Báo cáo khoa học: Recombinant derivatives of the human high-mobility group protein HMGB2 mediate efficient nonviral gene delivery pptx

Recombinant derivatives of the human high-mobility group
protein HMGB2 mediate efficient nonviral gene delivery
Arjen Sloots and Winfried S. Wels
Chemotherapeutisches Forschungsinstitut, Georg-Speyer-Haus, Frankfurt am Main, Germany
Virus-based vectors have been the gene delivery vehi-
cles of choice in most gene therapy approaches to date,
and use of these vectors has led to significant successes
in a number of clinical trials [1]. Nevertheless, recent
adverse events in patients treated with different viral
vectors have revived interest in alternative, nonviral
delivery systems for gene therapy [2,3]. Although still
less efficient than most viral vectors, nonviral gene
delivery vehicles are not usually associated with serious
safety concerns.
In addition to synthetic nonviral vectors such as
lipids and polycationic reagents, certain natural
peptides and proteins are able to bind and condense
plasmid DNA, a prerequisite for the formation of
transfection-competent complexes [4]. Consequently,
cellular DNA-binding proteins including histones [5–9]
and high-mobility group (HMG) proteins [10,11] have
been investigated for their potential as nonviral gene
delivery reagents. In these studies, DNA-binding pro-
teins were extracted from tissues such as calf thymus,
which requires large amounts of starting material and
can yield heterogeneous protein fractions that display
reduced DNA-binding activity because of exposure to
acid during purification [12,13]. Therefore recombinant
Keywords
gene delivery; high-mobility group protein;
importin-a; nuclear localization signal;

protein transduction domain
Correspondence
W. S. Wels, Chemotherapeutisches
Forschungsinstitut, Georg-Speyer-Haus,
Paul-Ehrlich-Straße 42–44, D-60596
Frankfurt am Main, Germany
Fax: +49 69 63395 189
Tel: +49 69 63395 188
Email:
(Received 13 April 2005, revised 23 June
2005, accepted 24 June 2005)
doi:10.1111/j.1742-4658.2005.04834.x
Certain natural peptides and proteins of mammalian origin are able to bind
and condense plasmid DNA, a prerequisite for the formation of transfec-
tion-competent complexes that facilitate nonviral gene delivery. Here we
have generated recombinant derivatives of the human high-mobility group
(HMG) protein HMGB2 and investigated their potential as novel protein-
based transfection reagents. A truncated form of HMGB2 encompassing
amino acids 1–186 of the molecule was expressed in Escherichia coli at high
yield. This HMGB2
186
protein purified from bacterial lysates was able to
condense plasmid DNA in a concentration-dependent manner, and medi-
ated gene delivery into different established tumor cell lines more efficiently
than poly(l-lysine). By attaching, via gene fusion, additional functional
domains such as the HIV-1 TAT protein transduction domain (TAT
PTD
-
HMGB2
186

), the nuclear localization sequence of the simian virus 40
(SV40) large T-antigen (SV40
NLS
-HMGB2
186
), or the importin-b-binding
domain (IBB) of human importin-a (IBB-HMGB2
186
), chimeric fusion pro-
teins were produced which displayed markedly improved transfection effi-
ciency. Addition of chloroquine strongly enhanced gene transfer by all four
HMGB2
186
derivatives studied, indicating cellular uptake of protein–DNA
complexes via endocytosis. The IBB-HMGB2
186
molecule in the presence
of the endosomolytic reagent was the most effective. Our results show that
recombinant derivatives of human HMGB2 facilitate efficient nonviral
gene delivery and may become useful reagents for applications in gene
therapy.
Abbreviations
eGFP, enhanced green fluorescent protein; HMG protein, human high-mobility group protein; IBB, importin-b-binding domain; NLS, nuclear
localization sequence; PEI, polyethyleneimine; PTD, protein transduction domain; SV40, simian virus 40.
FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS 4221
DNA-binding proteins such as human histone H1
expressed in bacteria and rat HMGB1 produced in
yeast cells are suitable alternatives [12,14].
On its own the ability of a nonviral vector to con-
dense DNA is not sufficient to mediate gene delivery

with high efficiency. Eukaryotic cells are protected
against the uptake of exogenous nucleic acids by a
series of cellular barriers that must be overcome before
a delivered gene can be expressed in the target cell
nucleus. In particular, ineffective escape from endo-
somal compartments and poor nuclear trafficking are
considered major limiting factors for many nonviral
gene-transfer systems [15,16]. The production of pro-
tein-based gene-delivery vectors in recombinant form
in principle allows their activities to be modified by inclu-
ding heterologous sequences that help to overcome
these cellular barriers by improving cellular uptake,
endosome escape, and intracellular routing [17].
Here we report the construction of recombinant
derivatives of the human nonhistone chromatin protein
HMGB2 and their functional characterization as non-
viral gene delivery vectors. Vertebrate HMGB proteins
such as HMGB1 and HMGB2 are composed of three
structurally defined regions [18,19]. They contain two
homologous but distinct DNA-binding motifs, termed
HMG boxes, and an acidic C-terminal domain. In
human HMGB2, the HMG boxes A and B, inter-
spaced by basic amino acids, are connected by another
basic region to a stretch of 22 acidic amino acids at
the C-terminus of the protein [20,21] (schematically
shown in Fig. 1A). These basic regions together with
some basic amino-acid residues at the N-terminus of
HMG box A have been suggested to function as a
nuclear localization signal (NLS) [22].
We generated a truncated HMGB2 derivative

which lacks the acidic tail previously reported to
decrease the affinity of HMG proteins for DNA
[23]. This bacterially expressed HMGB2
186
fragment
formed complexes with plasmid DNA, and mediated
gene delivery into different established tumor cell
lines more efficiently than poly(l-lysine). Further-
more, by including additional functional domains
such as the HIV-1 TAT protein transduction domain
(PTD), the NLS of the simian virus 40 (SV40) large
T-antigen, or the importin-b-binding domain (IBB)
of human importin-a2, alternative gene delivery vec-
tors were produced that displayed markedly enhanced
transfection efficiency.
A
B
Fig. 1. Construction and bacterial expression of HMGB2
186
. (A) Schematic representation of the human HMGB2 protein and the expression
construct encoding truncated HMGB2
186
. Full-length HMGB2 consists of HMG box A, a linker region (L), HMG box B, a joiner region (J) and
an acidic C-terminal tail. The bacterial expression vector pSW5-HMGB2
186
encodes under the control of the isopropyl b-D-thiogalactopyrano-
side-inducible tac promoter (tac) amino acids 1–186 of human HMGB2 fused to C-terminal Myc (M) and polyhistidine (H) tags. (B)
SDS ⁄ PAGE (lanes 1–4) and immunoblot analysis (lanes 5–8) of bacterial lysate (lanes 1, 5), flow through (lanes 2, 6), wash (lanes 3, 7) and
eluate fraction (lanes 4, 8) during purification of HMGB2
186

by Ni
2+
affinity chromatography. HMGB2
186
was identified with Myc-tag-specific
antibody 9E10 followed by horseradish peroxidase-coupled secondary antibody and chemiluminescent detection.
Recombinant HMGB2 mediates gene delivery A. Sloots and W. S. Wels
4222 FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS
Results
Truncated human HMGB2
186
is readily expressed
in bacteria
For bacterial expression of a truncated derivative of
HMGB2 that lacks the acidic tail, a cDNA fragment
encoding amino acids 1–186 (HMGB2
186
) was derived
by PCR and inserted into the expression vector pSW5
[24]. In the resulting pSW5-HMGB2
186
plasmid, cyto-
plasmic expression of HMGB2
186
fused to C-terminal
Myc and His-tags is controlled by an isopropyl b-d-thio-
galactopyranoside-inducible tac promoter (Fig. 1A).
HMGB2
186
protein was expressed in Escherichia coli

strain BL21(kDE3)trxB
–
[25] and purified from bacterial
lysates under native conditions by Ni
2+
-affinity chro-
matography as described in Experimental Procedures.
In SDS⁄ PAGE analysis, purified HMGB2
186
could be
detected as a single band with an apparent molecular
mass slightly larger than calculated from its sequence
(24.2 kDa; Fig. 1B). Similar results were obtained in
immunoblot analysis with mAb 9E10 specific for the
Myc tag included in the molecule, further confirming
the identity of the recombinant protein (Fig. 1B).
Recombinant HMGB2
186
binds to plasmid DNA
and cell surfaces
Binding of HMGB2
186
protein to plasmid DNA was
investigated in a gel retardation experiment (Fig. 2A).
Increasing amounts of purified HMGB2
186
were incu-
bated with 0.5 lg reporter gene plasmid pCMS-eGFP-
E2C-Luc. The electrophoretic mobility of the resulting
protein–DNA complexes in 1% agarose gel was then

determined. In the absence of HMGB2
186
, plasmid
DNA migrated as expected, with supercoiled and
relaxed forms as separate bands (Fig. 2A, lane 1). The
addition of 0.7 or 1.4 lg HMGB2
186
already retarded
the plasmid DNA substantially (Fig. 2A, lanes 3, 4),
and with 2.8 lg or more, maximal retardation of plas-
mid DNA was achieved (Fig. 2A, lane 5). In contrast,
BSA, included as a control protein, had no effect on
the electrophoretic mobility of plasmid DNA. Plasmid
DNA in HMGB2
186
complexes containing 2.8 lgor
more purified protein similar to DNA in poly(l-lysine)
complexes was completely protected against degrada-
tion by DNase I (data not shown).
Another important determinant of successful gene
delivery is the binding of DNA–vector complexes to
target cells. This can occur by direct interaction of
transfection complexes with integral components of
the cell membrane, or by binding to other molecules
expressed on the cell surface [15,16]. Binding of recom-
binant HMGB2
186
to target cells was investigated by
FACS analysis using human HeLa cells and COS-7
African green monkey kidney cells as a model. The

cells were incubated with increasing concentrations of
purified HMGB2
186
, and bound protein was detected
with mAb 9E10 recognizing the Myc tag included in
the molecule, followed by fluorescein isothiocyanate-
conjugated or phycoerythrin-conjugated secondary anti-
bodies. As shown in Fig. 2B, concentration-dependent
and saturable binding of HMGB2
186
to the cell surface
was detected, suggesting specific interaction with an as
yet unidentified target molecule.
HMGB2
186
facilitates gene delivery into COS-7
cells
To investigate HMGB2
186
-mediated gene transfer, the
pCMS-eGFP-E2C-Luc reporter gene plasmid was used
which encodes enhanced green fluorescent protein
(eGFP) and an optimized form of firefly luciferase under
the control of the SV40 enhancer ⁄ promoter and the
cytomegalovirus immediate early promoter, respectively.
Transfection complexes containing increasing amounts
of purified HMGB2
186
protein and 2.3 lg pCMS-eGFP-
E2C-Luc DNA were added to COS-7 cells in standard

growth medium with serum, and left on the cells for 4 h
before the medium was exchanged. Cells were lysed 40 h
later, and luciferase activity was measured. For compar-
ison, control cells were transfected with poly(l-lysine)–
DNA complexes containing 2.3 lg reporter plasmid
and a 60-fold molar excess of poly(l-lysine) as described
[26]. The results are shown in Fig. 2C. Concentration-
dependent HMGB2
186
-mediated gene delivery was
found, which was more efficient than poly(l-lysine)-
mediated transfection at an HMGB2
186
amount of
21.5 lg (representing a protein ⁄ DNA mass ratio of 9.5)
or higher. In this experiment, maximal reporter gene
expression was reached with 43 lg HMGB2
186
, with
luciferase activity eight times higher than in the poly
(l-lysine) control. Interestingly, in COS-7 cells, effi-
ciency of HMGB2
186
-mediated gene transfer decreased
again at higher HMGB2
186
concentrations, possibly
because of saturation of cell surface molecules occupied
by excess free HMGB2
186

protein.
To investigate the possible involvement of the endo-
cytic pathway in the internalization of HMGB2
186
–
DNA complexes, cells were also transfected in the
presence of the endosomolytic agent chloroquine
[27,28]. As shown in Fig. 2D, the efficiency of
HMGB2
186
-mediated gene transfer into COS-7 cells
was substantially increased by chloroquine. Luciferase
A. Sloots and W. S. Wels Recombinant HMGB2 mediates gene delivery
FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS 4223
activity was enhanced  15-fold and 55-fold for com-
plexes containing 43 and 64.5 lg HMGB2
186
, respect-
ively. Taken together, these data show that HMGB2
186
on its own is able to mediate nonviral gene delivery,
and strongly suggest that HMGB2
186
–DNA complexes
enter the cells through the endocytic pathway.
Construction of HMGB2
186
derivatives containing
the TAT PTD or the SV40 T-antigen NLS
To investigate whether HMGB2

186
-mediated gene
transfer can be improved by including in the molecule
a cell-penetrating peptide, the PTD of the HIV-1 TAT
A
B
CD
Fig. 2. Functional characterization of purified HMGB2
186
. (A) DNA binding was analyzed by agarose gel electrophoresis of 0.5 lg pCMS-
eGFP-E2C-Luc plasmid DNA in the absence of protein (lane 1), or after incubation with the indicated amounts of recombinant HMGB2
186
(lanes 2–5). The positions of supercoiled and open circular DNA, and protein–DNA complexes are indicated by arrows. (B) Binding of
HMGB2
186
to the surface of HeLa cells (left panel) and COS-7 cells (middle panel) was investigated by FACS analysis. Cells were incubated
with the indicated concentrations of purified HMGB2
186
protein. Then bound protein was detected with Myc-tag-specific antibody 9E10 fol-
lowed by fluorescein isothiocyanate-conjugated (HeLa) or phycoerythrin-conjugated (COS-7) anti-mouse IgG. Control cells were incubated
with antibodies in the absence of HMGB2
186
(open area). Mean fluorescence values (arbitrary units) were calculated from the COS-7 FACS
experiments and plotted against the protein concentrations used (16.5 n
M to 18.59 lM) (right panel). HMGB2
186
-mediated gene transfer into
COS-7 cells in the absence (C) or presence of 100 l
M chloroquine (D). Cells were seeded at a density of 7 · 10
4

cells per well 24 h before
transfection. Transfection complexes were formed by incubating the indicated amounts of purified HMGB2
186
with pCMS-eGFP-E2C-Luc
reporter plasmid before addition to the cells in normal growth medium (2.3 lg DNA per well). Control cells were treated with poly(
L-lysine)
(pL)–DNA complexes (open bar). After 4 h the medium was exchanged and cells were grown for another 40 h before they were harvested
for analysis. Luciferase activity is expressed in relative light units (RLU)Æ(mg total protein)
)1
.
Recombinant HMGB2 mediates gene delivery A. Sloots and W. S. Wels
4224 FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS
protein (amino acids 47–57) flanked by additional gly-
cine residues was fused to the N-terminus of the
HMGB2
186
fragment (Fig. 3A; TAT
PTD
sequence
shown in Table 1). This TAT fragment also includes a
nonclassical NLS [29]. Therefore, to examine the pos-
sible effect of a heterologous NLS on HMGB2
186
-
mediated gene transfer independent of cell-penetrating
activity, a similar HMGB2
186
derivative was construc-
ted which carries the classical NLS of the SV40 large
T-antigen [30] at the N-terminus (Fig. 3A; SV40

NLS
sequence shown in Table 1). TAT
PTD
-HMGB2
186
and
SV40
NLS
-HMGB2
186
proteins were expressed in E. coli
and purified from bacterial lysates as described above
for unmodified HMGB2
186
. As expected, in compari-
son with HMGB2
186
, a slight increase in the apparent
molecular mass was found for the fusion proteins in
SDS ⁄ PAGE and immunoblot analysis with antibody
against HMGB2 (Fig. 3B).
A
B
C
Fig. 3. (A, B) Bacterial expression of HMGB2
186
derivatives carrying the TAT PTD (TAT
PTD
; amino acids 47–57) or the NLS of SV40 large
T-antigen (SV40

NLS
). (A) Schematic representation of the TAT
PTD
-HMGB2
186
and SV40
NLS
-HMGB2
186
expression cassettes also encoding
C-terminal Myc (M) and polyhistidine (H) tags, inserted into plasmid pSW5. (B) SDS ⁄ PAGE (lanes 1, 2) and immunoblot analysis (lanes 3–5)
of purified SV40
NLS
-HMGB2
186
(lanes 1, 5) and TAT
PTD
-HMGB2
186
(lanes 2, 4) in comparison with unmodified HMGB2
186
(lane 3). The pro-
teins were identified with an HMGB2-specific antibody followed by horseradish peroxidase-coupled secondary antibody and chemilumines-
cent detection. (C) DNA binding was analyzed by agarose gel electrophoresis of 0.5 lg pCMS-eGFP-E2C-Luc plasmid DNA in the absence of
protein (lanes 1, 6), or after incubation with the indicated amounts of recombinant SV40
NLS
-HMGB2
186
(lanes 2–5) or TAT
PTD

-HMGB2
186
(lanes 7–10). The positions of supercoiled and open circular DNA, and protein–DNA complexes are indicated by arrows.
Table 1. N-terminal amino acid sequence and calculated isoelectric point of HMGB2 derivatives.
HMGB2 derivative N-terminal sequence
a
Calculated IEP
b
HMGB2
186
MGT-GKGD 9.75
TAT
PTD
-HMGB2
186
MG-YGRKKRRQRRR-GT-GKGD 9.99
SV40
NLS
-HMGB2
186
MPR-PKKKRKVEDP-GT-GKGD. 9.84
IBB-HMGB2
186
MPRHHHHHH-AARLHRFKNKGKDSTEMRRRRIEVNVELRKAKKDDQMLKRRNVSSFPD-GT-GK GD 9.99
a
Sequences of functional peptide domains are underlined. Positively charged amino acids are indicated in bold. The first four residues of the
HMGB2 1–186 fragment are shown in italics. The N-terminal Met residue, His-tag and sequences encoded by cloning linkers are also inclu-
ded.
b
Isoelectric points were calculated using Expasy ( />A. Sloots and W. S. Wels Recombinant HMGB2 mediates gene delivery

FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS 4225
The DNA-binding activity of purified TAT
PTD
-
HMGB2
186
and SV40
NLS
-HMGB2
186
proteins was
analyzed in gel retardation experiments as described
above. As shown in Fig. 3C, in the presence of 1.7 or
1.8 lg of the proteins, 0.5 lg pCMS-eGFP-E2C-Luc
plasmid DNA was markedly retarded in the agarose
gels, indicating effective DNA binding and complex
formation.
TAT
PTD
-HMGB2
186
and SV40
NLS
-HMGB2
186
mediate gene transfer
To examine the gene-transfer activity of HMGB2
186
fusion proteins, protein–DNA complexes were pre-
pared by mixing increasing amounts of HMGB2

186
,
TAT
PTD
-HMGB2
186
and SV40
NLS
-HMGB2
186
with
pCMS-eGFP-E2C-Luc DNA (2.3 lg per well) as des-
cribed above. Transfection complexes were added to
COS-7 and human HepG2 cells in complete growth
medium containing serum for 4 h. Luciferase activities
were determined 40 h later. In both, COS-7 and
HepG2 cells, gene delivery mediated by TAT
PTD
-
HMGB2
186
and SV40
NLS
-HMGB2
186
was more
efficient than HMGB2
186
-mediated transfection
(Fig. 4A,D). Maximal luciferase activity in COS-7 cells

was achieved using 16.5 lg TAT
PTD
-HMGB2
186
(molar protein to DNA ratio of 1270 : 1), which was
38 times higher than after HMGB2
186
-mediated
transfection with 13 lg of the unmodified protein
(molar protein to DNA ratio of 1070 : 1), and still
A
BC
DE
Fig. 4. Gene transfer mediated by SV40
NLS
-
HMGB2
186
and TAT
PTD
-HMGB2
186
. COS-7
(A, B), HeLa (C) or HepG2 cells (D, E) were
transfected with protein–DNA complexes
containing 2.3 lg pCMS-eGFP-E2C-Luc
reporter plasmid and the indicated amounts
of purified SV40
NLS
-HMGB2

186
or TAT
PTD
-
HMGB2
186
proteins in the absence (A, D) or
presence of 100 l
M chloroquine (B, C, E) as
described in the legend of Fig. 2. Protein–
DNA complexes prepared with unmodified
HMGB2
186
were included for comparison.
Luciferase activity is expressed in relative
light units (RLU)Æ(mg total protein)
)1
.
Recombinant HMGB2 mediates gene delivery A. Sloots and W. S. Wels
4226 FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS
seven times higher than with 52 lg HMGB2
186
(molar
protein to DNA ratio of 4200 : 1; Fig. 4A, left and
right panels). In HepG2 cells, TAT
PTD
-HMGB2
186
was more than three times more effective than
HMGB2

186
at the highest protein amounts used
(Fig. 4D). Unexpectedly, luciferase activities measured
after SV40
NLS
-HMGB2
186
-mediated gene delivery into
COS-7 and HepG2 cells were very similar to those
obtained after TAT
PTD
-HMGB2
186
-mediated transfec-
tion at comparable molar protein to DNA ratios. For
gene delivery into HepG2 cells, SV40
NLS
-HMGB2
186
was even slightly more effective than TAT
PTD
-
HMGB2
186
(Fig. 4A,D, middle panels).
As inclusion of the TAT
PTD
domain did not
enhance transfection efficiency more than inclusion of
the SV40

NLS
domain, the NLS function of TAT rather
than its membrane-translocating properties may be
responsible for the enhanced reporter gene expression
observed in comparison with HMGB2
186
. Therefore,
to analyze the possible involvement of the endocytic
pathway during TAT
PTD
-HMGB2
186
-mediated and
SV40
NLS
-HMGB2
186
-mediated gene transfer, the effect
of chloroquine on transfection efficiency was investi-
gated, with HeLa cells included in the analysis in
addition to COS-7 and HepG2 cells. As shown in
Fig. 4B,C,E, in all three cell lines not only HMGB2
186
-
mediated and SV40
NLS
-HMGB2
186
-mediated, but
also TAT

PTD
-HMGB2
186
-mediated gene delivery was
enhanced by chloroquine to a similar degree in com-
parison with transfection by the respective proteins in
the absence of an endosomolytic reagent. These results
suggest that TAT
PTD
-HMGB2
186
protein–DNA com-
plexes may indeed enter cells primarily via an endo-
somal pathway, rather than by direct membrane
translocation as originally hypothesized for TAT
PTD
-
containing fusion proteins.
An HMGB2
186
derivative carrying the IBB of
importin-a displays enhanced gene-delivery
activity
The viral TAT
PTD
domain can function as a nonclassi-
cal NLS by direct interaction with importin-b [29]. To
investigate whether attachment of an endogenous cellu-
lar importin-b binding sequence to HMGB2
186

enhan-
ces its gene-delivery activity to a similar extent, the
IBB of human importin-a2 (amino acids 11–58)
together with an N-terminal His-tag was fused to the
HMGB2
186
sequence (Fig. 5A; IBB sequence shown in
Table 1). The resulting IBB-HMGB2
186
fusion protein
was expressed in E. coli as described above for
HMGB2
186
. After purification, IBB-HMGB2
186
could
be identified as the major band on SDS ⁄ PAGE and
immunoblot analysis, and DNA-binding activity sim-
ilar to that of HMGB2
186
was confirmed in a gel retar-
dation assay.
The gene-transfer activity of IBB-HMGB2
186
was
investigated in transfection experiments as described
above using COS-7 cells. HMGB2
186
, SV40
NLS

-
HMGB2
186
and TAT
PTD
-HMGB2
186
containing com-
plexes were included for comparison. As controls,
gene-transfer complexes were also prepared with
poly(l-lysine) and polyethyleneimine (PEI). Surpris-
ingly, in the absence of chloroquine, gene transfer
mediated by IBB-HMGB2
186
was clearly less efficient
than SV40
NLS
-HMGB2
186
-mediated and TAT
PTD
-
HMGB2
186
-mediated transfection at similar protein to
DNA ratios, and was comparable to gene delivery
mediated by unmodified HMGB2
186
(Fig. 5B). In stri-
king contrast, in the presence of chloroquine, IBB-

HMGB2
186
-containing complexes were remarkably
effective, with ensuing luciferase activities higher than
those achieved after transfection with TAT
PTD
-
HMGB2
186
–DNA complexes (Fig. 5C). Importantly,
with the exception of HMGB2
186
, which in this experi-
ment was used at suboptimal protein concentrations,
gene transfer mediated by the recombinant HMGB2
186
derivatives was comparable to or more efficient than
poly(l-lysine)-mediated transfection. Not unexpectedly,
transfection of cells with PEI, which is considered to
be one of the most efficient nonviral gene-delivery
agents currently available, was still more effective than
IBB-HMGB2
186
-mediated gene transfer in the presence
of chloroquine. However, the differences were not
dramatic, with luciferase activities measured after
IBB-HMGB2
186
-mediated and PEI-mediated transfec-
tion being of the same order of magnitude (less than

fourfold difference; Fig. 5C).
To analyze whether IBB-HMGB2
186
-mediated gene
delivery is dependent on specific cell binding similar to
that found for uncomplexed HMGB2
186
, competition
experiments were performed. COS-7 cells were pre-
treated for 15 min with increasing amounts of IBB-
HMGB2
186
or unmodified HMGB2
186
protein before
IBB-HMGB2
186
–DNA complexes at an optimal pro-
tein to DNA mass ratio of 8.2 were added. Control
cells were treated with IBB-HMGB2
186
–DNA com-
plexes in the absence of competitor. As shown in
Fig. 6, in the presence of an amount of free IBB-
HMGB2
186
comparable to the amount of protein in
the complex, transfection efficiency was only 37% of
controls, and was reduced further to 22% if the con-
centration of free IBB-HMGB2

186
was doubled. Free
unmodified HMGB2
186
also affected transfection effi-
ciency of IBB-HMGB2
186
–DNA complexes, but to a
lesser extent than identical molar concentrations of
A. Sloots and W. S. Wels Recombinant HMGB2 mediates gene delivery
FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS 4227
free IBB-HMGB2
186
(reduction to 66% and 43% of
controls). These data suggest that IBB-HMGB2
186
–
DNA complexes bind to the cell surface primarily via
the HMGB2
186
domain and to the same structures
recognized by uncomplexed HMGB2
186
derivatives.
The pCMS-eGFP-E2C-Luc reporter plasmid in
addition to luciferase also encodes eGFP, which allows
identification of transfected cells individually. In a
separate experiment, COS-7 cells were incubated in the
presence of chloroquine with protein–DNA complexes
at a molar protein to DNA ratio of  1200 : 1 as des-

cribed above. At 20 h after transfection, cells were first
analyzed by fluorescence microscopy (Fig. 7A), fol-
lowed another 20 h later by FACS analysis for quanti-
fication of eGFP-expressing cells (Fig. 7B). Confirming
the results obtained in the luciferase assays, in the
presence of chloroquine, IBB-HMGB2
186
was again the
most effective HMGB2
186
derivative resulting in suc-
cessful transfection and measurable eGFP expression
in 13% of the cells, which compares well with PEI-
mediated transfection (19% of eGFP-positive cells).
Taken together, these data suggest that IBB has no
effect during uptake of protein–DNA complexes via
the endocytic pathway, but upon release from endo-
somes with the help of chloroquine may serve as a
pure NLS, greatly improving transport of plasmid
DNA to the nucleus and enabling efficient gene expres-
sion. In contrast, SV40
NLS
and TAT
PTD
may also con-
tribute other activities that improve nonviral gene
delivery, as indicated by their ability to enhance trans-
fection efficiency already in the absence of an endo-
somolytic reagent.
A

B
C
Fig. 5. (A) Bacterial expression of an
HMGB2
186
derivative containing IBB (amino
acids 11–58) of human importin-a2. The
IBB-HMGB2
186
expression cassette also
encodes a C-terminal Myc (M), and N-ter-
minal and C-terminal polyhistidine (H) tags,
inserted into plasmid pSW5. Shown below
are SDS ⁄ PAGE (lane 1) and immunoblot
analysis (lane 3) of purified IBB-HMGB2
186
in comparison with unmodified HMGB2
186
(lanes 2, 4). The proteins were identified
with an HMGB2-specific antibody followed
by horseradish peroxidase-coupled secon-
dary antibody and chemiluminescent detec-
tion. DNA binding was analyzed by agarose
gel electrophoresis of 0.5 lg pCMS-eGFP-
E2C-Luc plasmid DNA in the absence of
protein (lane 1), or after incubation with the
indicated amounts of recombinant IBB-
HMGB2
186
(lanes 2, 3). The positions of

supercoiled and open circular DNA, and
protein–DNA complexes are indicated by
arrows. (B, C) IBB-HMGB2
186
-mediated
gene transfer. COS-7 cells were transfected
with protein–DNA complexes containing
2.3 lg pCMS-eGFP-E2C-Luc reporter plas-
mid and the indicated amounts of purified
IBB-HMGB2
186
, or HMGB2
186
, SV40
NLS
-
HMGB2
186
,orTAT
PTD
-HMGB2
186
proteins
for comparison in the absence (B) or pres-
ence of 100 l
M chloroquine (C) as described
in the legend of Fig. 2. Control cells were
treated with poly(
L-lysine) (pL)–DNA com-
plexes, or PEI–DNA complexes at an N ⁄ P

ratio of 10 (open bars). Luciferase activity is
expressed in relative light units (RLU)Æ(mg
total protein)
)1
.
Recombinant HMGB2 mediates gene delivery A. Sloots and W. S. Wels
4228 FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS
Discussion
The ability to condense DNA is essential for a nonviral
vector to be successful as a gene-delivery reagent [4].
Although synthetic vectors are most commonly used
for nonviral gene transfer, certain DNA-condensing
proteins of mammalian origin have also been shown to
facilitate cellular uptake of plasmid DNA. Previous
studies on histones and HMG proteins as gene-delivery
reagents mainly used full-length proteins purified or
enriched from animal tissues [5–11], whereas only a few
groups have so far attempted to utilize such proteins in
recombinant form [12,14]. Here we have generated a
novel recombinant derivative of the human HMG pro-
tein HMGB2, which facilitates nonviral delivery of
plasmid DNA into tumor cells. By complementing
the DNA-binding activity of this HMGB2
186
variant
with additional functional domains from heterologous
proteins, we achieved a marked increase in protein-
mediated transfection.
HMGB2 is a member of the HMGB subfamily of
nonhistone chromatin proteins, which also includes

HMGB1 and the more recently discovered HMGB3
[31,32]. HMGB proteins have little or no sequence spe-
cificity and bind preferentially to certain (distorted)
DNA structures [32,33]. Thereby the acidic C-terminus
appears to control DNA binding, as truncated
HMGB1 and HMGB2 lacking this sequence displayed
increased affinity for DNA [23,34–37]. Deletion of the
acidic tail also largely abolished the differences in
DNA binding between the three HMGB proteins [23].
Consequently, as a reagent for nonviral gene delivery,
we constructed a truncated HMGB2 derivative that
encompasses amino acids 1–186 of the human protein,
but lacks the acidic C-terminal part. Whereas in a
previous report bacterial expression of full-length
HMGB1 had only resulted in very low amounts of
recombinant protein [38], here we encountered no
problems with regard to expression of truncated
HMGB2
186
in E. coli, and high yields of soluble
recombinant protein could be obtained after purifica-
tion from bacterial lysates under native conditions
[up to 4 mgÆ(L culture)
)1
].
Purified HMGB2
186
was able to condense plasmid
DNA in a concentration-dependent manner, indicated
by marked retardation of the resulting protein–DNA

complexes in an agarose gel. Starting at a molar pro-
tein to DNA ratio of 1050 : 1 (representing an
HMGB2
186
protein to DNA mass ratio of 5.6), maxi-
mum retardation of plasmid DNA was achieved. Pre-
viously for recombinant full-length HMGB1 expressed
in yeast, in a similar assay, a protein to DNA ratio of
7000 : 1 was required [12]. This suggests that removal
of the acidic tail indeed facilitated enhanced DNA
binding of the protein, even if general differences
between HMGB1 and HMGB2 may have partially
contributed to this effect. As shown by FACS analysis,
HMGB2
186
also bound to the surface of established
tumor cell lines in a concentration-dependent and satu-
rable manner. Although at present the exact nature of
this interaction remains unclear, our data suggest spe-
cific binding of HMGB2
186
to defined target molecules
rather than unspecific attachment to the cell mem-
brane. This also appears to be the case for HMGB2
186
derivatives complexed with DNA, as transfection
efficiency of preformed IBB-HMGB2
186
–DNA com-
plexes was significantly reduced when free HMGB2

186
or IBB-HMGB2
186
proteins were added as competi-
tors. Likewise, gene transfer was decreased when
HMGB2
186
derivatives were present in too high
amounts in transfection complexes. For HMGB2, in
contrast with the related HMGB1 molecule, so far no
extracellular activity has been reported [32,39]. Never-
theless, owing to the high homology between these
proteins (80% amino-acid sequence identity), HMGB2
may bind to the same or similar cell surface molecules
as HMGB1, which include the receptor for advanced
glycation end products (RAGE) [40] and syndecan-1
[41]. Interestingly, PEI–DNA complexes have also
recently been found to be internalized by adherent cells
after binding to syndecans [42].
Fig. 6. Effect of uncomplexed HMGB2
186
derivatives on transfec-
tion efficiency of IBB-HMGB2
186
–DNA complexes. COS-7 cells
were transfected in the presence of 100 l
M chloroquine with
protein–DNA complexes containing 2.3 lg pCMS-eGFP-E2C-Luc
reporter plasmid and 18.8 lg purified IBB-HMGB2
186

protein
(protein ⁄ DNA mass ratio of 8.2) as described in the legend of
Fig. 2. Before transfection, cells were treated for 15 min with
uncomplexed IBB-HMGB2
186
or HMGB2
186
proteins as indicated.
A. Sloots and W. S. Wels Recombinant HMGB2 mediates gene delivery
FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS 4229
In in vitro transfection experiments, treatment of dif-
ferent tumor cell lines with HMGB2
186
–DNA com-
plexes resulted in transient expression of luciferase
and eGFP reporter genes. Depending on the protein
amounts used, gene transfer was more efficient than
with poly(l-lysine)–DNA complexes, and was not inhib-
ited by the serum in the culture medium. Although in
these experiments poly(l-lysine) was used at amounts
favoring the formation of electroneutral complexes, and
no attempt was made to optimize transfection by this
control reagent, gene delivery by modified HMGB2
186
derivatives compared well with PEI-mediated transfec-
tion at optimal N⁄ P ratios previously shown to be far
superior to poly(l-lysine) [43]. At the concentrations
tested, cell viability was not affected by protein–DNA
complexes containing HMGB2
186

or HMGB2
186
fusion proteins (data not shown). Addition of the endo-
somolytic reagent, chloroquine, strongly enhanced the
A
B
Fig. 7. Analysis of eGFP expression after transfection with protein–DNA complexes containing HMGB2
186
derivatives. COS-7 cells were
transfected with protein–DNA complexes containing 2.3 lg pCMS-eGFP-E2C-Luc reporter plasmid and the indicated amounts of purified
HMGB2
186
, SV40
NLS
-HMGB2
186
, TAT
PTD
-HMGB2
186
, or IBB-HMGB2
186
proteins in the presence of 100 lM chloroquine as described in the
legend of Fig. 2. Control cells were treated with poly(
L-lysine) (pL)–DNA complexes, or PEI–DNA complexes at an N ⁄ P ratio of 12. (A) Micro-
scopic analysis of eGFP expressing cells 20 h after transfection. Corresponding representative fields after fluorescence and bright field micro-
scopy are shown. (B) Quantification of eGFP-expressing cells by FACS analysis. At 40 h after transfection, cells were collected and analyzed
by flow cytometry. Untreated COS-7 cells were used as a control. The cut-off for eGFP expression was set at the fluorescence intensity at
which 99.84% of the control cells displayed a lower fluorescent signal. The bars represent the percentage of eGFP expressing cells (1 · 10
4

cells per well analyzed in duplicate). Magnification, 100·.
Recombinant HMGB2 mediates gene delivery A. Sloots and W. S. Wels
4230 FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS
transfection efficiency of HMGB2
186
–DNA complexes,
indicating cellular uptake via endocytosis. To improve
gene transfer by nonviral vectors that do not possess
intrinsic endosome escape mechanisms, these vectors
can be modified to include functional entities that confer
active endosome escape, such as bacterial translocation
domains [26,44,45] or membrane active peptides [46]. As
an alternative, here we fused the PTD of HIV-1 TAT,
initially reported to shuttle heterologous proteins or
protein domains into cells by direct membrane trans-
location [47], to the N-terminus of HMGB2
186
. As this
TAT
PTD
peptide also contains an NLS [29], allowing
transport of cargo molecules into the nucleus, it may
be especially attractive for gene-delivery applications.
For comparison, a similar HMGB2
186
derivative was
constructed that contained the NLS of the SV40 large
T-antigen [30].
As expected, TAT
PTD

-HMGB2
186
clearly proved
more effective in mediating gene transfer than
HMGB2
186
. Importantly, this was also the case for
SV40
NLS
-HMGB2
186
, which displayed transfection effi-
ciencies comparable to those of TAT
PTD
-HMGB2
186
.
Because no cell-penetrating activity has been reported
for the SV40 NLS, these results suggest that the NLS
in the TAT
PTD
peptide rather than TAT
PTD
-dependent
membrane translocation may be responsible for
increased gene transfer. Recently, internalization of
TAT fusion proteins via lipid raft-dependent macro-
pinocytosis, a specialized form of endocytosis, has
been demonstrated, questioning the possibility of
TAT-mediated direct penetration of the lipid bilayer

[48]. Uptake of TAT
PTD
-HMGB2
186
–DNA complexes
by endocytosis is supported by the data obtained in
transfection experiments in the presence of chloro-
quine. Like HMGB2
186
and SV40
NLS
-HMGB2
186
,
TAT
PTD
-HMGB2
186
-mediated gene delivery was mark-
edly enhanced by the endosomolytic reagent, indicating
that TAT
PTD
-HMGB2
186
utilizes predominantly the
endosomal pathway to transport plasmid DNA into
the cell. Similarly, recent reports on gene delivery
mediated by monomeric or oligomeric TAT
PTD
pep-

tides on their own suggest cellular uptake of peptide–
DNA complexes through mechanisms that involve
endocytosis [49,50]. This was also found for a fusion
peptide consisting of the TAT
PTD
domain and 15
lysine residues for DNA binding, which showed a
strong dependence on the presence of chloroquine [51].
For the sequence-specific yeast GAL4 DNA-binding
domain, it was reported that binding to the cognate
nucleotide sequence prevents simultaneous association
via its nonclassical NLS with the nuclear import recep-
tor importin-b [52]. Similarly, in HMGB2
186
complexed
with DNA the functionality of the endogenous
HMGB2 NLS could be compromised, as this sequence
is part of the HMG boxes that are directly involved in
DNA binding [22]. Reconstitution or enhancement of
nuclear tropism may therefore at least partly explain
the enhanced gene-transfer activity of the SV40
NLS
-
HMGB2
186
and TAT
PTD
-HMGB2
186
proteins that add

heterologous NLS sequences to HMGB2
186
.
To further investigate the effect of a protein domain
that strongly enhances nuclear localization, we gener-
ated an HMGB2
186
derivative carrying at its N-termi-
nus the IBB of human importin-a2. The classical
nuclear import pathway in eukaryotic cells is initiated
by binding of an NLS to importin-a in the cytoplasm.
Via the IBB of importin-a, the resulting complex then
interacts with importin-b, which facilitates active trans-
port of the trimeric complex into the nucleus [53,54].
Consequently, once in the cytoplasm of the target cell,
IBB-HMGB2
186
, which is similar to proteins such as
HIV-1 TAT that contain a nonclassical NLS, can be
expected to bind directly to importin-b, thereby bypas-
sing the requirement to first recruit importin-a. Indeed,
in the presence of chloroquine, IBB-HMGB2
186
at simi-
lar protein to DNA ratios was found to be the most
potent gene-delivery reagent of the four HMGB2
186
derivatives tested.
IBB, TAT
PTD

and SV40
NLS
are cationic peptides
and increase the overall positive charge of the corres-
ponding HMGB2
186
fusion proteins, which can be
expected to result in improved DNA binding. Indeed,
IBB-HMGB2
186
, TAT
PTD
-HMGB2
186
and SV40
NLS
-
HMGB2
186
displayed a comparable increase in DNA-
condensing activity in gel retardation assays (maximum
DNA retardation by the fusion proteins at protein to
DNA mass ratios of 3.4–5.4), which may contribute to
the enhanced transfection efficiency observed. At least
for IBB-HMGB2
186
, however, the increase in positive
charge alone cannot explain the beneficial effect of the
NLS domain on gene transfer, because, despite its cati-
onic nature, it was no more effective than unmodified

HMGB2
186
in transfection experiments without the
addition of chloroquine. This supports the interpre-
tation that the addition of an NLS to HMGB2
186
enhances gene transfer primarily by improving nuclear
delivery of the DNA cargo. Data from previous
studies on transfection by DNA constructs to which a
single or multiple NLS were attached suggest that
covalently linked peptide–DNA complexes may be able
to directly utilize the endogenous nuclear import
machinery [55,56]. Although still speculative at the
moment, it appears possible that, in a similar fashion,
enough of the NLS-HMGB2
186
derivatives remained
bound to plasmid DNA in the intracellular milieu to
facilitate transfer through nuclear pores.
A. Sloots and W. S. Wels Recombinant HMGB2 mediates gene delivery
FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS 4231
Although originally a nuclear protein, HMGB1 can
also be released by certain cell types such as activated
monocytes and macrophages, and appears to play an
important role in inflammation [57]. It induces matur-
ation of dendritic cells [58], and can act as a potent
vaccine adjuvant and enhance tumor-specific immune
responses [59]. At present, it is unclear whether HMGB2
shares this immunomodulatory activity. However,
HMGB2 displays very high homology to HMGB1

in the area where its cytokine-like activity has been
localized (first 20 residues of HMG box B, amino
acids 89–108) [60]. Owing to the much more restricted
expression pattern of HMGB2 in the adult in compari-
son with HMGB1 [39], this may not be physiologically
relevant. Nevertheless, as this part of the protein is
also present in HMGB2
186
, it could be important for
potential in vivo applications of HMGB2-based gene
delivery vectors. Although immunostimulatory activity
may limit their application in cases where the induction
of an immune response is undesirable, it could be bene-
ficial if HMGB2 derivatives are used as components of
DNA vaccines where the protein could play a dual role
as a DNA-delivery reagent and a vaccine adjuvant.
Our results demonstrate that bacterially expressed
derivatives of human HMGB2 facilitate efficient non-
viral gene delivery. Endogenous DNA-condensing and
cell-binding activities of HMGB2 could successfully be
complemented by the addition of NLS from different
heterologous proteins, suggesting the resulting mole-
cules as versatile building blocks for further optimiza-
tion as protein-based transfection reagents for
applications in gene therapy.
Experimental procedures
Cell culture
SV40-transformed COS-7 African Green monkey kidney
cells and human HeLa cervical carcinoma cells were main-
tained in Dulbecco’s modified Eagle’s medium (Cambrex,

Verviers, Belgium), and human HepG2 hepatocellular carci-
noma cells were cultured in RPMI 1640 medium (Camb-
rex). The established cell lines were obtained from
American Type Culture Collection (ATCC) (Manassas,
VA, USA). Each medium was supplemented with 10% fetal
bovine serum, 2 mml-glutamine, 100 U ÆmL
)1
penicillin
and 100 lgÆmL
)1
streptomycin. Cells were cultivated at
37 °C in a humidified atmosphere of air and 5% CO
2
.
Construction of HMGB2 proteins
A cDNA fragment encoding amino acids 1–186 of
human HMGB2 was generated by PCR using plasmid
pGEX-5X-1-hHMGB2 [61] as a template and two oligo-
nucleotide primers complementary to regions in the
human HMGB2 cDNA, 5¢-AAAAAAACATATGGGTA
CCGGTAAAGGAGACCCCA-3¢ and 5¢-AAATCTAGAA
GCTTTGGTTCGTTCTTCTTCT-3¢, introducing NdeI
and KpnI, and HindIII and XbaI restriction sites at the
5¢ and 3¢ ends of the amplified fragment, respectively.
Digestion of the PCR product with NdeI and XbaI
restriction enzymes resulted in two fragments of 436 and
139 bp because of an internal NdeI site. The fragments
were inserted simultaneously into NdeI ⁄ XbaI-digested
bacterial expression plasmid pSW5 [24]. Integrity of the
resulting plasmid pSW5-HMGB2

186
was confirmed by
restriction analysis and DNA sequencing. The plasmid
encodes under the control of an isopropyl b-d-thiogal-
actopyranoside-inducible tac promoter amino acids 1–186
of human HMGB2 fused to a C-terminal Myc tag [62]
and a cluster of six histidine residues (His-tag).
For the construction of HMGB2 derivatives containing
functional domains from heterologous proteins, an HIV-1
TAT sequence encoding the PTD (TAT
PTD
; amino acids
47–57) of the protein flanked by glycine residues was
derived by annealing the oligonucleotides 5¢-TATGGGT
TATGGCAGGAAGAAGCGGAGACAGCGACGAAGA
GGTACCTGCT-3¢ and 5¢-CTAGAGCAGGTACCTCTT
CGTCGCTGTCTCCGCTTCTTCCTGCCATAACCCA-3¢.
Similarly, a double-stranded oligonucleotide encoding the
NLS (amino acids 126–135) of the SV40 large tumor anti-
gen (SV40
NLS
) was derived by annealing the oligonucleo-
tides 5¢-TATGCCTAGGCCGAAGAAAAAGCGTAAAG
TTGAAGACCCGGGTAC-3¢ and 5¢-CCGGGTCTTCAA
CTTTACGCTTTTTCTTCGGCCTAGGCA-3¢. A cDNA
fragment encoding the IBB of human importin-a2 (amino
acids 11–58), including an additional N-terminal His-tag,
was generated by PCR using plasmid pQE70-importin-a
(kindly provided by D. Go
¨

rlich, Heidelberg, Gemany) as a
template and the two oligonucleotide primers 5¢-TT
TCCTAGGCACCATCATCACCATCACGCTGCCCGT
CTTCACAGATTC-3¢ and 5¢-TTTGGTACCATCAGG
AAATGAGCTTAC-3¢. The expression plasmids pSW5-
TAT
PTD
-HMGB2
186
, pSW5-SV40
NLS
-HMGB2
186
and
pSW5-IBB-HMGB2
186
, which encode TAT
PTD
-HMGB2
186
,
SV40
NLS
-HMGB2
186
and IBB-HMGB2
186
fusion proteins
each carrying C-terminal Myc and His-tags, were derived
by stepwise assembly of the respective double-stranded

oligonucleotides or cDNA fragments with the HMGB2
186
fragment in plasmid pSW5.
Bacterial expression and purification of
recombinant proteins
E. coli BL21(kDE3)trxB
–
[25] carrying the respective
expression plasmids was grown to D
600
of 0.8 at 20 ° C
in Luria–Bertani medium containing 0.6% glucose and
Recombinant HMGB2 mediates gene delivery A. Sloots and W. S. Wels
4232 FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS
100 lgÆmL
)1
ampicillin. Then protein expression was
induced by the addition of isopropyl thio-b-d-galactoside to
a final concentration of 0.25 mm followed by another 2 h
of incubation at 20 °C. Cells were harvested by centrifuga-
tion at 6000 g for 15 min at 4 °C. Bacterial pellets were
resuspended in NaCl ⁄ P
i
, pH 8.0, containing 1 m NaCl,
10 mm imidazole, and 0.75 mm phenylmethylsulfonyl
fluoride, and lysed in a French pressure cell. Lysates were
cleared by centrifugation at 40 000 g for 30 min at 4 °C,
and loaded on a Ni
2+
-saturated Chelating Sepharose Fast

Flow column (Amersham Biosciences, Freiburg, Germany)
connected to an FPLC system (Amersham Biosciences).
Bound, nonspecific proteins were removed by washing the
column with NaCl ⁄ P
i
, pH 8.0, containing 1 m NaCl and
40 mm imidazole. Specifically bound proteins were eluted
with NaCl ⁄ P
i
, pH 8.0, containing 1 m NaCl and 250 mm
imidazole and dialyzed immediately against NaCl ⁄ P
i
(pH 7.4) ⁄ 20% glycerol at 4 °C overnight. Dialyzed proteins
were passed through a 0.45-lm filter (Millipore, Eschborn,
Germany), divided into aliquots, and stored at )80 °C
until use.
Purified proteins were analyzed by SDS ⁄ PAGE followed
by Coomassie staining or immunoblotting on to Immobi-
lon-P membranes (Millipore). Recombinant proteins were
detected with Myc tag-specific monoclonal antibody 9E10
[62] or HMGB2-specific polyclonal antibody (BD Bio-
sciences, Heidelberg, Germany) followed by horseradish
peroxidase-conjugated species-specific secondary antibodies
and chemiluminescent detection using the ECL kit (Amer-
sham Biosciences).
Construction of the reporter gene construct
pCMS-eGFP-E2C-Luc
For DNA mobility-shift assays and transfection experi-
ments, the reporter gene construct pCMS-eGFP-E2C-Luc
was used, which contains the eGFP-encoding gene under

the control of the SV40 enhancer ⁄ promoter, and the firefly
Photinus pyralis luciferase gene under the control of the
cytomegalovirus immediate early promoter. Briefly, the
luciferase gene was obtained by XhoI ⁄ XbaI digestion of
plasmid pGL3-enhancer (Promega, Mannheim, Germany),
and inserted into XhoI ⁄ Xba I-digested plasmid pCMS-eGFP
(BD Biosciences Clontech, Heidelberg, Germany). Subse-
quently, part of the filamentous phage f1 origin of pCMS-
eGFP was replaced with the DNA-recognition sequence for
the synthetic DNA-binding protein E2C (not relevant for
the experiments performed with pCMS-eGFP-E2C-Luc in
this study).
Flow cytometric and microscopic analysis
Binding of HMGB2
186
protein to HeLa and COS-7 cells
was investigated by incubating 5 · 10
5
cells with increasing
concentrations of the protein in a total volume of 20 or
100 lL NaCl ⁄ P
i
⁄ 20% glycerol for 40 min at 4 °C.
Unbound proteins were removed, and bound proteins were
detected by incubation with the Myc-tag-specific mAb 9E10
(20 lL hybridoma supernatant), followed by fluorescein
isothiocyanate-labeled (HeLa) or phycoerythrin-labeled
(COS-7) secondary antibodies (Jackson Immunoresearch,
West Grove, PA, USA). Fluorescence of cells was then ana-
lyzed with FACScan or FACSCalibur fluorescence-activa-

ted cell sorters (Becton Dickinson, Heidelberg, Germany).
Intracellular expression of eGFP upon transfection with
the pCMS-eGFP-E2C-Luc reporter gene construct was
determined by directly measuring green fluorescence in
FACS analysis, or by microscopic analysis with a Nikon
eclipse TE300 fluorescence microscope (Nikon, Du
¨
sseldorf,
Germany).
DNA mobility-shift assay
Protein–DNA complexes were prepared by mixing 0.5 lg
plasmid pCMS-eGFP-E2C-Luc with increasing amounts of
purified HMGB2
186
or HMGB2
186
fusion proteins in a final
volume of 22.5 lL dialysis buffer (NaCl ⁄ P
i
⁄ 20% glycerol,
pH 7.4) to which 2.5 lL10· Hepes buffer (500 mm Hepes,
pH 7.5, 500 mm KCl, 50 mm MgCl
2
,1mm ZnCl
2
) had
been added, followed by incubation for 30 min at room
temperature. After complex formation, samples were mixed
with standard DNA loading buffer and analyzed by
agarose gel electrophoresis (1%) in the presence of ethidium

bromide to visualize the DNA.
Transfection of cells
Plasmid pCMS-eGFP-E2C-Luc (6963 bp) was isolated from
E. coli XL1-blue (Stratagene, Heidelberg, Germany) using a
plasmid purification kit (Qiagen, Hilden, Germany) accord-
ing to the manufacturer’s instructions. For transfection,
cells were seeded in 12-well tissue culture plates at a density
of 3.5 or 7 · 10
4
cells per well and grown overnight at
37 °C. The growth medium was exchanged with 1 mL per
well fresh growth medium containing standard serum con-
centrations 1 h before addition of protein–DNA complexes.
Transfection complexes were prepared in microcentrifuge
tubes by incubating different amounts of purified fusion
proteins with 5 lg plasmid pCMS-eGFP-E2C-Luc for
30 min at room temperature in a final volume of 220 lL
transfection buffer containing 50 mm Hepes, pH 7.5,
50 mm KCl, 5 mm MgCl
2
, and 100 lm ZnCl
2
[26].
As controls, pCMS-eGFP-E2C-Luc complexes with
poly(l-lysine)-HBr with an average degree of polymeriza-
tion of 236 residues [poly(l-lysine)
236
; Sigma, Deisenhofen,
Germany] were prepared as described [26]. For PEI-medi-
ated transfection, a 100 mm stock solution of branched PEI

(average molecular mass 25 kDa; Sigma) in water was pre-
pared, the pH was adjusted to 7.0 with HCl, and the solu-
tion was passed through a 0.22-lm filter. The working
A. Sloots and W. S. Wels Recombinant HMGB2 mediates gene delivery
FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS 4233
concentration of 10 mm was obtained by 10-fold dilution
with water. PEI–DNA complexes were derived by mixing
5 lg plasmid pCMS-eGFP-E2C-Luc in 110 lL Dulbecco’s
NaCl ⁄ P
i
(Cambrex) with an amount of 10 mm PEI required
to obtain the indicated N ⁄ P ratios [43] diluted in 110 lL
Dulbecco’s NaCl ⁄ P
i
, and incubation for 30 min at room
temperature. Protein–DNA, poly(l-lysine)–DNA and PEI–
DNA complexes were then added to the cells (100 lL per
well). After 4 h of incubation at 37 °C, the growth medium
was exchanged, and the cells were incubated for another
40 h before they were harvested for analysis.
Luciferase assay
After removal of growth medium, cells were washed and
then lysed in 100 lL buffer containing 25 mm glycylglycine,
pH 7.8, 15% glycerol, 8 mm MgSO
4
,1mm dithiothreitol,
1mm EDTA, and 1% Triton X-100. Lysates were cleared
by centrifugation, and protein concentrations were deter-
mined using the Bio-Rad Protein Assay (Bio-Rad, Munich,
Germany). Ten microliters of cleared lysates were mixed

with the same volume of dilution buffer containing 25 mm
glycylglycine, pH 7.8, 10 mm MgSO
4
, and 5 mm ATP in
96-well microlyteä2 plates (ThermoLabsystems, Franklin,
MA, USA). Luciferase activity was then measured at room
temperature for 30 s in a luminometer Microlumat LB 96P
(Berthold, Munich, Germany) with automatic injection of
50 lL luciferin solution containing 250 lm luciferin (Sig-
ma), 25 mm glycylglycine, pH 7.8, and 0.5 mm CoA (Roche
Molecular Biochemicals, Mannheim, Germany). Luciferase
activity was calculated as relative light unitsÆ(mg total cellu-
lar protein)
)1
.
Acknowledgements
This work was supported in part by a grant from the
Bundesministerium fu
¨
r Bildung und Forschung
(BMBF) FKZ 0312174 ⁄ 0. We thank Drs Jutta Schu
¨
ller
and Norbert Lehming for plasmid pGEX-5X-1-
hHMGB2, Dr Dirk Go
¨
rlich for plasmid pQE70-impor-
tin-a, and Dr Christoph Uherek for many helpful
discussions and suggestions.
References

1 Ferber D (2001) Gene therapy. Safer and virus-free?
Science 294, 1638–1642.
2 Dewey RA, Morrissey G, Cowsill CM, Stone D,
Bolognani F, Dodd NJ, Southgate TD, Klatzmann D,
Lassmann H, Castro MG, et al. (1999) Chronic brain
inflammation and persistent herpes simplex virus 1
thymidine kinase expression in survivors of syngeneic
glioma treated by adenovirus-mediated gene therapy:
implications for clinical trials. Nat Med 5, 1256–1263.
3 Hacein-Bey-Abina S, von Kalle C, Schmidt M, Le Deist
F, Wulffraat N, McIntyre E, Radford I, Villeval JL,
Fraser CC, Cavazzana-Calvo M, et al. (2003) A serious
adverse event after successful gene therapy for X-linked
severe combined immunodeficiency. N Engl J Med 348,
255–256.
4 Wagner E, Cotten M, Foisner R & Birnstiel ML (1991)
Transferrin-polycation-DNA complexes: the effect of
polycations on the structure of the complex and DNA
delivery to cells. Proc Natl Acad Sci USA 88, 4255–4259.
5 Chen J, Stickles RJ & Daichendt KA (1994) Galactosy-
lated histone-mediated gene transfer and expression.
Hum Gene Ther 5, 429–435.
6 Balicki D & Beutler E (1997) Histone H2A significantly
enhances in vitro DNA transfection. Mol Med 3, 782–
787.
7 Zaitsev SV, Haberland A, Otto A, Vorob’ev VI, Haller
H & Bottger M (1997) H1 and HMG17 extracted from
calf thymus nuclei are efficient DNA carriers in gene
transfer. Gene Ther 4, 586–592.
8 Demirhan I, Hasselmayer O, Chandra A, Ehemann M

& Chandra P (1998) Histone-mediated transfer and
expression of the HIV-1 tat gene in Jurkat cells. J Hum
Virol 1, 430–440.
9 Balicki D, Reisfeld RA, Pertl U, Beutler E & Lode HN
(2000) Histone H2A-mediated transient cytokine gene
delivery induces efficient antitumor responses in murine
neuroblastoma. Proc Natl Acad Sci USA 97, 11500–
11504.
10 Bottger M, Vogel F, Platzer M, Kiessling U, Grade K
& Strauss M (1988) Condensation of vector DNA by
the chromosomal protein HMG1 results in efficient
transfection. Biochim Biophys Acta 950 , 221–228.
11 Bottger M, Zaitsev SV, Otto A, Haberland A &
Vorob’ev VI (1998) Acid nuclear extracts as mediators
of gene transfer and expression. Biochim Biophys Acta
1395, 78–87.
12 Mistry AR, Falciola L, Monaco L, Tagliabue R,
Acerbis G, Knight A, Harbottle RP, Soria M, Bianchi
ME, Coutelle C, et al. (1997) Recombinant HMG1
protein produced in Pichia pastoris: a nonviral gene
delivery agent. Biotechniques 22, 718–729.
13 Wagner JP, Quill DM & Pettijohn DE (1995) Increased
DNA-bending activity and higher affinity DNA binding
of high mobility group protein HMG-1 prepared with-
out acids. J Biol Chem 270, 7394–7398.
14 Fritz JD, Herweijer H, Zhang G & Wolff JA (1996)
Gene transfer into mammalian cells using histone-
condensed plasmid DNA. Hum Gene Ther 7, 1395–1404.
15 Nishikawa M & Huang L (2001) Nonviral vectors in
the new millennium: delivery barriers in gene transfer.

Hum Gene Ther 12, 861–870.
16 Wiethoff CM & Middaugh CR (2003) Barriers to non-
viral gene delivery. J Pharm Sci 92 , 203–217.
Recombinant HMGB2 mediates gene delivery A. Sloots and W. S. Wels
4234 FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS
17 Uherek C & Wels W (2000) DNA-carrier proteins for
targeted gene delivery. Adv Drug Deliv Rev 44, 153–
166.
18 Reeck GR, Isackson PJ & Teller DC (1982) Domain
structure in high molecular weight high mobility group
nonhistone chromatin proteins. Nature 300, 76–78.
19 Cary PD, Turner CH, Leung I, Mayes E & Crane-
Robinson C (1984) Conformation and domain structure
of the non-histone chromosomal proteins HMG 1 and
2. Domain interactions. Eur J Biochem 143, 323–330.
20 Shirakawa H, Tsuda K & Yoshida M (1990) Primary
structure of non-histone chromosomal protein HMG2
revealed by the nucleotide sequence. Biochemistry 29,
4419–4423.
21 Yoshioka K, Saito K, Tanabe T, Yamamoto A, Ando
Y, Nakamura Y, Shirakawa H & Yoshida M (1999)
Differences in DNA recognition and conformational
change activity between boxes A and B in HMG2 pro-
tein. Biochemistry 38, 589–595.
22 Shirakawa H, Tanigawa T, Sugiyama S, Kobayashi M,
Terashima T, Yoshida K, Arai T & Yoshida M (1997)
Nuclear accumulation of HMG2 protein is mediated by
basic regions interspaced with a long DNA-binding
sequence, and retention within the nucleus requires the
acidic carboxyl terminus. Biochemistry 36 , 5992–5999.

23 Lee KB & Thomas JO (2000) The effect of the acidic
tail on the DNA-binding properties of the HMG1,2
class of proteins: insights from tail switching and tail
removal. J Mol Biol 304, 135–149.
24 Rohrbach F, Gerstmayer B, Biburger M & Wels W
(2000) Construction and characterization of bispecific
costimulatory molecules containing a minimized CD86
(B7–2) domain and single-chain antibody fragments for
tumor targeting. Clin Cancer Res 6, 4314–4322.
25 Proba K, Ge L & Pluckthun A (1995) Functional anti-
body single-chain fragments from the cytoplasm of
Escherichia coli: influence of thioredoxin reductase
(TrxB). Gene 159, 203–207.
26 Uherek C, Fominaya J & Wels W (1998) A modular
DNA carrier protein based on the structure of
diphtheria toxin mediates target cell-specific gene
delivery. J Biol Chem 273, 8835–8841.
27 Cotten M, Langle-Rouault F, Kirlappos H, Wagner E,
Mechtler K, Zenke M, Beug H & Birnstiel ML (1990)
Transferrin-polycation-mediated introduction of DNA
into human leukemic cells: stimulation by agents that
affect the survival of transfected DNA or modulate
transferrin receptor levels. Proc Natl Acad Sci USA 87,
4033–4037.
28 Erbacher P, Roche AC, Monsigny M & Midoux P
(1996) Putative role of chloroquine in gene transfer into
a human hepatoma cell line by DNA ⁄ lactosylated poly-
lysine complexes. Exp Cell Res 225, 186–194.
29 Truant R & Cullen BR (1999) The arginine-rich
domains present in human immunodeficiency virus type

1 Tat and Rev function as direct importin beta-depen-
dent nuclear localization signals. Mol Cell Biol 19,
1210–1217.
30 Kalderon D, Richardson WD, Markham AF & Smith
AE (1984) Sequence requirements for nuclear location
of simian virus 40 large-T antigen. Nature 311, 33–38.
31 Bustin M (1999) Regulation of DNA-dependent activit-
ies by the functional motifs of the high-mobility-group
chromosomal proteins. Mol Cell Biol 19, 5237–5246.
32 Muller S, Scaffidi P, Degryse B, Bonaldi T, Ronfani L,
Agresti A, Beltrame M & Bianchi ME (2001) New
EMBO members’ review: the double life of HMGB1
chromatin protein: architectural factor and extracellular
signal. Embo J 20 , 4337–4340.
33 Thomas JO & Travers AA (2001) HMG1 and 2, and
related ‘architectural’ DNA-binding proteins. Trends
Biochem Sci 26, 167–174.
34 Sheflin LG, Fucile NW & Spaulding SW (1993) The
specific interactions of HMG 1 and 2 with negatively
supercoiled DNA are modulated by their acidic C-term-
inal domains and involve cysteine residues in their
HMG 1 ⁄ 2 boxes. Biochemistry 32, 3238–3248.
35 Stros M, Stokrova J & Thomas JO (1994) DNA looping
by the HMG-box domains of HMG1 and modulation
of DNA binding by the acidic C-terminal domain.
Nucleic Acids Res 22, 1044–1051.
36 Muller S, Bianchi ME & Knapp S (2001) Thermody-
namics of HMGB1 interaction with duplex DNA. Bio-
chemistry 40, 10254–10261.
37 Bonaldi T, Langst G, Strohner R, Becker PB & Bianchi

ME (2002) The DNA chaperone HMGB1 facilitates
ACF ⁄ CHRAC-dependent nucleosome sliding. EMBO J
21, 6865–6873.
38 Bianchi ME (1991) Production of functional rat HMG1
protein in Escherichia coli. Gene 104, 271–275.
39 Muller S, Ronfani L & Bianchi ME (2004) Regulated
expression and subcellular localization of HMGB1, a
chromatin protein with a cytokine function. J Intern
Med 255, 332–343.
40 Hori O, Brett J, Slattery T, Cao R, Zhang J, Chen JX,
Nagashima M, Lundh ER, Vijay S, Nitecki D, et al.
(1995) The receptor for advanced glycation end pro-
ducts (RAGE) is a cellular binding site for amphoterin.
Mediation of neurite outgrowth and co-expression of
rage and amphoterin in the developing nervous system.
J Biol Chem 270, 25752–25761.
41 Salmivirta M, Rauvala H, Elenius K & Jalkanen M
(1992) Neurite growth-promoting protein (amphoterin,
p30) binds syndecan. Exp Cell Res 200, 444–451.
42 Kopatz I, Remy JS & Behr JP (2004) A model for
non-viral gene delivery: through syndecan adhesion
molecules and powered by actin. J Gene Med 6, 769–
776.
43 Boussif O, Lezoualc’h F, Zanta MA, Mergny MD,
Scherman D, Demeneix B & Behr JP (1995) A versatile
A. Sloots and W. S. Wels Recombinant HMGB2 mediates gene delivery
FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS 4235
vector for gene and oligonucleotide transfer into cells in
culture and in vivo: polyethylenimine. Proc Natl Acad
Sci USA 92, 7297–7301.

44 Fominaya J & Wels W (1996) Target cell-specific DNA
transfer mediated by a chimeric multidomain protein.
Novel non-viral gene delivery system. J Biol Chem 271,
10560–10568.
45 Gaur R, Gupta P, Goyal A, Wels W & Singh Y (2002)
Delivery of nucleic acid into mammalian cells by
anthrax toxin. Biochem Biophys Res Commun 297,
1121–1127.
46 Cho YW, Kim JD & Park K (2003) Polycation gene
delivery systems: escape from endosomes to cytosol.
J Pharm Pharmacol 55 , 721–734.
47 Snyder EL & Dowdy SF (2004) Cell penetrating pep-
tides in drug delivery. Pharm Res 21, 389–393.
48 Wadia JS, Stan RV & Dowdy SF (2004) Transducible
TAT-HA fusogenic peptide enhances escape of TAT-
fusion proteins after lipid raft macropinocytosis. Nat
Med 10, 310–315.
49 Ignatovich IA, Dizhe EB, Pavlotskaya AV, Akifiev BN,
Burov SV, Orlov SV & Perevozchikov AP (2003) Com-
plexes of plasmid DNA with basic domain 47–57 of the
HIV-1 Tat protein are transferred to mammalian cells
by endocytosis-mediated pathways. J Biol Chem 278,
42625–42636.
50 Rudolph C, Plank C, Lausier J, Schillinger U, Muller
RH & Rosenecker J (2003) Oligomers of the arginine-
rich motif of the HIV-1 TAT protein are capable of
transferring plasmid DNA into cells. J Biol Chem 278,
11411–11418.
51 Hashida H, Miyamoto M, Cho Y, Hida Y, Kato K,
Kurokawa T, Okushiba S, Kondo S, Dosaka-Akita H

& Katoh H (2004) Fusion of HIV-1 Tat protein trans-
duction domain to poly-lysine as a new DNA delivery
tool. Br J Cancer 90, 1252–1258.
52 Chan CK, Hubner S, Hu W & Jans DA (1998) Mutual
exclusivity of DNA binding and nuclear localization sig-
nal recognition by the yeast transcription factor GAL4:
implications for nonviral DNA delivery. Gene Ther 5,
1204–1212.
53 Gorlich D, Henklein P, Laskey RA & Hartmann E
(1996) A 41 amino acid motif in importin-alpha confers
binding to importin-beta and hence transit into the
nucleus. EMBO J 15, 1810–1817.
54 Gorlich D & Kutay U (1999) Transport between the cell
nucleus and the cytoplasm. Annu Rev Cell Dev Biol 15,
607–660.
55 Sebestyen MG, Ludtke JJ, Bassik MC, Zhang G, Budker
V, Lukhtanov EA, Hagstrom JE & Wolff JA (1998)
DNA vector chemistry: the covalent attachment of signal
peptides to plasmid DNA. Nat Biotechnol 16, 80–85.
56 Zanta MA, Belguise-Valladier P & Behr JP (1999) Gene
delivery: a single nuclear localization signal peptide is
sufficient to carry DNA to the cell nucleus. Proc Natl
Acad Sci USA 96, 91–96.
57 Wang H, Bloom O, Zhang M, Vishnubhakat JM,
Ombrellino M, Che J, Frazier A, Yang H, Ivanova S,
Borovikova L, et al. (1999) HMG-1 as a late mediator
of endotoxin lethality in mice. Science 285, 248–251.
58 Messmer D, Yang H, Telusma G, Knoll F, Li J,
Messmer B, Tracey KJ & Chiorazzi N (2004) High
mobility group box protein 1: an endogenous signal

for dendritic cell maturation and Th1 polarization.
J Immunol 173, 307–313.
59 Rovere-Querini P, Capobianco A, Scaffidi P, Valentinis
B, Catalanotti F, Giazzon M, Dumitriu IE, Muller S,
Iannacone M, Traversari C, et al. (2004) HMGB1 is an
endogenous immune adjuvant released by necrotic cells.
EMBO Rep 5, 825–830.
60 Li J, Kokkola R, Tabibzadeh S, Yang R, Ochani M,
Qiang X, Harris HE, Czura CJ, Wang H, Ulloa L, et al.
(2003) Structural basis for the proinflammatory cytokine
activity of high mobility group box 1. Mol Med 9, 37–45.
61 Lehming N, Le Saux A, Schuller J & Ptashne M (1998)
Chromatin components as part of a putative transcrip-
tional repressing complex. Proc Natl Acad Sci USA 95,
7322–7326.
62 Evan GI, Lewis GK, Ramsay G & Bishop JM (1985)
Isolation of monoclonal antibodies specific for human
c-myc proto-oncogene product. Mol Cell Biol 5, 3610–
3616.
Recombinant HMGB2 mediates gene delivery A. Sloots and W. S. Wels
4236 FEBS Journal 272 (2005) 4221–4236 ª 2005 FEBS