-helix serves as a platform to facilitate DNA bending. The structural principles underlying DNA bending converge

with those of the unrelated HU family proteins, which have analogous architectural roles in organizing bacterial nucleoids.

The functional importance of this extreme DNA bending is promoter specific and seems to be related to the orientation of

Tfam on the promoters.

nature structur al & molecular biology VOLUME 18 NUMBER 11 NOVEMBER 2011 1291

The recombinant Tfam used (residues 43–246) represents the full-

length, mature Tfam after cleavage of the N-terminal mitochondrial

leader sequence

. The DNA fragment includes a ~22-bp sequence

that was identified as a high-affinity Tfam-binding site by DNase I

footprinting and has two half-sites that interact with the HMG-box

domains

. A selenomethionine-substituted Tfam–mtDNA complex

was used for structure determination at 2.5 Å by multiwavelength

anomalous diffraction (MAD) analysis. The crystallographic

statistics of data collection and refinement are presented in Table 1.

The electron density map was of sufficient quality to build almost

all of the protein (residues 43–237) and all 28 base pairs of mtDNA.

Model building and refinement produced a final structure with

excellent stereochemistry, with an R

free

of 24.7% and an R

work

of 19.8%.

This crystal structure is the first one of a native tandem HMG-box

protein in complex with DNA. In a previous study

, the NMR struc-

ture was solved of a chimeric molecule consisting of the HMG-box

domain of Sex-determining region Y protein (SRY, a single HMG-

box protein) fused to one of the two HMG-box domains of HMGB1

(a tandem HMG-box protein). This artificial molecule is nonphysi-

ological, and its structure in complex with DNA does not resemble

the structure described here.

Tfam imposes a severe bend on LSP mtDNA

The most striking feature of the structure is that binding of a Tfam

monomer dramatically distorts the DNA into a U-shape, causing a

reversal in the direction of the DNA helical axis (Fig. 1c,d). Each

HMG-box folds into a three-helix motif

with a concave surface that intercalates

between the bases in the minor groove of

an LSP half-site (Fig. 1c). These two inter-

calations result in two sharp kinks on one

face of the DNA helix. The buried contact

area of the first HMG-box domain (box A)

with DNA is 1,566 Å

, and the correspond-

ing surface area of the second HMG-box

(box B) is nearly as extensive at 1,404 Å

(Fig. 1e,f). The linker connecting the two

HMG-box domains forms an α-helix around

which the DNA wraps (contact area 864 Å

)

(Fig. 1c,d,f). As described in detail later,

basic side chains in the linker interact with

the negatively charged phosphates in the

bent DNA backbone. The C-terminal tail

also contacts DNA (580 Å

), and the first

part of this region extends the third helix of

the second HMG-box domain. Therefore,

all four regions (Fig. 1a) of Tfam—the two

HMG-box domains, the linker and the

C-terminal tail—make extensive contact

with the DNA.

The structure agrees well with previ-

ous DNase I footprinting and methyla-

tion interference experiments probing the

binding of Tfam to LSP DNA

3,18

. The Tfam

monomer accounts for the large recognition

site identified by a combination of DNase I

footprinting and sequence analysis

3,4,8

. Each

HMG-box domain binds to one of the two

half-sites identified by sequence analysis

In previous methylation interference experi-

ments

, the methylation by dimethylsulfate (DMS) of selected

adenines was associated with reduced binding of Tfam. DMS meth-

ylates adenine at the N3 atom, which is located within the minor

groove and would sterically block subsequent Tfam binding. In

our crystal structure, all of the adenines identified by Clayton and

colleagues

reside in a position where Tfam contacts the DNA minor

groove and causes widening (Supplementary Fig. 1). By contrast,

methylation of adenines located outside the contact area did not

affect Tfam binding.

Our crystal structure indicates that Tfam binds mtDNA as a mono-

mer. Human Tfam without DNA is monomeric, but it has been sug-

gested that Tfam assembles into dimers on DNA binding

. The latter

conclusion is tenuous, because it was based on a gel mobility assay

that used extremely high concentrations of Tfam and DNA and does

not give a definitive assessment of stoichiometry. To independently

test the 1:1 stoichiometry found in our crystal structure, we analyzed

Tfam and the Tfam–mtDNA complex in solution by size exclusion

chromatography with in-line multi-angle light scattering analysis

(SEC-MALS). The measured molar masses indicated that Tfam in

isolation is monomeric and, when complexed with DNA, forms a 1:1

complex (Fig. 2).

Protein-DNA interactions

As in other HMG-box structures, each Tfam HMG-box domain folds

into an L shape composed of three α-helices, with the third helix

forming the long axis (Fig. 1c). A hydrophobic core composed of

Tyr57, Phe60, Trp88 and Tyr99 stabilizes the L-shaped configuration

Table 1 Data collection, phasing and refinement statistics

Crystal 1

Crystal 2

Data collection

Space group C222

C222

Cell dimensions

a, b, c (Å) 68.36, 81.35, 160.63 68.44, 81.91, 161.25

(°) 90, 90, 90 90, 90, 90

Peak Inflection Remote

Resolution (Å) 28.8–3.0

(3.17–3.0)

28.8–3.0

(3.17–3.0)

28.8–3.0

(3.17–3.0)

24.2–2.5

(2.64–2.50)

merge

0.048 (0.120) 0.046 (0.114) 0.049 (0.215) 0.060 (0.454)

I / σI 21.9 (9.7) 23.3 (10.0) 21.3 (5.8) 12.2 (2.8)

Completeness 97.7 (91.4) 97.4 (88.8) 98.5 (97.4) 98.3 (99.5)

Redundancy 5.8 (4.2) 5.8 (4.0) 5.8 (4.6) 3.5 (3.6)

Refinement

Resolution (Å) 24.2–2.5

No. reflections 15,795

work

/ R

free

19.8 / 24.7

No. atoms

Total 2,888

Protein 1,641

DNA 1,148

Water 80

B-factors (Å)

Protein 59.1

DNA 68.7

Water 55.3

R.m.s. deviations

Bond lengths (Å) 0.007

Bond angles (°) 1.03

Two crystals were used for the structure.

Values in parentheses are for highest-resolution shell.

a r t i c l e s

1292 VOLUME 18 NUMBER 11 NOVEMBER 2011 nature structur al & molecular biology

a r t i c l e s

of the first HMG-box (Fig. 3a; Supplementary Fig. 2). Similarly,

buried residues Tyr162, Tyr165, Trp189 and Tyr200 stabilize the

second HMG-box domain (Fig. 3b; Supplementary Fig. 2). The

overall folds of both HMG-box domains superimpose well with

other HMG boxes whose structures in complex with DNA have been

solved (Fig. 3c).

In the Tfam–mtDNA complex, most of the side-chain–DNA con-

tacts are not sequence specific and occur on the sugar-phosphate

backbone of the DNA. However, a small number of contacts to bases

within the minor groove can be seen. HMG-box domains generally

contain one or two hydrophobic residues that intercalate into the

minor groove (highlighted in Supplementary Fig. 3). Consistent

with this generalization, the HMG-box A of Tfam contains the first

of these intercalating residues at position 58 (Leu58), which inter-

acts with A8 (strand B) (Fig. 3d, red residue). A previous crystal

structure of the isolated, HMG-box B of Tfam raised the issue of

whether it was a noncanonical HMG-box domain with unusual

binding properties

, because it seemed to lack both intercalat-

ing hydrophobic residues. Our Tfam–mtDNA structure clarifies

this issue by showing that HMG-box B does contain DNA-binding

residues at these same positions, even though the residues are not

nonpolar. In the first position, HMG-box B contains Asn163, which

reaches into the minor groove and contacts the underlying thymine

(T7, strand A). In the second position, Pro178 similarly inserts into

the minor groove and contacts a guanine (G9, strand A) (Fig. 3e, red

residues). In comparison to the previous structure of HMG-box B

without DNA

, Pro178 has shifted >2 Å to make this contact with

the DNA base.

Besides the interactions indicated above, several other contacts

to DNA bases are apparent. In HMG-box A, contacts are observed

between Ile81 and T19 (strand A), Tyr57 and G20 (strand A), and

Ser61 and G20 (strand A) (Fig. 3d; Supplementary Fig. 2c). In

addition, Ser61 and Ser55 indirectly interact with C9 (strand B)

and T21 (strand A), respectively, through water molecules. In

the HMG-box B (Fig. 3e; Supplementary Fig. 2d), contacts are

observed between Arg157 and T24 (strand B), and Gln179 and

C19 (strand B). The linker does not directly interact with DNA

bases. However, it makes substantial contacts with DNA via charged

or polar interactions (Fig. 3f,g). Lys147 contacts G16 (strand A).

His137 and Arg140 both make contacts to the phosphate back-

bone. Other lysine residues in the linker region (Lys136, Lys139

and Lys146) make longer-range contacts (>3.35 Å) with the sugar-

phosphate backbone.

Similarity to HU and IHF nucleoid proteins

The conformations of the two half-sites bound by Tfam deviate sub-

stantially from canonical B-DNA (Fig. 4a–d). At each location, inter-

calation by the HMG box results in substantial widening of the minor

groove (Fig. 4a). There is local DNA unwinding, as indicated by sharp

e f

43 122

Strand B

Strand A

90°

180°

Transcriptional start

HMG-box A HMG-box B + C-tail HMG-box B + C-tailHMG-box ALinker Linker

152 222

HMG-box A HMG-box B C-tailLinker

246

HSP LSP

Tfam

419446

Strand A

Strand B

5′

Tfam

Figure 1 Overview of the Tfam–mtDNA complex.

(a) The domain structure of mature Tfam. Residues

1–42 constitute the mitochondrial targeting

sequence that is cleaved upon import of Tfam

into the mitochondrial matrix. (b) Organization

of the LSP and HSP1 promoters. Comparative

sequence analysis showed that the two Tfam

binding sites are oriented in opposite directions

relative to the direction of transcription

4,8

. The

sequence of the LSP DNA fragment used for

crystallization is indicated. (c) Side view of the

Tfam–mtDNA complex. The Tfam domains are

color coded as in a, and DNA is colored in gray.

The LSP transcriptional start site would be located

away from the DNA end on the left, as indicated

by the arrow. Note that HMG-box B binds to the

half-site further away from the transcriptional start

site. (d) A view of the Tfam–mtDNA complex from

the top. The protein and DNA are color coded

as in c. (e) Electrostatic surface potential plot of

Tfam. Surface areas of Tfam that are buried on

DNA binding are highlighted in yellow mesh. The

HMG-box A, linker, HMG-box B and C-terminal tail

(C-tail) regions are labeled. Regions of negative

electrostatic potential are indicated in red and

regions of positive electrostatic potential in blue.

(f) Electrostatic surface potential plot of Tfam,

viewed in the same orientation as in d and flipped

180° from e. This view emphasizes that the

surface of the linker contacts the DNA.

0.8

0.6

8.0 8.5 9.0 9.5

Volume (ml)

10.0

31 kDa

45 kDa

Molar mass (Da)

66 kDa

1 × 10

8 × 10

6 × 10

4 × 10

2 × 10

Tfam + mtDNA

Tfam

BSA

10.5 11.0 11.5

Rayleigh ratio (cm

–1

) (10

–4

)

0.4

0.2

Figure 2 Molecular mass of the Tfam–mtDNA complex determined by

SEC-MALS. Elution profiles of Tfam, the Tfam–mtDNA complex and

BSA (control) examined by SEC-MALS. The horizontal black, red and

blue lines correspond to SEC-MALS calculated masses for BSA, Tfam

and Tfam-mtDNA, respectively. The corresponding theoretical masses

are 28,075 Da (Tfam), 45,410 Da (Tfam–mtDNA; 1:1 complex) and

66,776 Da (BSA).

nature structur al & molecular biology VOLUME 18 NUMBER 11 NOVEMBER 2011 1293

a r t i c l e s

minima in the base twist (base step parameter) at the sites of inter-

calation (Fig. 4c). Globally, however, the DNA is not underwound,

with an average helical twist of ~36°. The roll angle profile (Fig. 4b)

shows two sharp peaks, which reflect distortions of base stacking

owing to acute DNA bending.

The mode of DNA bending in the Tfam–mtDNA structure shows

remarkable parallels with the HU protein family, which consists of

DNA minor groove–binding proteins that have architectural roles

in prokaryotic DNA nucleoids

20,21

. Integration host factor (IHF),

HU and Hbb are HU-family proteins that contort their bound DNA

into a U-shape

20,21

. These proteins form dimers in which each sub-

unit uses a ‘β-ribbon arm’ to intercalate into the DNA minor groove

(

Fig. 4a,e). The dimerization interface between the two subunits is

rich in positive residues and serves to neutralize the negative charges

of the bent DNA backbone. The DNA fragments in the Tfam and

Hbb complexes show similar profiles in the minor groove width,

with two broad peaks corresponding to minor groove intercalations

(Fig. 4a). The roll angles also show two peaks that signify the sharp

bending of DNA. The peaks are slightly closer together in the Hbb

(~9 bp apart) versus the Tfam structure (

Fig. 4b). Superimposition

of the DNA fragments reveals the similarity in overall

geometry (Fig. 4f).

Both HMG boxes and the linker are crucial for DNA bending

To monitor DNA bending by Tfam, we developed a fluorescence

resonance energy transfer (FRET)-based assay. The crystal struc-

ture shows that after binding of Tfam, the ends of the 28 bp LSP

DNA fragment are brought to within 55 Å of each other (measuring

from the 5′-phosphate of one strand to the 5′-phosphate of the other

strand), whereas there is a 95-Å separation in a rod-like DNA frag-

ment of identical length. To construct the FRET sensor, Cy3 (donor)

and Cy5 (acceptor) fluorophores were covalently attached to opposite

ends of the LSP fragment. Addition of Tfam to the labeled, double-

stranded DNA resulted in a dose-dependent increase in acceptor

emission and a decrease in donor emission (Table 2; Supplementary

Fig. 4a,b). Control experiments confirmed that the acceptor

Phe60

Tyr200

Tyr165 Tyr162

Trp189

Strand A Strand B

Arg233

Arg232

Arg157

Trp189

Tyr162

Gln179

Pro178

Arg159

Arg157

5′

Asn163

Thr150

Thr77

Lys51

Leu58

Arg140

Ser61

Lys52

His137

Tyr103

Gln100

Ser56

Trp88

Ser55

Ser61

Tyr57

Arg82

IIe81

Thr78

Thr234

Tyr211

Pro178

Gln179

C19

3.1

3.6

G10

A22

3.0

3.1

Tyr162

T24

3.1

Lys146

His137

Lys139

K136

2.8

3.2

3.4

G16

3.2

Lys147

Arg140

T15

Thr77

G11

T19

3.3

IIe81

Ser61

Tyr57

G20

Leu58

3.8

2.8

2.9

2.8

Tyr57

Tyr99

Trp88

Tfam box A

Tfam box B

Hmgd

Lef1

Sox2

Hmgb1 box A

Tfam box B, no DNA

Asn163

Arg157

Lys147

Figure 3 Interactions of Tfam with DNA. (a) A ribbon diagram of HMG-box A. Hydrophobic residues that stabilize

the core are highlighted, with the 2F

– F

electron density map contoured at 1.5 σ. (b) HMG-box B, highlighted

as in a. (c) Superimposition of HMG-boxes A and B of Tfam with other HMG-boxes. Structures correspond to the

following accession numbers, and r.m.s. deviation values, relative to HMG-box A of Tfam, are provided in

parentheses: HMG-box B of Tfam (without DNA), 3fgh

(0.974); Hmgb1 box A, 1ckt

(1.101); Lef1, 2lef

(1.162);

Sox2, 1gt0 (ref.29) (1.152); Hmgd, 1qrv

(1.127). (d) Interactions of HMG-box A with DNA (gray). Tyr57, Leu58,

Ser61 and Ile81 make contacts with the DNA bases and sugar phosphate backbone, as indicated by dashed lines

with distances (in angstroms). Thr77 contacts a deoxyribose in the DNA backbone, and a mutant containing alanine

at this position shows reduced DNA bending (Table 2). (e) Interactions of HMG-box B with DNA (gray). Arg157,

Asn163, Gln179 and Pro178 make contacts with the bases, as indicated by the dashed lines. Tyr162 contacts a

deoxyribose in the DNA backbone, and the Y162A mutant shows reduced DNA bending (Table 2). (f) Interactions of the

α-helical linker with DNA (gray). The backbone of the linker helix is traced in magenta. (g) Interactions between Tfam and

DNA, analyzed by NUCPLOT

. Blue (dotted) and red (dashed) lines represent hydrogen-bonded and unbonded contacts (<3.35 Å) to DNA, respectively.

Circles labeled W indicate water-mediated interaction with DNA. Stereo views of a, b, d and e are provided in Supplementary Figure 2.

1294 VOLUME 18 NUMBER 11 NOVEMBER 2011 nature structur al & molecular biology

a r t i c l e s

emission depended on the presence of

both the donor fluorophore and Tfam.

The maximum FRET efficiency measured

with wild-type Tfam corresponds to a cal-

culated fluorophore separation of 59 Å, in

good agreement with the crystal structure.

The ability of Tfam to bend DNA is not

restricted to the LSP template. Tfam was able

to bend a DNA template lacking promoter sequences, and also one

corresponding to HSP1 (Supplementary Fig. 4c,d).

Analysis of a panel of Tfam mutants indicated that coordinated

binding of both HMG-box domains is important for effective bend-

ing of DNA (Table 2). HMG-box A alone bound to LSP DNA with

the same affinity as full-length Tfam, consistent with previous

studies

19,22

. However, it showed a large reduction in DNA bending.

HMG-box B alone showed much weaker affinity for LSP DNA

~400 nM) and also showed a large reduction in DNA bending.

In addition, we tested Tfam mutants with single point mutations in

HMG-box residues that contact DNA (Table 2; Supplementary

Fig. 4b). T77A, which contains a mutation in HMG-box A, and

Y162A, which contains a mutation in HMG-box B, showed mod-

erate reductions in DNA bending. Each of these residues makes

contacts with the DNA backbone (Fig. 3d,e). Finally, we tested the

effect of mutations in the positively charged residues in the linker

helix. Single point mutations had little effect (data not shown). We

therefore made a mutant, L6, in which six positively charged resi-

dues in the linker region were replaced by alanine. The L6 mutant

showed a >30% reduction in FRET, indicating that the linker region

is important for DNA bending. All of these mutants were well folded,

as established through secondary structure analysis by circular dichro-

ism (Supplementary Fig. 4e). In addition, the T77A, Y162A and

L6 mutants retained high affinity for DNA, as indicated by an assay

monitoring the quenching of intrinsic tryptophan fluorescence upon

DNA binding (Table 2; Supplementary Fig. 5).

Tfam bending mutants show promoter-specific defects

With mutants that reduce DNA bending by Tfam, we used in vitro

transcription reactions containing mitochondrial RNA polymerase

(Polrmt) and Tfb2m to test whether DNA bending is important for its

transcriptional activation function (Fig. 5). Neither HMG-box A nor

HMG-box B alone was able to activate transcription from LSP or HSP

templates. In addition, Tfam containing both HMG boxes but lacking

the C-terminal tail was unable to activate transcription (Fig. 5a,b).

These results are expected, because previous studies indicated that

the C-terminal tail of Tfam is essential for transcriptional activation

Notably, we found that both the T77A and Y162A mutants were less

efficient in promoting transcription from the LSP template. Y162A,

which has a more severe bending defect, was more affected. Finally,

the mutant L6, which has the strongest bending defect, showed a

severe defect in transcriptional activation. The transcriptional defects

were similar whether full-length or truncated LSP transcripts were

quantified (Fig. 5c,d).

To test whether these Tfam mutants were generally defective

in transcriptional activation, we examined their activity with

an HSP1 template (Fig. 5b,e). In DNA-bending measurements,

these mutants showed defects in bending the HSP1 DNA template

(Supplementary Fig. 4d), as was found previously with the LSP

template. In transcriptional activation assays, however, the Y77A,

Y162A and L6 mutants were all efficient at stimulating transcripts

from HSP1 (Fig. 5e). Quantification showed that all three mutants

showed a similar transcriptional activation profile compared to

wild-type Tfam.

a d

Minor groove (

)

T G T T A G T T G G G G G G G GA C T T T A A A A G TT

A A T A C T A T A T G T C A T A T A G T A T T A A A T

Tfam

Hbb

Roll (°)

T G T T A G T T G G G G G G G GA C T T T A A A A G TT

A A T A C T A T A T G T C A T A T A G T A T T A A A T

–20

Tfam

Hbb

Twist (°)

T G T T A G T T G G G G G G G GA C T T T A A A A G TT

A A T A C T A T A T G T C A T A T A G T A T T A A A T

Tfam

Hbb

Figure 4 Comparison of Tfam and Hbb

structures. (a) Profiles of minor groove width

in the Tfam–mtDNA (blue) and Hbb–DNA (red)

structures. (b) Roll angle profiles in the Tfam–

mtDNA (blue) and Hbb–DNA (red) structures.

(blue) and Hbb–DNA (red) structures. (d) Side

view of the Tfam–mtDNA complex. The protein

is shown in green, and DNA is shown in blue.

(e) Side view of the Hbb–DNA complex. Hbb is

shown in light blue, and DNA is shown in red.

(f) Manual overlay of DNA in the Tfam–mtDNA

and Hbb–DNA structures. DNAs from the

two structures are color coded as in d and e.

Analyses of the helical parameters of the DNA

molecules were carried out using 3DNA

Table 2 DNA bending and binding of Tfam variants

Tfam mutant FRET efﬁciency

(%) Distance

(Å) K

(nM)

No protein 6.5 ± 0.7 84 -

Wild type 36.5 ± 0.8 59 6.0 ± 0.9

T77A 31.4 ± 0.2 61 7.6 ± 1.1

Y162A 29.4 ± 0.6 63 12.3 ± 1.8

L6 26.1 ± 0.6 64 10.1 ± 1.2

HMG-box A 12.2 ± 0.4 75 6.5 ± 1.2

HMG-box B 6.2 ± 0.4 84 411.3 ± 46

The DNA bending and binding properties of Tfam and the indicated mutants were

measured. DNA bending was measured with a FRET assay using Cy3–Cy5-labeled LSP

DNA. The measured FRET efﬁciency was used to calculate the distance between the

DNA ends. The afﬁnity of Tfam mutants to DNA was monitored through the change

in intrinsic tryptophan ﬂuorescence on DNA binding. s.d. from three independent

experiments are indicated. L6: K136A, H137A, K139A, R140A, K146A and K147A;

HMG-box A: residues 43–122; HMG-box B: residues 153–222.

The FRET efﬁciency (E) was calculated with the following equation: E = (F

corr

)/(F

corr

+ D

corr

where F

corr

and D

corr

are the corrected FRET and donor signals at 662 and 562 nm, respec-

tively.

The distance was calculated from the FRET efﬁciency using the following equation:

E = R

/ (R

+ R

) with R

= 54 Å.

was calculated from an LSP DNA-binding assay, as

detailed in the Supplementary Methods.

nature structur al & molecular biology VOLUME 18 NUMBER 11 NOVEMBER 2011 1 2 9 5

a r t i c l e s

DISCUSSION

Previous structural studies have indicated that a single HMG-box

domain can bind to the DNA minor groove and sometimes cause

bending of the DNA double helix. For example, the prototypical

HMG-box protein Sry, which contains a single HMG-box domain,

bends DNA ~70–80° on binding to the minor groove

. This mode

of DNA bending (Fig. 6a) superficially resembles that of TATA-

box-binding protein (Tbp), in which binding of a β-sheet to the

DNA minor groove again induces moderate bending toward the

opposite direction

24,25

In comparison to these structures, the Tfam–mtDNA complex

illustrates how spatial coordination of tandem HMG-box domains

can be harnessed to impose even more extreme distortion onto

DNA (Fig. 6b). Tfam belongs to the subset of HMG-box proteins

that contain tandem HMG-box domains. These HMG-box proteins

generally show broad DNA binding and have important roles in

regulating chromatin structure and function

. For example, Hmgb1

is an architectural protein on chromatin that has been implicated

in transcription, chromatin organization and genome stability

. In

Tfam, the α-helical linker plays a key part by spatially coordinating

the two HMG-box domains, so that they bind the DNA minor groove

at sites located approximately one helical turn apart. Moreover, the

linker further facilitates DNA bending by neutralizing the negative

charges on the DNA backbone. Intriguingly, all of the other dual

HMG-box proteins in the human genome contain a cluster of 5–8

positively charged residues in the short region between the HMG-box

domains (Supplementary Fig. 6). It will be interesting to determine

whether these residues have a role analogous to that of the linker

region in Tfam.

Although Tfam and the HU family of nucleoid proteins do not

share sequence or structural homology, our studies indicate that

they use remarkably analogous strategies to impose extreme bend-

ing onto DNA (Fig. 6b). The similarities between the Tfam–DNA

and HU-family–DNA structures are intriguing, given that both pro-

teins are thought to control the architecture of DNA in nucleoids.

The DNA in our structure is from LSP and therefore is more directly

related to mitochondrial transcriptional activation. However, the

structure is likely to also be relevant for the role of Tfam in nucle-

oid organization, given the ability of Tfam to bend generic DNA

(Supplementary Fig. 4c).

Our results indicate that the relative importance of extreme

DNA bending by Tfam depends on the mitochondrial promoter.

LSP

Wild type

T77A

Y162A

No Tfam

HMG-box B

HMG-box A

No C-tail

420 nt

120 nt

Wild type

Y162A

T77A

10 20 30 40 60 80 (nM)

Wild type

Y162A

T77A

0 10 20 30 40 60 80 (nM)

Wild type

Y162A

T77A

0 10 20 30

60 80 (nM)

Wild type

T77A

Y162A

No Tfam

HMG-box B

HMG-box A

No C-tail

180 nt

HSP

LSP, full-length

Wild type

100

7060

4030

Concentration (nM)

50 80

Y162A

T77A

Transcript level (%)

Wild type

Y162A

T77A

Concentration (nM)

LSP, truncated

100

7060403020100 50 80

Transcript level (%)

Wild type

Y162A

T77A

Concentration (nM)

HSP1

100

7060403020100 50 80

Transcript level (%)

Figure 5 Tfam mutants with a selective defect at LSP. (a) In vitro

transcription reactions using an LSP template. Reactions contained 100 nM

Tfam or the indicated mutant. HMG-box A, residues 43–122; HMG-box B,

residues 153–222; no C-tail, residues 43–222; L6, K136A, H137A, K139A,

R140A, K146A and K147A. The LSP template generates a 420-nucleotide

(nt) full-length (run-off) transcript and a truncated 120 nt transcript.

(b) Same as a, except using an HSP1 template. (c) Generation of full-length

LSP transcripts by Tfam and mutants. The left panel shows representative

reactions, using the indicated concentrations of protein. Quantification is

presented in the right panel, with error bars representing s.d. from three

independent experiments. (d) Same as in c, except that truncated LSP

transcripts are shown and quantified. A fraction of LSP transcripts are

known to terminate prematurely at the conserved sequence block II (CSBII)

site located downstream of the start site

. (e) Same as in c, except that an

HSP1 template was used.

Box B

Box A

LSP

C-tail

Figure 6 Models for DNA bending and transcriptional activation.

(a) DNA bending by a single HMG box. The DNA (blue) is moderately

bent by wedging of the HMG box (triangle) on one face of the DNA.

Dashes indicate negative charges on the opposite face of the DNA

backbone. (b) Extreme DNA bending by Tfam and HU family proteins.

Two wedges (triangles) applied to one face of DNA result in two acute

kinks. A positively charged platform (circle) on the opposite face helps to

neutralize the negative charges of the DNA backbone. (c) Transcriptional

activation at LSP. Based on our crystal structure, HMG-box B binds the

half-site further away from the transcriptional start site. The C-terminal

tail (C-tail) nevertheless faces the transcriptional start site because of the

extreme DNA bend. (d) With Tfam mutants, we suggest that the defect in

DNA bending prevents proper orientation of the C-terminal tail.

1296 VOLUME 18 NUMBER 11 NOVEMBER 2011 nature structur al & molecular biology

Previous studies indicated that the C-terminal tail of Tfam is essential

for transcriptional activation

and physical interaction with Tfb2m

In the crystal structure, when Tfam is bound to the LSP promoter, the

HMG-box B domain binds at the half-site further upstream from the

transcription start site (Figs. 1c and 6c). Without DNA bending,

the C-terminal tail would face away from the transcriptional start

site (Fig. 6d). However, the DNA U-turn redirects the C-terminal

tail toward the transcriptional machinery (Fig. 6c). We speculate

that one of the functions of DNA bending by Tfam is to enable the

C-terminal tail to interact with the rest of transcriptional machinery.

Based on previous results

, Tfb2m is a favored candidate for such an

interaction. Remarkably, transcription from HSP1 is much less sensi-

tive to DNA bending by Tfam. Based on sequence analysis, the Tfam

binding sites in HSP1 versus LSP are in reverse orientations relative to

the direction of transcription

4,8

(Fig. 1b). When Tfam is bound to the

HSP1 promoter, HMG-box B would be expected to bind the half site

adjacent to the transcriptional start. The C-terminal tail is therefore

in proximity to the transcriptional machinery, regardless of whether

the DNA is bent or not. In future studies, it will be important to test

this proposal by determining the structure of Tfam in complex with

HSP1 promoter DNA.

METHODS

Methods and any associated references are available in the online

version of the paper at http://www.nature.com/nsmb/.

Accession codes. Protein Data Bank: atomic coordinates and

structure factors for the Tfam–mtDNA complex have been deposited

under the accession code 3TMM.

Note: Supplementary information is available on the Nature Structural

Molecular

Biology website.

ACKNOWLEDGMENTS

We thank N. Chan (California Institute of Technology) for making some

mutant constructs, Y. Zhang and Z. Liu (California Institute of Technology)

for suggestions on phase determination and structure refinement, T. Walton

(California Institute of Technology) for advice on SEC-MALS, S. Shan (California

Institute of Technology) for use of equipment and insightful discussions, the

staff at the Stanford Synchrotron Radiation Lightsource (SSRL) for technical

support with crystal screening and data collection, and members of the Chan

laboratory for critical reading of the manuscript. We acknowledge the Gordon

and Betty Moore Foundation for support of the Molecular Observatory at Caltech.

SSRL is supported by the US Department of Energy and National Institutes of

Health (NIH). This work was supported by NIH grants GM083121 (D.C.C.) and

GM062967 (D.C.C.).

AUTHOR CONTRIBUTIONS

H.B.N. and D.C.C. designed the experiments, analyzed the data and wrote the

paper. H.B.N. carried out the crystallography and performed the experimental

work. J.T.K. helped with the crystallographic analysis.

COMPETING FINANCIAL INTERESTS

The authors declare no competing financial interests.

Published online at http://www.nature.com/nsmb/.

Reprints and permissions information is available online at http://www.nature.com/

reprints/index.html.

1. Falkenberg, M., Larsson, N.G. & Gustafsson, C.M. DNA replication and transcription

in mammalian mitochondria. Annu. Rev. Biochem. 76, 679–699 (2007).

2. Bonawitz, N.D., Clayton, D.A. & Shadel, G.S. Initiation and beyond:

multiple functions of the human mitochondrial transcription machinery. Mol. Cell

24, 813–825 (2006).

3. Fisher, R.P. & Clayton, D.A. Puriﬁcation and characterization of human mitochondrial

transcription factor 1. Mol. Cell. Biol. 8, 3496–3509 (1988).

4. Fisher, R.P., Topper, J.N. & Clayton, D.A. Promoter selection in human mitochondria

involves binding of a transcription factor to orientation-independent upstream

regulatory elements. Cell 50, 247–258 (1987).

5. Larsson, N.G. et al. Mitochondrial transcription factor A is necessary for mtDNA

maintenance and embryogenesis in mice. Nat. Genet. 18, 231–236 (1998).

6. Parisi, M.A. & Clayton, D.A. Similarity of human mitochondrial transcription factor

1 to high mobility group proteins. Science 252, 965–969 (1991).

7. Stros, M., Launholt, D. & Grasser, K.D. The HMG-box: a versatile protein domain

occurring in a wide variety of DNA-binding proteins. Cell. Mol. Life Sci. 64, 2590–2606

(2007).

8. Dairaghi, D.J., Shadel, G.S. & Clayton, D.A. Addition of a 29 residue carboxyl-

terminal tail converts a simple HMG box-containing protein into a transcriptional

activator. J. Mol. Biol. 249, 11–28 (1995).

9. McCulloch, V. & Shadel, G.S. Human mitochondrial transcription factor B1 interacts

with the C-terminal activation region of h-mtTFA and stimulates transcription

independently of its RNA methyltransferase activity. Mol. Cell. Biol. 23, 5816–5824

(2003).

10. Kang, D., Kim, S.H. & Hamasaki, N. Mitochondrial transcription factor A (TFAM):

roles in maintenance of mtDNA and cellular functions. Mitochondrion 7, 39–44

(2007).

11. Kaufman, B.A. et al. The mitochondrial transcription factor TFAM coordinates the

assembly of multiple DNA molecules into nucleoid-like structures. Mol. Biol. Cell

18, 3225–3236 (2007).

12. Spelbrink, J.N. Functional organization of mammalian mitochondrial DNA in

nucleoids: history, recent developments, and future challenges. IUBMB Life 62,

19–32 (2010).

13. Bogenhagen, D.F., Rousseau, D. & Burke, S. The layered structure of human

mitochondrial DNA nucleoids. J. Biol. Chem. 283, 3665–3675 (2008).

14. Alam, T.I. et al. Human mitochondrial DNA is packaged with TFAM. Nucleic Acids

Res. 31, 1640–1645 (2003).

15. Ekstrand, M.I. et al. Mitochondrial transcription factor A regulates mtDNA copy

number in mammals. Hum. Mol. Genet. 13, 935–944 (2004).

16. Kucej, M., Kucejova, B., Subramanian, R., Chen, X.J. & Butow, R.A. Mitochondrial

nucleoids undergo remodeling in response to metabolic cues. J. Cell Sci. 121,

1861–1868 (2008).

17. Stott, K., Tang, G.S., Lee, K.B. & Thomas, J.O. Structure of a complex of tandem

HMG boxes and DNA. J. Mol. Biol. 360, 90–104 (2006).

18. Fisher, R.P., Parisi, M.A. & Clayton, D.A. Flexible recognition of rapidly evolving

promoter sequences by mitochondrial transcription factor 1. Genes Dev. 3,

2202–2217 (1989).

19. Gangelhoff, T.A., Mungalachetty, P.S., Nix, J.C. & Churchill, M.E. Structural analysis

and DNA binding of the HMG domains of the human mitochondrial transcription

factor A. Nucleic Acids Res. 37, 3153–3164 (2009).

20. Mouw, K.W. & Rice, P.A. Shaping the Borrelia burgdorferi genome: crystal structure

and binding properties of the DNA-bending protein Hbb. Mol. Microbiol. 63,

1319–1330 (2007).

21. Rice, P.A., Yang, S., Mizuuchi, K. & Nash, H.A. Crystal structure of an IHF-DNA

complex: a protein-induced DNA U-turn. Cell 87, 1295–1306 (1996).

22. Wong, T.S. et al. Biophysical characterizations of human mitochondrial transcription

factor A and its binding to tumor suppressor p53. Nucleic Acids Res. 37,

6765–6783 (2009).

23. Werner, M.H., Huth, J.R., Gronenborn, A.M. & Clore, G.M. Molecular basis of human

46X,Y sex reversal revealed from the three-dimensional solution structure of the

human SRY-DNA complex. Cell 81, 705–714 (1995).

24. Kim, J.L., Nikolov, D.B. & Burley, S.K. Co-crystal structure of TBP recognizing the

minor groove of a TATA element. Nature 365, 520–527 (1993).

25. Kim, Y., Geiger, J.H., Hahn, S. & Sigler, P.B. Crystal structure of a yeast TBP/TATA-

box complex. Nature 365, 512–520 (1993).

26. Stros, M. HMGB proteins: interactions with DNA and chromatin. Biochim. Biophys.

Acta 1799, 101–113 (2010).

27. Ohndorf, U.M., Rould, M.A., He, Q., Pabo, C.O. & Lippard, S.J. Basis for recognition

of cisplatin-modiﬁed DNA by high-mobility-group proteins. Nature 399, 708–712

(1999).

28. Love, J.J. et al. Structural basis for DNA bending by the architectural transcription

factor LEF-1. Nature 376, 791–795 (1995).

29. Reményi, A. et al. Crystal structure of a POU/HMG/DNA ternary complex

suggests differential assembly of Oct4 and Sox2 on two enhancers. Genes Dev. 17,

2048–2059 (2003).

30. Murphy, F.V.IV., Sweet, R.M. & Churchill, M.E. The structure of a chromosomal high

mobility group protein-DNA complex reveals sequence-neutral mechanisms

important for non-sequence-speciﬁc DNA recognition. EMBO J. 18, 6610–6618

(1999).

31. Luscombe, N.M., Laskowski, R.A. & Thornton, J.M. NUCPLOT: a program to generate

schematic diagrams of protein-nucleic acid interactions. Nucleic Acids Res. 25,

4940–4945 (1997).

32. Lu, X.J. & Olson, W.K. 3DNA: a software package for the analysis, rebuilding and

visualization of three-dimensional nucleic acid structures. Nucleic Acids Res. 31,

5108–5121 (2003).

33. Pham, X.H. et al. Conserved sequence box II directs transcription termination

and primer formation in mitochondria. J. Biol. Chem. 281, 24647–24652

(2006).

a r t i c l e s

nature structur al & molecular biology

doi:10.1038/nsmb.2159

ONLINE METHODS

Tfam purification. The human TFAM gene was cloned into the pET28a expression

vector (Novagen) between the BamHI and XhoI sites. This construct encodes resi-

dues 43–246, corresponding to full-length Tfam after cleavage of the N-terminal

mitochondrial leader sequence (residues 1–42). Tfam mutants were constructed

using PCR with oligonucleotides encoding mutations. Plasmids were transformed

into BL21 (DE3) Escherichia coli (Invitrogen). LB medium (20 ml) containing

50 µg ml

−1

kanamycin was inoculated with a single colony and grown overnight at

37 °C. The overnight culture was diluted to 4 l and grown until an OD

600

it reached of

1.0. After induction with 1 mM isopropyl β-

-1-thiogalactopyranoside, the culture

was grown overnight at room temperature (24 °C). The cells were harvested and

stored at −80 °C. Five grams of cells were resuspended in 50 ml lysis buffer (20 mM

Tris-HCl, 500 mM NaCl, pH 7.5) and sonicated for 5 min (10 s on and 20 s off)

on ice. After centrifugation at 4.3 × 10

g for 30 min at 4 °C, His-tagged Tfam was

purified from the supernatant with 3 ml of Talon Cobalt resin (Clontech). The

protein was eluted (20 mM Tris-HCl, 500 mM NaCl, 300 mM imidazole, pH 7.5)

and further purified by gel filtration chromatography using a Hi-Load Superdex

200 16/60 column (GE Healthcare) pre-equilibrated with running buffer (20 mM

Tris-HCl, 300 mM NaCl, 1 mM dithiothreitol (DTT), pH 7.5) in an AKTA Purifier

(Amersham). The peak fraction was collected and concentrated to 17–20 mg ml

−1

using Amicon Ultra-15 concentrators (Millipore) with a molecular weight cutoff

of 10 kDa. The protein was flash-frozen in liquid nitrogen and stored at −80 °C.

Selenomethionine-substituted Tfam was produced by the metabolic inhibition

method

, and preparative buffers contained 5 mM β-mercaptoethanol instead

of DTT. Proteins were analyzed by DNA binding and circular dichroism analysis,

as detailed in the Supplementary Methods.

Crystallization, data collection and structure determination. The duplex LSP

fragment was made by annealing complementary oligonucleotides (5′-TGTTA

GTTGGGGGGTGACTGTTAAAAGT-3′ and 5′-ACTTTTAACAGTCACCCC

CCAACTAACA-3′) in buffer (10 mM Tris-HCl (pH 7.5), 50 mM NaCl, 1 mM

EDTA) at a concentration of 0.9 mM. The mixture was incubated at 95 °C for

5 min, 75 °C for 5 min and room temperature for >5 h.

To form Tfam–mtDNA complexes, Tfam was mixed with duplex DNA in a

1.3:1 molar ratio. The mixture was incubated at room temperature for 30 min and

then on ice for 2 h. Crystallization trials by hanging drop-vapor diffusion at room

temperature identified a condition (29% (w/v) PEG 400, 0.15 M calcium acetate,

0.1 M sodium acetate (pH 4.2), 400 mM NDSB211 (dimethyl-2(-hydroxyethyl)-

(3-sulfopropyl)-ammonium)) that yielded rod-shaped crystals. Diffraction data

were collected on frozen crystals on beamline 12-2 at the Stanford Synchrotron

Radiation Lightsource. All data were processed with IMOSFLM

or XDS

, and

merged using SCALA

as implemented in CCP4 (ref. 38). A selenomethionine-

substituted Tfam–mtDNA complex was used for phasing. Using intensity data at

3.0 Å from three wavelengths, all five selenium sites were located with PHENIX

After solvent flattening and density modification in PHENIX, the map revealed

clear density for the protein and DNA. Manual model building in COOT

using

the 3.0-Å experimental map generated a starting model. Refinement of the best

solutions was carried out using PHENIX, with an initial round of rigid body

refinement followed by a round of simulated annealing. Refinement against a

2.5-Å data set produced an excellent map with density for most of the side chains.

After a few rounds of model adjustment and refinement with TLS obtained from

the TLSMD server

, the R

work

converged to 19.8% and the R

free

to 24.7%. The

final model includes residues 43–237 of Tfam and all the nucleotides. The current

model has excellent stereochemistry with no Ramachandran outliers, as assessed

by MOLPROBITY

FRET experiments. To generate LSP, HSP and non-promoter templates,

the following complementary oligonucleotides were annealed as described above:

LSP, 5′-Cy3-TGTTAGTTGGGGGGTGACTGTTAAAAGT-3′ and 5′-Cy5-ACT

TTTAACAGTCACCCCCCAACTAACA-3′; HSP1, 5′-Cy3-GGTTGGTTCGG

GGTATGGGGTTAGCAGC-3′ and 5′-Cy5-GCTGCTAACCCCATACCCCGA

ACCAACC-3′; non-promoter DNA, 5′-Cy3-GACATTGGAACACTATACCTA

TTATTCG-3′ and 5′-Cy5-cgaataataggtatagtgttccaatgtc-3′.

Additional details of the FRET measurements and analysis of the FRET data

are described in the Supplementary Methods.

SEC-MALS. SEC-MALS experiments were performed at room temperature

by loading samples on a Shodex KW 803 column with a Dawn Heleos MALS

detector (Wyatt Technology). The column was eluted with buffer containing

20 mM Tris-HCl (pH7.5), 300 mM NaCl and 1 mM DTT. A dn/dc (refractive

index increment) value of 0.185 ml mg

−1

was used. Bovine serum albumin

was used as an isotropic scatterer for detector normalization. The light scat-

tered by a protein is directly proportional to its weight-average molecular mass

and concentration.

In vitro transcription reactions. DNA fragments corresponding to LSP (posi-

tions 1–477) and HSP1 (positions 499–741) of human mtDNA were cloned into

the pSP65 vector at the BamHI and SalI sites. After digestion with BamHI for

LSP and SalI for HSP1, the linearized plasmids were used as templates in a tran-

scriptional run-off assay. Transcription reactions were carried out as described

with modifications. Template DNA (5 nM) was added to the reaction mix

(10 mM HEPES (pH 7.5), 10 mM MgCl

, 1 mM DTT, 100 µg ml

−1

BSA and

40 units of RNaseOut (Invitrogen)) for 5 min, and then Tfam, Tfb2m (30 nM,

Enzymax) and Polrmt (30 nM, Enzymax) were sequentially added, with a 1-min

incubation between each addition. After addition of rNTPs (400 µM rATP,

150 µM rCTP, 150 µM rGTP, 15 µM rUTP (Promega), 0.2 µM [α-

P]rUTP

(3,000 Ci mmol

−1

, PerkinElmer)), the reaction was incubated for 3 h at 33 °C,

and stopped by addition of 25 µL of stop buffer (80% formamide (v/v), 10 mM

EDTA, pH 8.0, 0.025% xylene cyanol (w/v), 0.025% bromophenol blue (w/v)).

Samples were heated to 90 °C for 5 min and separated on 5% polyacrylamide

gels (w/v) containing 8 M urea in 1× TBE buffer. The gels were fixed in 7%

(v/v) acetic acid, dried and exposed to a phosphorimager screen. The data were

collected on a Storm 880 phosphorimager (Molecular Dynamics) and quantified

using ImageQuant 5.2 Software.

34. Van Duyne, G.D., Standaert, R.F., Karplus, P.A., Schreiber, S.L. & Clardy, J. Atomic

structures of the human immunophilin FKBP-12 complexes with FK506 and

rapamycin. J. Mol. Biol. 229, 105–124 (1993).

35. Leslie, A.G.W. Joint CCP4 and ESF-EACBM Newsletter on Protein Crystallography

(Warrington WA4 4AD, UK, 1992).

36. Kabsch, W. XDS. Acta Crystallogr. D Biol. Crystallogr. 66, 125–132 (2010).

37. Evans, P.R. Data reduction. in Proceedings of the CCP4 Study Weekend. Data

Collection and Processing (eds. Sawyer, L., Isaacs, N. & Bailey, S.) 114–122

(Daresbury Laboratory, Warrington, UK, 1993).

38. Collaborative Computational Project. N. The CCP4 Suite: programs for

protein crystallography. Acta Crystallogr. D Biol. Crystallogr. 50, 760–763

(1994).

39. Adams, P.D. et al. PHENIX: building new software for automated crystallographic

structure determination. Acta Crystallogr. D Biol. Crystallogr. 58, 1948–1954

(2002).

40. Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics.

Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132 (2004).

41. Painter, J. & Merritt, E.A. TLSMD web server for the generation of multi-group TLS

models. J. Appl. Crystallogr. 39, 109–111 (2006).

42. Davis, I.W. et al. MolProbity: all-atom contacts and structure validation for proteins

and nucleic acids. Nucleic Acids Res. 35, W375–W383 (2007).

43. Lodeiro, M.F. et al. Identiﬁcation of multiple rate-limiting steps during the human

mitochondrial transcription cycle in vitro. J. Biol. Chem. 285, 16387–16402

(2010).