Báo cáo khoa học: Human mitochondrial transcription factor A possesses multiple subcellular targeting signals pptx

Human mitochondrial transcription factor A possesses
multiple subcellular targeting signals
Viktoriya Pastukh
1
, Inna Shokolenko
1
, Bin Wang
2
, Glenn Wilson
1
and Mikhail Alexeyev
1,3
1 Department of Cell Biology and Neuroscience, University of South Alabama, Mobile, AL, USA
2 Department of Mathematics and Statistics, University of South Alabama, Mobile, AL, USA
3 Institute of Molecular Biology and Genetics, Kyiv, Ukraine
Mitochondrial transcription factor A (TFAM, mtTFA)
is a member of a high-mobility group (HMG) of pro-
teins named on the basis of their electrophoretic mobil-
ity in polyacrylamide gels. This group is composed of
nonhistone chromatin proteins and transcription factors
that can bind DNA either nonspecifically or in a
sequence-dependent manner [1]. TFAM is encoded
in the nucleus and is synthesized on cytoplasmic
ribosomes as a precursor, which is converted, upon
mitochondrial importation, into a 24.4 kDa, 204 amino
acid mature form. The N-terminal sequence of the pre-
cursor has not been determined, and therefore it is possi-
ble that translation can start on either of two
N-terminal methionines, resulting in either 246 amino
acid (29 kDa) or 240 amino acid (28.4 kDa) precursors
[2]. The mature form contains two HMG boxes,

Keywords
chemotherapy; cisplatin; etoposide;
mitochondrial transcription factor A; nuclear
localization sequence
Correspondence
M. Alexeyev, Department of Cell Biology
and Neuroscience, University of South
Alabama, 307 University Blvd., MSB1201,
Mobile, AL 36688, USA
Fax: +1 251 460 6771
Tel: +1 251 460 6789
E-mail:
(Received 29 July 2007, revised 12 October
2007, accepted 25 October 2007)
doi:10.1111/j.1742-4658.2007.06167.x
The mitochondrial transcription factor A (TFAM) is a member of a high-
mobility group (HMG) family represented mostly by nuclear proteins.
Although nuclear localization of TFAM has been demonstrated in some
tumors and after treatment of tumor cells with anticancer drugs, the signifi-
cance of these observations has not been fully elucidated. Here we report
that both TFAM overexpression and impairment of its mitochondrial tar-
geting can result in nuclear accumulation of the protein. Both M1 and M7
methionines of human TFAM (hTFAM) can be used for translation initia-
tion with almost equal efficiency resulting in two polypeptides. The shorter
polypeptide, however, is not located in the nucleus, despite truncation in
the mitochondrial targeting sequence, and both isoforms are targeted to
mitochondria with similar efficiency. We further demonstrate that nuclear
TFAM confers significant cytoprotection against the chemotherapeutic
drugs etoposide, camptothecin, and cisplatin. Three regions of hTFAM
[HMG-like domain 1 (HMG1) and HMG-like domain 2 (HMG2), as well

as the tail region] can effect nuclear accumulation of enhanced green fluo-
rescent protein (EGFP) fusions. The HMG1 domain contains a bipartite
nuclear localization sequence whose identity is supported by site-directed
mutagenesis. However, this bipartite nuclear localization sequence is weak,
and both N-terminal and C-terminal flanking sequences enhance the
nuclear targeting of EGFP. Finally, several mutations in the HMG1
domain increased the mitochondrial targeting of the EGFP fusions, sug-
gesting that the mitochondrial targeting sequence of hTFAM may extend
beyond the cleavable presequence.
Abbreviations
EGFP, enhanced green fluorescent protein; HMG, high-mobility group; HMG1 and HMG2, HMG-like domains of human mitochondrial
transcription factor A; hTFAM, human mitochondrial transcription factor A; MTS, mitochondrial targeting sequence; NLS, nuclear localization
sequence; N ⁄ C, nucleus-to-cytoplasm; SOD2, manganese superoxide dismutase; Tc, tetracycline; TFAM, mitochondrial transcription
factor A.
6488 FEBS Journal 274 (2007) 6488–6499 ª 2007 The Authors Journal compilation ª 2007 FEBS
HMG-like domain 1 (HMG1) and HMG-like domain 2
(HMG2) (Fig. 1A), joined by a basic 36 amino acid lin-
ker and followed by a basic 27 amino acid tail. The gene
for TFAM spans about 10 kb and consists of seven ex-
onsandsixintrons[3,4].Inhumanandrat,exon 5cansplice
alternatively, resulting in two TFAM isoforms [4,5].
TFAM is required for mtDNA transcription and
maintenance. Inactivation of both TFAM alleles
results in embryonic lethality accompanied by severe
depletion of mtDNA [6]. Tissue-specific inactivation of
TFAM in cardiomyocytes, skeletal muscle cells, pan-
creatic b-cells and pyramidal neurons is associated
with mtDNA depletion, reduced levels of mitochon-
drial transcripts, and severe respiratory chain defi-
ciency [7–11]. TFAM levels generally correlate well

with mtDNA content, and upon transient depletion of
mtDNA with ethidium bromide, cellular TFAM con-
tent diminishes as well [12]. Conversely, both mtDNA
and TFAM levels are restored upon ethidium bromide
withdrawal, although TFAM appears to lag behind
mtDNA [12].
Like many other members of the HMG family,
TFAM can bind DNA in a nonsequence-specific man-
ner, although it appears to show a higher affinity for
mitochondrial heavy strand promoter and light strand
promoter [13]. TFAM binding to DNA induces
unwinding and bending [14], and the mitochondrial
TFAM content (approximately one TFAM molecule
per 10 bp) has been suggested to be high enough for
TFAM to cover mtDNA completely [15,16]. This,
together with TFAM’s high affinity for DNA contain-
ing cisplatin adducts and 8-oxo-7,8-dihydroguanine
raises the possibility of its involvement in recognition
and ⁄ or repair of mtDNA damage [17].
TFAM effects could be modulated by its interaction
with p53 [18] and acetylation [19]. Another interesting
possibility is the regulation of the effects of TFAM by
its subcellular targeting. In the mouse and chicken, but
not in the human, expression of a special nuclear iso-
form of TFAM was demonstrated. This isoform is
generated by alternative splicing of the duplicated first
exons, resulting in a protein that lacks a mitochondrial
A
B
C

D
Fig. 1. Nuclear localization of hTFAM and hTFAM–EGFP fusion pro-
teins upon overexpression. (A) Structure of TFAM. The figure is
drawn to scale. The domain boundaries are in accordance with the
Entrez Protein Database entry Q00059. Numbers indicate the
amino acid position. (B) Structures of constructs 1760 and 2463
encoding constitutively expressed and Tc-inducible hTFAM–EGFP
fusion proteins, respectively. Numbers on the right indicate plasmid
designations. Crossed ATG, deleted or mutated translation initiation
site. (C) Top row: Flp-in T-Rex cells were transiently transfected
with constitutively expressed hTFAM–EGFP fusion construct (con-
struct 1760). Bottom two rows: inducible expression of hTFAM–
EGFP construct in stably transfected (single copy) Flp-in T-Rex
cells. Left images: green, EGFP fusion proteins. Middle images:
red, MitoTracker Red (mitochondrial stain). Right images: overlay;
yellow, regions of colocalization. – Tc and + Tc, cells were either left
uninduced, or induced with 2 lgÆmL
)1
Tc for 48 h. (D) Accumula-
tion of hTFAM in the nuclei of transfected cells upon overexpres-
sion. The Flp-in T-Rex293 cells were stably transfected with
construct 2462, which encodes full-length hTFAM. Nuclear frac-
tions (12 lg of total protein) from the parental cell line (T-Rex), un-
induced (2462 unind) and induced (2462 ind) 2462 cell line as well
as the mitochondrial fraction (Mito) from the induced 2462 cell line
were subjected to western blot analysis, using antibodies against
nuclear lamin A ⁄ B (loading control) and SOD2 to verify the purity
of the fractions (top panel), as well as with antibodies against
lamin A ⁄ B (loading control) and antibody to hTFAM to determine
levels of hTFAM in the nuclei (the bottom panel).

V. Pastukh et al. Nuclear localization of hTFAM
FEBS Journal 274 (2007) 6488–6499 ª 2007 The Authors Journal compilation ª 2007 FEBS 6489
targeting sequence (MTS). However, the order of
‘nuclear’ and ‘mitochondrial’ exons in genomic DNA
of these species is opposite [20–22]. Nuclear localiza-
tion of TFAM was observed in rat hepatoma, where it
correlates with 10-fold overexpression of this protein
[23]. Also, TFAM was isolated recently from rat liver
nuclei, where it was found to be bound to chromatin
[24]. Presently, both the mechanism(s) and the physio-
logical consequences of the nuclear localization of
TFAM remain unclear. Here we identify the nuclear
localization sequences (NLSs) of human TFAM
(hTFAM), and demonstrate that nuclearly localized
hTFAM can exert significant sensitizing and cytopro-
tective effects in response to chemotherapeutic drugs.
Results and Discussion
HTFAM overexpression results in nuclear
localization of hTFAM–EGFP fusion proteins
As nuclear localization of TFAM correlates with ele-
vated levels of this protein in rat hepatoma cells [23],
we were interested in whether TFAM overexpression,
by itself, is sufficient for the relocalization of a fraction
of this protein to the nucleus. To this end, the con-
struct encoding TFAM–enhanced green fluorescent
protein (EGFP) fusion protein under the control of
the CMV promoter (construct 1760) was assembled
and introduced into HeLa and Flp-in T-Rex293 cell
lines by transient transfection. In both cases, a fraction
of the fusion protein accumulated in both the cyto-

plasm and the nucleus (Figs 1 and 2B). To confirm
that nuclear localization of the fusion protein was
indeed due to overexpression, we stably integrated an
identical fusion construct under the control of a CMV-
tet promoter (construct 2463) into the genome of the
Flp-in T-Rex293 cell line. The Flp-integrase-mediated
insertion occurs in a single defined site in the Flp-in
T-Rex293 genome. Therefore, our stable integrants,
unlike cells that received a similar construct by tran-
sient transfection, contained a single copy of the fusion
construct, and expressed lower levels of the fusion pro-
tein. In agreement with our hypothesis, the lower levels
of expression attained in the Flp-in T-Rex293-2463 cell
Fig. 2. Schematic diagrams (A) and subcellu-
lar localization (B) of hTFAM deletion con-
structs. The constructs were generated by
PCR and transfected using Polyfect reagent
as described in Experimental procedures.
Left images: green, EGFP fusion proteins.
Middle images: red, MitoTracker Red (mito-
chondrial stain). Right images: overlay;
yellow, the regions of colocalization.
EF1-alpha, a construct expressing unfused
EGFP.
Nuclear localization of hTFAM V. Pastukh et al.
6490 FEBS Journal 274 (2007) 6488–6499 ª 2007 The Authors Journal compilation ª 2007 FEBS
line in response to induction did not lead to nuclear
accumulation of the hTFAM–EGFP fusion protein, as
detectable by confocal microscopy, and instead com-
plete colocalization of the fusion protein and mito-

chondria was observed (Fig. 1, yellow color in the
overlay). To rule out the possibility that EGFP non-
specifically interferes with mitochondrial targeting, we
established a similar inducible stable cell line that
expressed unfused hTFAM (construct 2462), and used
subcellular fractionation techniques in combination
with more sensitive detection by western blotting. This
experiment also revealed accumulation of the hTFAM
in the nucleus in response to increased expression
(induction; Fig. 1D, lower panel). This accumulation
was not due to contamination of the nuclear fraction
with mitochondria, as shown by blotting for a mito-
chondrial marker, manganese superoxide dismutase
(SOD2; Fig. 1D, upper panel).
Translation of hTFAM can be initiated on both
N-terminal methionines with similar efficiency
The above observations suggest either that the two
specialized isoforms of hTFAM, nuclear and mito-
chondrial, are produced from a single cDNA, or that
a single hTFAM polypeptide possesses an intrinsic
nuclear localization signal and is unevenly partitioned
between the mitochondria and the nucleus. Indeed,
DNA ligase III has been shown to produce both
nuclear and mitochondrial isoforms by using alterna-
B
EF1alpha
1760
1788
1789
1790

1804
1805
2078
1819
1818
1817
1807
1806
Fig. 2. (Continued).
V. Pastukh et al. Nuclear localization of hTFAM
FEBS Journal 274 (2007) 6488–6499 ª 2007 The Authors Journal compilation ª 2007 FEBS 6491
tive translation initiation signals. In this case, a
shorter, nuclear isoform lacks the first 87 amino acids
encoding the MTS [25]. Similarly, hTFAM has two
methionines in its N-terminal region, M1 and M7, and
either one can potentially be used for translation initia-
tion. To verify whether this is indeed the case, the
5¢-region of hTFAM cDNA, including the 5¢-UTR,
was fused in frame with the luciferase gene, and either
M1 or M7, or both, were mutated to isoleucine (Fig. 3
and Experimental procedures). The luciferase assays
demonstrated that although translation initiation on
M7 occurs with somewhat lower efficiency as com-
pared to M1, these differences do not reach the level
of statistical significance (n ¼ 3, two-tailed t-test;
Fig. 3). To evaluate the subcellular distribution of the
shorter hTFAM variant with a truncated MTS, amino
acids 7–246 were fused to EGFP and the resulting con-
struct was transiently transfected into HeLa cells. No
substantial differences in the subcellular distribution

were detected between the full-length and the truncated
hTFAM variants (supplementary Fig. S1, 7–246). The
S12T polymorphism in the MTS of hTFAM has been
identified as a risk factor for Alzheimer’s disease [26].
We attempted to link this risk with altered nuclear tar-
geting of the shorter hTFAM variant containing the
S12T mutation. However, the patterns of subcellular
targeting of hTFAM(7–246) and hTFAM(7–246)S12T
were essentially identical (supplementary Fig. S1). As
both full-length and truncated [hTFAM(7–246)]
products are efficiently targeted to mitochondria, the
existence of a shorter hTFAM variant cannot account
for the nuclear accumulation of the EGFP fusion pro-
teins. Therefore, it is more likely that intrinsic NLS(s)
mediate the nuclear accumulation of hTFAM.
HTFAM possesses multiple NLSs
To determine whether hTFAM possesses intrinsic
NLS(s), a series of 5¢- and 3¢-deletions were introduced
into the hTFAM gene (Fig. 2). All three 3¢-deletions
(constructs 1788, 1789, and 1790) retained both the
MTS and HMG1 domain and demonstrated promi-
nent mitochondrial localization of the fusion proteins
with some nuclear fluorescence. In contrast, all 5¢-dele-
tions lacked the MTS and exhibited predominantly
nuclear and ⁄ or cytoplasmic fluorescence. We further
fused individual hTFAM segments (HMG1, linker,
HMG2, tail) to EGFP to locate putative NLS(s). Sur-
prisingly, three of the four constructs tested (HMG1,
HMG2, and tail fusions) accumulated in the nucleus,
suggesting the presence of NLSs. The strength of these

signals can be ranked on the basis of nuclear ⁄ cyto-
plasmic partitioning of the fusion proteins as
HMG1 > tail > HMG2 (Fig. 2B). Another unex-
pected result was that in approximately 2% of the cells
expressing the HMG1–EGFP fusion protein, a fraction
of the fusion protein was localized to mitochondria,
suggesting that mitochondrial targeting determinants
of hTFAM may extend beyond the cleavable MTS.
Lys96 and Lys97 are critical for the nuclear
targeting of HMG1
Human SRY protein, a nuclear transcription factor
expressed early in embryonic development, is arguably
the best studied member of the HMG family [27].
SRY contains two distinct NLSs, at either end of a
single HMG box. Both NLSs are highly conserved in
SRY among mammals and are believed to be required
for complete nuclear localization [28]. The N-terminal
NLS is bipartite and consists of two clusters of
Fig. 3. Translation initiation efficiency at M1 versus M7 of hTFAM.
(A) The structure of the reporter constructs. The bent arrow indi-
cates the initiating methionine. (B) Levels of luciferase activity
when initiated at either M1 or M7. Activities were normalized for
transfection efficiency using cotransfection with Renilla luciferase
and a dual luciferase assay system.
Nuclear localization of hTFAM V. Pastukh et al.
6492 FEBS Journal 274 (2007) 6488–6499 ª 2007 The Authors Journal compilation ª 2007 FEBS
positively charged residues separated by 12 amino
acids (Fig. 4B). Whereas the first of these clusters is
conserved in both HMG1 and HMG2 as well as in all
TFAMs aligned in Fig. 4B, the second cluster is absent

in mammalian TFAMs. As both HMG1 and HMG2
of hTFAM appear to possess NLSs, amino acid resi-
dues conserved between either HMG1 and HMG2
(Fig. 4A), or between HMG1 domains of TFAMs
from different species (Fig. 4B), were interrogated by
site-directed mutagenesis to identify residues that may
constitute the HMG1 NLS (Table 1; Fig. 4; supple-
mentary Fig. S2). Mutations in only four (K51, E63,
P73, and E106) of the 12 residues that are invariant
between HMG1 domains in all TFAMs aligned in
Fig. 4B had no effect on subcellular redistribution of
EGFP fusion proteins (Table 1; supplementary
Fig. S2). Of the eight remaining residues, mutations in
five (P50, P53, Y57, R104, and Y110) resulted in a sig-
nificant impairment of nuclear accumulation of fusion
proteins, implying involvement of these residues in
nuclear targeting. Interestingly, mutations in three
invariant residues (P66, W88, and K96) resulted in a
significant increase in the proportion of cells that
displayed mitochondrial partitioning of EGFP fusion
proteins (Table 1; Fig. 5A). On closer examination, a
putative bipartite NLS that consists of the R82-R83
duet and the K95-K96-K97 triplet separated by a
spacer of 11 amino acids (Fig. 4A) was found in the
HMG1 domain. Consistent with this observation, a
double mutation K96A + K97A completely elimi-
nated nuclear localization of the HMG1 domain
(Table 1). The nucleus-to-cytoplasm (N ⁄ C) index in
cells transfected with this mutant was not statistically
different from that of the cells transfected with a con-

struct encoding unfused EGFP (results not shown).
However, this putative NLS, by itself, was unable to
effect nuclear accumulation of EGFP fusion proteins
(Fig. 4A, construct 2163), and both N-terminal and
C-terminal flanking sequences enhanced nuclear target-
ing of EGFP by this NLS (Fig. 4A, constructs 2208
and 2209, respectively). Unlike mutations W88R and
Y99A in HMG1, which resulted in increased mito-
chondrial localization of the fusion proteins, the corre-
sponding mutations W189R and Y200A in HMG2 did
not result in any detectable mitochondrial localization,
and led instead to increased nuclear accumulation
Fig. 4. Alignments of HMG domains from various sources. (A) Alignment of HMG1 and HMG2 domains of hTFAM. The invariant amino acid
residues are in bold and italic. The solid black lines below alignment indicate boundaries of deletion constructs, whose designations appear
to the left. The P-values, which appear in the brackets next to construct designations, are from one-way
ANOVA with Dunnett’s post hoc test
comparisons with a construct expressing cytoplasmic EGFP. The statistically significant difference indicates nuclear accumulation of corre-
sponding EGFP fusion constructs. The HMG1 and HMG2 amino acid residues interrogated by site-directed mutagenesis are indicated by
arrows above and below the alignment, respectively. The brackets above the alignment designate components of the putative HMG1 NLS.
(B) Alignment of human SRY versus TFAMs from various species. The invariant amino acid residues are in bold and italic. Amino acid resi-
dues interrogated by site-directed mutagenesis are indicated by arrows below the alignment. The components of SRY bipartite NLS are indi-
cated by the brackets above the alignment, and corresponding amino acid residues in TFAMs from different species are indicated by
brackets below the alignment.
V. Pastukh et al. Nuclear localization of hTFAM
FEBS Journal 274 (2007) 6488–6499 ª 2007 The Authors Journal compilation ª 2007 FEBS 6493
(Table 1; Fig. 5; supplementary Fig. S3). In general,
unlike HMG1 mutations, none of the mutations in the
HMG2 domain led to detectable mitochondrial parti-
tioning of EGFP fusion proteins (Table 1). Mutations
D207A and E214A in HMG2, which affect residues cor-

responding to E106 and E113, respectively, behaved like
their HMG1 counterparts and did significantly affect
nuclear localization (Table 1; supplementary Fig. S3).
Mitochondrial targeting determinants of hTFAM
may extend beyond the cleavable MTS
Perhaps the most unexpected finding of the site-direc-
ted mutagenesis experiments was that several HMG1
mutations resulted in a significant increase in mito-
chondrial targeting of the HMG1–EGFP fusion
proteins. As mentioned above, in 2% of cells, HMG1–
EGFP fusion proteins are partially localized to mito-
chondria. Mutations P66R, P66A, W88R, W88A,
W88A + D93A, K96I and Y99A (Table 1; Fig. 5A)
significantly increased the fraction of cells with mito-
chondrial partitioning of the HMG1 fusion proteins.
The effect of the P66R and W88R mutations was addi-
tive, and the HMG1 P66R + W88R double mutant
localized to mitochondria in 60% of transfected cells,
as judged by subcellular distribution of its EGFP
fusion protein (Fig. 5A). Mitochondrial relocalization
of mutatnt HMG1–EGFP fusion proteins did not pre-
vent nuclear targeting of the same constructs, and dual
mitochondrial and nuclear localization was typically
observed (Fig. 5A). Interestingly, the P66E mutation,
unlike P66R and P66A, did not cause increased
mitochondrial localization of fusion proteins (Fig. 5A).
This is consistent with the notion that N-terminal posi-
tively charged amphiphilic a-helices, which are poor in
aspartic and glutamic acid residues, serve as mitochon-
drial targeting signals [29]. Further supporting the

notion that mutant HMG1 domains are targeted to
mitochondria using determinants similar to those
found in the N-terminal presequences, the placement
of HMG1 P66R + W88R at the C-terminus of EGFP
completely abolished the mitochondrial targeting effect
of the mutations, while having no effect on nuclear
targeting of this fusion protein (Table 1; Fig. 5A).
Finally, subcellular fractionation of cells transfected
with either wild-type HMG1–EGFP fusion construct
or with the double P66R + W88R mutant HMG1–
EGFP fusion construct has revealed increased mito-
chondrial accumulation of EGFP in cells transfected
with mutant construct. This accumulation was accom-
panied by the presence of putative processing products
in both whole cell lysates and in purified mitochondria
(Fig. 5B). Such products are characteristic of precursor
proteins cleaved by mitochondrial processing pepti-
dase, which removes presequences to produce mature
mitochondrial proteins. Collectively, these results
Table 1. Effects of mutations in HMG domains on their subcellular
distribution. ND, the N ⁄ C ratio was not determined for mutants dis-
playing mitochondrial retargeting, due to the existence of two dis-
crete populations of transfected cells; flN, decreased nuclear
accumulation; ›N, increased nuclear accumulation; ›M, mitochon-
drial redistribution of fusion proteins; ››M, strong mitochondrial
redistribution of fusion proteins; NS, not significantly different from
the wild type (WT).
Mutation(s)
N ⁄ C index
(mean ± SEM) P-value (n)

Trend in the
subcellular
redistribution
of mutants
EGFP (EF1-alpha) 1.01 ± 0.04
HMG1–EGFP fusion proteins
WT 6.3 ± 1.1
P50G 2.63 ± 0.31 P<0.05 (8) flN
P50G + Y110C 2.71 ± 0.32 P<0.05 (8) flN
P50G + E112A 8.17 ± 2.2 P > 0.05 (6) NS
K51A 3.81 ± 0.34 P > 0.05 (8) NS
K52A 0.37 ± 0.08 P<0.01 (6) flN
P53A 3.33 ± 0.65 P<0.01 (8) flN
Y57R 2.59 ± 0.4 P<0.05 (8) flN
E63A 4.41 ± 0.84 P > 0.05 (8) NS
P66R ND ND ›M
P66A ND ND ›M
P66R + W88R ND ND ››M
P73A 3 ± 0.39 P > 0.05 (8) NS
P73E 2.7 ± 0.31 P > 0.05 (8) NS
K76A 2.12 ± 0.2 P > 0.05 (8) NS
W88R ND ND ›M
W88A ND ND ›M
W88A + D93A ND ND ›M
R89A 1.82 ± 0.1 P > 0.05 (8) NS
D93A 4.12 ± 0.41 P > 0.05 (8) NS
S94W 7.26 ± 0.75
P > 0.05 (6) NS
K96A + K97A 1.12 ± 0.08 P<0.01 (8) flN
K96I ND ND ›M

Y99A ND ND ›M
R104A 2.02 ± 0.18 P<0.01 (8) flN
E106A 6.05 ± 0.59 P > 0.05 (8) NS
E106A, E113A 2.24 ± 0.24 P<0.01 (8) flN
Y110C 1.79 ± 0.2 P<0.01 (8) flN
E113A 7.78 ± 0.47 P > 0.05 (6) NS
HMG2–EGFP fusion proteins
WT 1.48 ± 0.06
W189R 2.74 ± 0.18 P<0.01 (8) ›N
E196K 1.91 ± 0.12 P<0.05 (8) ›N
Y200A 1.78 ± 0.11 P > 0.05 (8) NS
K205A 1.38 ± 0.07 P > 0.05 (8) NS
D207A 1.77 ± 0.14 P > 0.05 (8) NS
E214A 1.55 ± 0.06 P > 0.05 (8) NS
D207A, E214A 1.63 ± 0.06 P > 0.05 (8) NS
EGFP–HMG1 fusion proteins
WT 5.51 ± 0.44
P66R + W88R 4.36 ± 0.68 P > 0.05 (8) NS
Nuclear localization of hTFAM V. Pastukh et al.
6494 FEBS Journal 274 (2007) 6488–6499 ª 2007 The Authors Journal compilation ª 2007 FEBS
A
B
Fig. 5. Increased mitochondrial partitioning
of some HMG1 mutants. (A) Partitioning as
observed by fluorescence microscopy. Cells
with mitochondrial localization of mutants
are indicated by white arrows. Note that
P66A and P66R mutations, but not P66E
mutations, increase mitochondrial targeting
of HMG1–EGFP fusion proteins. Left

images: green, EGFP fusions. Middle
images: red, MitoTracker Red (mitochondrial
stain). Right images: overlay; yellow,
regions of colocalization. (B) Partitioning as
observed by subcellular fraction-
ation ⁄ western blotting. HEK293FT cells
were transfected with either construct 1817
(wild-type HMG1–EGFP fusion construct,
Fig. 3) or construct 1925 [P66R + W88R
HMG1–EGFP fusion construct (A)], and 48 h
after transfection, cells were lysed to pro-
duce whole cell (wc) lysates, or mitochon-
dria were isolated using a Pierce
mitochondrial isolation kit. Twenty micro-
grams of wc lysates and 10 l g of mitochon-
drial fraction (mito) were separated by
SDS ⁄ PAGE and subjected to western blot-
ting with antibody to mitochondrial HSP60
(a-HSP60, loading control) or antibody to
GFP (a-GFP). Asterisk: putative processing
products cleaved by mitochondrial process-
ing peptidase, which removes MTS from
mitochondrial precursor proteins.
V. Pastukh et al. Nuclear localization of hTFAM
FEBS Journal 274 (2007) 6488–6499 ª 2007 The Authors Journal compilation ª 2007 FEBS 6495
indicate that a cryptic mitochondrial targeting determi-
nant may be present in the HMG1 domain. This deter-
minant is likely to play an accessory role in the
context of the full-length protein. Taken out of that
context, this signal, by itself, is insufficient to effect

mitochondrial localization of EGFP fusion proteins.
However, as a consequence of mutations in the HMG1
domain, this determinant can be strengthened, result-
ing in the retargeting of the EGFP fusion proteins to
mitochondria. The role of this cryptic determinant in
the mitochondrial import of hTFAM remains to be
determined. It is likely that it acts cooperatively with
the MTS to effect the mitochondrial localization of the
mature polypeptide. Importantly, we found no evi-
dence for the presence of a similar cryptic determinant
in the HMG2 domain.
Nuclearly targeted hTFAM exerts cytoprotective
effects
TFAM has been found to preferentially bind to DNA
damaged by the genotoxic drugs cisplatin and N-acet-
oxyacetylaminofluorene [17,24,30]. This, in combina-
tion with observations of TFAM accumulation in
transformed cells [23], and its presence in nuclear
extracts of normal liver cells [24], raises the possibility
of the involvement of hTFAM in cellular responses to
chemotherapy [31,32]. Indeed, HMG proteins have
been reported to both impede [33,34] and enhance [35–
37] repair of damaged DNA. Therefore, we established
a cell line with tetracycline (Tc)-inducible expression
of nuclear, MTS-less, hTFAM (construct 2476). Upon
the induction of nuclear hTFAM synthesis, the suscep-
tibility of this cell line to treatment with three different
chemotherapeutic drugs, etoposide, camptothecin, and
cisplatin, was tested. As compared to the similarly
treated parental cell line, the susceptibility to treatment

with etoposide, camptothecin and cisplatin was
decreased by 6.8%, 3.9% and 9.6%, respectively, in
the 2476 line (Table 2). Therefore, although nuclear
hTFAM may affect a tumor’s susceptibility to chemo-
therapy, and may represent a defensive mechanism, the
amplitude of this response with the drugs tested is too
low to be of practical significance.
The mitochondrial localization of hTFAM may rep-
resent an example of ‘eclipsed distribution’, the phe-
nomenon of uneven protein distribution between two
or more cellular compartments, where accumulation of
protein in one compartment impedes its detection in
another [38]. Nsf1 protein, which is involved in the
maturation of FeS proteins in mitochondria, represents
a prototypical example of such distribution. Similar to
that of hTFAM, the nuclear localization of Nsf1 pro-
tein is undetectable by physical means. However, it has
been demonstrated that Nsf1 possesses an internal
NLS, and that impairment of either nuclear or mito-
chondrial targeting of Nsf1 is lethal [39,40]. The
embryonic lethality of the TFAM knockout [6] appears
to extend the similarity between these two proteins.
However, more studies are needed to identify the exact
physiological role of nuclear TFAM.
Experimental procedures
Plasmids
pEF1a is a pcDNA3-derived plasmid in which the elonga-
tion factor 1a promoter drives expression of the EGFP
gene. The plasmid encoding full-length cDNA of hTFAM
was purchased from Open Biosystems (Huntsville, AL).

hTFAM fusion, deletion and mutant constructs were
assembled under the control of the CMV promoter. Con-
structs for generation of Tc-inducible cell lines were gener-
ated in a modified pcDNA5 ⁄ FRT ⁄ TO vector.
Site-directed mutagenesis and gene fusion
Site-directed mutagenesis was performed by an overlap
extension method [41] using Taq and Vent DNA poly-
merases. All mutations were verified by sequencing. For all
C-terminal fusions with HMG domains, an EGFP gene
lacking the initiating ATG codon was used. The ATG-less
EGFP gene was generated by PCR, cloned, and sequenced.
This was done to exclude expression of unfused EGFP by
means of leaky ribosomal scanning.
Cell culture and transfection
HeLa and Flp-in T-Rex293 cells were grown in DMEM
supplemented with 10% fetal bovine serum, 100 unitsÆmL
)1
penicillin, and 100 lgÆmL
)1
streptomycin. Cells were seeded
into 35 mm tissue culture dishes at a density of 3 · 10
5
cells
per dish, and transfections were performed using Polyfect
Table 2. Effect of nuclear hTFAM expression on susceptibility to
treatment with genotoxic drugs.
Drug
2476
Mean
a

(%)
SEM
(%)
P-value
(n ¼ 3)
Etoposide 106.8 2.1 < 0.0001
Camptothecin 103.9 1.7 0.0017
Cisplatin 109.6 2.6 < 0.0001
a
Viability as compared to the parental cell line.
Nuclear localization of hTFAM V. Pastukh et al.
6496 FEBS Journal 274 (2007) 6488–6499 ª 2007 The Authors Journal compilation ª 2007 FEBS
transfection reagent (Qiagen, Valencia, CA) according to
the manufacturer’s recommendations. Cells were observed
by confocal microscopy 40 h after transfection.
Generation of inducible cell lines
Cell lines with Tc-inducible expression of hTFAM or its
derivatives were generated with the help of a Flp-in T-Rex
system according to the manufacturer’s recommendations
(Invitrogen, Carlsbad, CA). Protein expression was induced
with 2 lgÆmL
)1
Tc for 48 h.
Subcellular fractionation
Cells were collected by trypsinization, washed with NaCl ⁄ P
i
,
and resuspended in buffer A (10 mm Hepes, pH 7.9, 10 mm
KCl, 5 mm MgCl
2

), to which NP40 was added to a final con-
centration of 0.4%. Cells were vortexed for 1 min, and 1 m
sucrose in buffer A was added to a final concentration of
200 mm to make the solution isotonic. Nuclei were collected
by centrifugation at 850 g for 3 min at 4 °C and washed in
the same buffer with sucrose. The supernatant was centri-
fuged at 15 000 g for 10 min at 4 °C to pellet mitochondria.
Nuclei and mitochondria were lysed in 10 mm Tris (pH 8.0),
1mm EDTA, and 0.5% SDS, and sonicated, and the protein
concentration was determined by the Bradford method.
Cell viability studies
The effect of expression of the hTFAM derivatives on cell
viability in response to various drug treatments was evalu-
ated using Alamar Blue fluorescence.
Microscopy
Confocal microscopy was performed on live cells using a
Leica DM RXE microscope and a TCS SP2 confocal system
(Leica Microsystems Inc., Bannockburn, IL) in combination
with a 63· water immersion objective. Prior to microscopy,
mitochondria were stained with 200 nm MitoTracker Red
(Invitrogen) for 15 min at 37 °C in an atmosphere of 5%
CO
2
. The nuclear accumulation of EGFP fusion proteins
was quantitated using the N ⁄ C distribution index. To calcu-
late this index, average fluorescence intensities (pixel densi-
ties) in nuclear and cytoplasmic regions were determined
with the image j program (National Institutes of Health),
and nuclear fluorescence was divided by cytoplasmic fluores-
cence. Statistical analyses of N ⁄ C indices were performed

using one-way anova with Dunnett’s post hoc test.
Luciferase assays
The pGL3 basic reporter plasmid was modified by intro-
ducing the CMV promoter and by removing the N-terminal
methionine of luciferase. The latter modification makes
luciferase expression dependent upon the upstream methio-
nine, which can be provided by a fusion partner. Then,
143 bp of hTFAM cDNA encompassing the 5¢-UTR and
the first 21 bp of the hTFAM gene was cloned upstream
of, and in frame with, the luciferase gene. Finally, three
constructs were generated by replacing either M1, M7 or
both with isoleucine (constructs 1969, 1970 and 1971,
respectively). Luciferase assays were performed using a
dual-luciferase reporter assay system (Promega, Madison,
WI). This system allows for the internal normalization of
results using cotransfection with a second plasmid encoding
Renilla luciferase. The light output was measured using a
TD-20 luminometer (Turner BioSystems, Inc., Sunnyvale,
CA).
Susceptibility to anticancer drugs
Flp-in T-Rex293 cells were stably transformed with con-
struct 2476, which encodes an MTS-less mature form of
hTFAM. The resulting cell line, 2476, accumulates hTFAM
in the nucleus in response to Tc induction. It was plated at
100 000 cells per well and pretreated with Tc for 24 h,
where necessary. Subsequently, cells were subjected to one
of four treatments: (a) carrier (dimethylsulfoxide) alone; (b)
Tc (2 lgÆmL
)1
) alone; (c) drug (etoposide, 20 lg Æ mL

)1
;
camptothecin, 20 lgÆmL
)1
; or cisplatin 75 lgÆmL
)1
) alone;
and (d) drug plus Tc for 24 h. Viability was determined
using Alamar Blue fluorescence. The fluorescence readings
from each cell line that received the four different treat-
ments (triplicate wells) were normalized as follows: dividing
the reading obtained under treatments 2, 3 and 4 in each
experiment by the average of the triplicate readings
obtained under treatment 1. This normalized the readings
in different experiments on different days and made the
changes in readings comparable. A four-way anova was
used to evaluate the effects of nuclear hTFAM expression.
The mean changes, together with the SEMs, were also com-
puted and listed on the basis of the normalized data
obtained under treatment 4 between Flp-in T-Rex293 and
2476. The analyses were performed with the sas 9.1 soft-
ware package (SAS Institute Inc., Cary, NC).
Acknowledgements
The work in G. L. Wilson’s laboratory was supported
by National Institutes of Health Grants ES03456 and
AG19602.
References
1 Bustin M (1999) Regulation of DNA-dependent activi-
ties by the functional motifs of the high-mobility-group
chromosomal proteins. Mol Cell Biol 19, 5237–5246.

V. Pastukh et al. Nuclear localization of hTFAM
FEBS Journal 274 (2007) 6488–6499 ª 2007 The Authors Journal compilation ª 2007 FEBS 6497
2 Parisi MA & Clayton DA (1991) Similarity of human
mitochondrial transcription factor 1 to high mobility
group proteins. Science 252, 965–969.
3 Reyes A, Mezzina M & Gadaleta G (2002) Human
mitochondrial transcription factor A (mtTFA): gene
structure and characterization of related pseudogenes.
Gene 291, 223–232.
4 Tominaga K, Hayashi J, Kagawa Y & Ohta S (1993)
Smaller isoform of human mitochondrial transcription
factor 1: its wide distribution and production by alter-
native splicing. Biochem Biophys Res Commun 194,
544–551.
5 Mezzina M, Reyes A, D’Errico I & Gadaleta G (2002)
Characterization of the mtTFA gene and identification
of a processed pseudogene in rat. Gene 286, 105–112.
6 Larsson NG, Wang J, Wilhelmsson H, Oldfors A,
Rustin P, Lewandoski M, Barsh GS & Clayton DA
(1998) Mitochondrial transcription factor A is necessary
for mtDNA maintenance and embryogenesis in mice.
Nat Genet 18, 231–236.
7 Li H, Wang J, Wilhelmsson H, Hansson A, Thoren P,
Duffy J, Rustin P & Larsson NG (2000) Genetic modifi-
cation of survival in tissue-specific knockout mice with
mitochondrial cardiomyopathy. Proc Natl Acad Sci
USA 97, 3467–3472.
8 Silva JP, Kohler M, Graff C, Oldfors A, Magnuson
MA, Berggren PO & Larsson NG (2000) Impaired insu-
lin secretion and beta-cell loss in tissue-specific knock-

out mice with mitochondrial diabetes. Nat Genet 26,
336–340.
9 Sorensen L, Ekstrand M, Silva JP, Lindqvist E, Xu B,
Rustin P, Olson L & Larsson NG (2001) Late-onset
corticohippocampal neurodepletion attributable to cata-
strophic failure of oxidative phosphorylation in MILON
mice. J Neurosci 21, 8082–8090.
10 Wang J, Wilhelmsson H, Graff C, Li H, Oldfors A,
Rustin P, Bruning JC, Kahn CR, Clayton DA, Barsh
GS et al. (1999) Dilated cardiomyopathy and atrioven-
tricular conduction blocks induced by heart-specific
inactivation of mitochondrial DNA gene expression.
Nat Genet 21, 133–137.
11 Wredenberg A, Wibom R, Wilhelmsson H, Graff C,
Wiener HH, Burden SJ, Oldfors A, Westerblad H &
Larsson NG (2002) Increased mitochondrial mass in
mitochondrial myopathy mice. Proc Natl Acad Sci USA
99, 15066–15071.
12 Seidel-Rogol BL & Shadel GS (2002) Modulation of
mitochondrial transcription in response to mtDNA
depletion and repletion in HeLa cells. Nucleic Acids Res
30, 1929–1934.
13 Fisher RP & Clayton DA (1988) Purification and
characterization of human mitochondrial transcription
factor 1. Mol Cell Biol 8, 3496–3509.
14 Fisher RP, Lisowsky T, Parisi MA & Clayton DA
(1992) DNA wrapping and bending by a mitochondrial
high mobility group-like transcriptional activator pro-
tein. J Biol Chem 267, 3358–3367.
15 Alam TI, Kanki T, Muta T, Ukaji K, Abe Y, Nakay-

ama H, Takio K, Hamasaki N & Kang D (2003)
Human mitochondrial DNA is packaged with TFAM.
Nucleic Acids Res 31, 1640–1645.
16 Takamatsu C, Umeda S, Ohsato T, Ohno T, Abe Y,
Fukuoh A, Shinagawa H, Hamasaki N & Kang D
(2002) Regulation of mitochondrial D-loops by tran-
scription factor A and single-stranded DNA-binding
protein. EMBO Report 3, 451–456.
17 Yoshida Y, Izumi H, Ise T, Uramoto H, Torigoe T,
Ishiguchi H, Murakami T, Tanabe M, Nakayama Y,
Itoh H et al. (2002) Human mitochondrial transcrip-
tion factor A binds preferentially to oxidatively dam-
aged DNA. Biochem Biophys Res Commun 295, 945–
951.
18 Yoshida Y, Izumi H, Torigoe T, Ishiguchi H, Itoh H,
Kang D & Kohno K (2003) P53 physically interacts
with mitochondrial transcription factor A and differen-
tially regulates binding to damaged DNA. Cancer Res
63, 3729–3734.
19 Dinardo MM, Musicco C, Fracasso F, Milella F,
Gadaleta MN, Gadaleta G & Cantatore P (2003) Acety-
lation and level of mitochondrial transcription factor A
in several organs of young and old rats. Biochem Bio-
phys Res Commun 301, 187–191.
20 Larsson NG, Barsh GS & Clayton DA (1997) Structure
and chromosomal localization of the mouse mitochon-
drial transcription factor A gene (Tfam). Mamm
Genome 8, 139–140.
21 Larsson NG, Garman JD, Oldfors A, Barsh GS &
Clayton DA (1996) A single mouse gene encodes the

mitochondrial transcription factor A and a testis-specific
nuclear HMG-box protein. Nat Genet 13, 296–302.
22 Matsushima Y, Matsumura K, Ishii S, Inagaki H,
Suzuki T, Matsuda Y, Beck K & Kitagawa Y (2003)
Functional domains of chicken mitochondrial transcrip-
tion factor A for the maintenance of mitochondrial
DNA copy number in lymphoma cell line DT40. J Biol
Chem 278, 31149–31158.
23 Dong X, Ghoshal K, Majumder S, Yadav SP & Jacob
ST (2002) Mitochondrial transcription factor A and its
downstream targets are up-regulated in a rat hepatoma.
J Biol Chem 277, 43309–43318.
24 Pietrowska M, Kolodziejczyk I & Widlak P (2006)
Mitochondrial transcription factor A is the major pro-
tein in rodent hepatocytes that recognizes DNA lesions
induced by N-acetoxy-acetylaminofluorene. Acta
Biochim Pol 53, 777–782.
25 Lakshmipathy U & Campbell C (1999) The human
DNA ligase III gene encodes nuclear and mitochondrial
proteins. Mol Cell Biol 19, 3869–3876.
26 Gunther C, von Hadeln K, Muller-Thomsen T, Alberici
A, Binetti G, Hock C, Nitsch RM, Stoppe G, Reiss J,
Nuclear localization of hTFAM V. Pastukh et al.
6498 FEBS Journal 274 (2007) 6488–6499 ª 2007 The Authors Journal compilation ª 2007 FEBS
Gal A et al. (2004) Possible association of mitochon-
drial transcription factor A (TFAM) genotype with spo-
radic Alzheimer disease. Neurosci Lett 369, 219–223.
27 Kanai Y, Hiramatsu R, Matoba S & Kidokoro T
(2005) From SRY to SOX9: mammalian testis differen-
tiation. J Biochem (Tokyo) 138, 13–19.

28 Harley VR, Layfield S, Mitchell CL, Forwood JK, John
AP, Briggs LJ, McDowall SG & Jans DA (2003) Defec-
tive importin beta recognition and nuclear import of the
sex-determining factor SRY are associated with XY
sex-reversing mutations. Proc Natl Acad Sci USA 100,
7045–7050.
29 Lemire BD, Fankhauser C, Baker A & Schatz G (1989)
The mitochondrial targeting function of randomly gen-
erated peptide sequences correlates with predicted heli-
cal amphiphilicity. J Biol Chem 264, 20206–20215.
30 Pietrowska M & Widlak P (2005) Characterization of a
novel protein that specifically binds to DNA modified
by N-acetoxy-acetylaminofluorene and cis-diamminedi-
chloroplatinum. Acta Biochim Pol 52, 867–874.
31 Reeves R & Adair JE (2005) Role of high mobility
group (HMG) chromatin proteins in DNA repair. DNA
Repair (Amst) 4, 926–938.
32 Widlak P, Pietrowska M & Lanuszewska J (2006) The
role of chromatin proteins in DNA damage recognition
and repair. Histochem Cell Biol 125, 119–126.
33 He Q, Liang CH & Lippard SJ (2000) Steroid hormones
induce HMG1 overexpression and sensitize breast
cancer cells to cisplatin and carboplatin. Proc Natl Acad
Sci USA 97, 5768–5772.
34 Zamble DB, Mikata Y, Eng CH, Sandman KE & Lipp-
ard SJ (2002) Testis-specific HMG-domain protein alters
the responses of cells to cisplatin. J Inorg Biochem 91,
451–462.
35 Nagatani G, Nomoto M, Takano H, Ise T, Kato K,
Imamura T, Izumi H, Makishima K & Kohno K (2001)

Transcriptional activation of the human HMG1 gene in
cisplatin-resistant human cancer cells. Cancer Res 61,
1592–1597.
36 Nagaki S, Yamamoto M, Yumoto Y, Shirakawa H,
Yoshida M & Teraoka H (1998) Non-histone chromo-
somal proteins HMG1 and 2 enhance ligation reaction
of DNA double-strand breaks. Biochem Biophys Res
Commun 246, 137–141.
37 Yuan F, Gu L, Guo S, Wang C & Li GM (2004) Evi-
dence for involvement of HMGB1 protein in human
DNA mismatch repair. J Biol Chem 279, 20935–20940.
38 Regev-Rudzki N, Karniely S, Ben-Haim NN & Pines O
(2005) Yeast aconitase in two locations and two meta-
bolic pathways: seeing small amounts is believing. Mol
Biol Cell 16, 4163–4171.
39 Nakai Y, Umeda N, Suzuki T, Nakai M, Hayashi H,
Watanabe K & Kagamiyama H (2004) Yeast Nfs1p is
involved in thio-modification of both mitochondrial and
cytoplasmic tRNAs. J Biol Chem 279, 12363–12368.
40 Muhlenhoff U, Balk J, Richhardt N, Kaiser JT, Sipos
K, Kispal G & Lill R (2004) Functional characteriza-
tion of the eukaryotic cysteine desulfurase Nfs1p from
Saccharomyces cerevisiae. J Biol Chem 279, 36906–
36915.
41 Ho SN, Hunt HD, Horton RM, Pullen JK & Pease LR
(1989) Site-directed mutagenesis by overlap extension
using the polymerase chain reaction. Gene 77, 51–59.
Supplementary material
The following supplementary material is available
online:

Fig. S1. Subcellular targeting of TFAM(7–246)–EGFP
and TFAM(7–246)S12T–EGFP fusion proteins in
transiently transfected HeLa cells.
Fig. S2. Subcellular distribution of mutant HMG1–
EGFP fusion proteins.
Fig. S3. The effect of HMG2 mutations on the subcel-
lular distribution of HMG2–EGFP fusion proteins.
This material is available as part of the online article
from
Please note: Blackwell Publishing is not responsible
for the content or functionality of any supplementary
materials supplied by the authors. Any queries (other
than missing material) should be directed to the corre-
sponding author for the article.
V. Pastukh et al. Nuclear localization of hTFAM
FEBS Journal 274 (2007) 6488–6499 ª 2007 The Authors Journal compilation ª 2007 FEBS 6499