Gene Therapy (2007) 14, 1175–1180
& 2007 Nature Publishing Group All rights reserved 0969-7128/07 $30.00
www.nature.com/gt
SHORT COMMUNICATION
Spontaneous cellular uptake of exogenous
messenger RNA in vivo is nucleic acid-specific,
saturable and ion dependent
J Probst1,2, B Weide3, B Scheel1,2, BJ Pichler4, I Hoerr2, H-G Rammensee1 and S Pascolo1,2
1
Department of Immunology, Institute for Cell Biology, University of Tübingen, Tübingen, Germany; 2CureVac GmbH, Tübingen,
Germany; 3Department of Dermatology, University of Tübingen, Tübingen, Germany and 4Laboratory for Preclinical Imaging and
Imaging Technology, Department of Radiology, University of Tübingen, Tübingen, Germany
The development of new treatments in the post-genomic era
requires methods for safe delivery of foreign genetic
information in vivo. As a transient, natural and controllable
alternative to recombinant viruses or plasmid DNA (pDNA),
purified or in vitro transcribed messenger RNA (mRNA)
can be used for the expression of any therapeutic protein
in vitro and in vivo. As it has been shown previously, the
simple injection of naked mRNA results in local uptake and
expression. We show here that this process, in the skin,
can greatly be modulated according to the injection solution
composition and blocked by an excess of competing
nucleic acids or a drug affecting cytosolic mobility. Different
cell types at the site of injection can take up the foreign
nucleic acid molecules and the protein translated from
this is detected for no more than a few days. To test this
gene transfer method in humans, we produced in vitro
transcribed mRNA under good manufacturing practice
(GMP) conditions in a dedicated facility. After injection
into the human dermis, we could document the translation
of the exogenous mRNA. Our results pave the way toward
the use of mRNA as a vehicle for transient gene delivery in
humans.
Gene Therapy (2007) 14, 1175–1180; doi:10.1038/
sj.gt.3302964; published online 3 May 2007
Keywords: RNA-transfection; DNA-transfection; endosomes; calcium; GMP
Messenger RNA (mRNA) is a transient copy of the
coding genomic information in any living organism. Its
potential as a gene therapy vehicle was hindered, on the
one hand, because of the technical difficulties and cost of
mRNA production in research laboratories and, on the
other because of the general belief that ubiquitous
intracellular and extracellular RNases would quickly
destroy the molecule and limit its efficacy as a genetic
tool in vivo. However, Wolff et al.1 have shown 16 years
ago that, in mice, the injection of naked genetic
information in the form of plasmid DNA (pDNA) or
mRNA can lead to protein expression. Following these
results, many studies were undertaken which demonstrated that naked pDNA can be used for vaccination.2,3
mRNA was rarely exploited until the late nineties when
Gilboa’s group showed that adoptive transfer of mRNAtransfected dendritic cells (DCs) primes a T-cell immune
response.4 But, the direct injection of naked mRNA for
vaccination or gene complementation remained quite
unexplored, being reported in only four articles from
three different teams5–8 (for a review on mRNA-based
vaccines, see Pascolo9). Because of the safety (minimal
Correspondence: Dr S Pascolo, Department of Immunology,
Institute for Cell Biology, University of Tübingen, Auf der
Morgenstelle 15, Tübingen 72076, Germany.
E-mail: [email protected]
Received 4 February 2007; revised 19 March 2007; accepted 20
March 2007; published online 3 May 2007
vector, completely and naturally catabolized in several
hours), ease (no need for infrastructures for cell culture)
and versatility (any mRNA of interest can be produced)
of naked mRNA-based therapies compared to DNAbased therapies,10,11 we decided to study and improve
the intradermal delivery of this genetic vehicle. As
shown in Figure 1, using intradermal injection of globin
UTR-stabilized (RNActive, CureVac GmbH, Tübingen,
Germany) luciferase-encoding mRNA in the ear pinna,12
we could investigate the effect of the injection solution
composition on the efficacy of in vivo mRNA transfer (for
method details, see Supplementary Information and
Supplementary Figure S1). We tested the injection
solutions used by authors of the few published works
that were phosphate-buffered saline (PBS) and Hepes/
NaCl, and in addition we tested Ringer lactate, which is a
standard injection solution in humans. Figure 1a shows
that RNActive dissolved in Ringer lactate gives a
significant higher amount of luciferase expression than
mRNA resuspended in Hepes/NaCl or PBS. Ringer
lactate and PBS have three differences: only the former
contains lactate and calcium, whereas the latter contains
phosphate. We tested which of the two ions present in
Ringer lactate is responsible for the observed enhanced
mRNA transfer. Self-made Ringer without lactate or
calcium was compared to self-made Ringer lactate. As
shown in Figure 1b, although Ringer without lactate
gave luciferase expression comparable to that observed
with complete Ringer lactate, the absence of calcium in
Uptake of naked mRNA
J Probst et al
1176
the injection solution lowered significantly (P ¼ 0.004) the
efficiency of vector transfer to a level similar to PBS or
Hepes/NaCl. When injecting naked mRNA solubilized
in solutions with higher or lower concentrations of
calcium compared to the standard Ringer lactate, we did
not see higher amounts of luciferase protein (data not
shown). Similarly, higher or lower Ringer lactate concentrations in the final injected solution did not increase
the amount of luciferase proteins detected in the ear
Gene Therapy
(Supplementary Information and Supplementary Figure
S2). The quantity of luciferase produced by injection of
RNActive in the skin can be higher than that obtained
when transfecting primary cells using optimal electroporation as described13 (Figure 1c). Thus, the uptake and
expression of injected naked RNActive is an efficacious
process. Using Ringer lactate, we studied the kinetics of
pDNA and mRNA expression after injection. As shown
in Figure 1d, the amount of luciferase detected after
Uptake of naked mRNA
J Probst et al
RNActive injection peaks at ca. 17 h and is undetectable
after 3 days. In comparison, pDNA injection gave a
delayed protein expression that peaks at 3 days after
injection and eventually lasted for more than 50 days. Of
note, at the peak of expression, RNActive can give more
protein expression than pDNA. The results were
confirmed when we studied the luciferase expression
using an optical imaging system (Figure 1e). This
technology demonstrates that luciferase activity is
detectable only at the site of RNActive injection. Thus,
after intradermal injection, neither the RNActive nor the
cells that take it up migrate to a distant site. Checking the
effect of RNA concentration, we observed a linear
increase in luciferase expression when increasing the
RNActive amount from 1 to 5 mg in 100 ml injection
volume (Figure 1f). Surprisingly, from 5 mg up to 80 mg of
RNActive in 100 ml Ringer lactate, no change in the
amount of luciferase activity was observed (Figure 1f).
These results are different to those reported by Wolf et al.
where the mRNA was injected intramuscularly. In Wolf’s
report, the more mRNA was injected, the more luciferase
activity was detected. In contrast, our results indicate
that the uptake of naked RNActive in the dermis is a
saturable process. Therefore, these results suggest that at
least one step in the uptake of exogenous RNActive by
some cells of the dermis is mediated by an active
biological mechanism. To investigate this mechanism, we
studied the capacity of different macromolecules to
compete with the RNActive uptake. A mass excess of
irrelevant RNActive (coding Influenza matrix M1),
fragmented (sonicated) luciferase-coding RNActive,
dsRNA, pDNA, CpG DNA oligonucleotide or peptide
was mixed with 10 mg of RNActive-coding luciferase and
injected in the ear pinna. As shown in Figure 1g, the
amount of resulting luciferase molecules was not
changed when a peptide was added to the RNActive,
but was significantly reduced (Po0.001 for CpG oligonucleotide, pDNA and irrelevant mRNA, P ¼ 0.017 for
sonicated luc mRNA and P ¼ 0.04 for dsRNA) when an
excess of any type of nucleic acid was added. Thus, the
uptake-limiting step for injected RNActive’s expression
is a mechanism specific for nucleic acids. We then
investigated further this active pathway that allows
RNActive to reach the cytosol by using available
inhibitors that block macropinocytose (Amiloride), clathrin-coated pits (Chlorpromazine), Caveolae (Nystatin)
or depolymerization of actin and thus cytosolic mobility
(Cytochalasin B). Each of these inhibitors is toxic and we
used the dose reported to be efficacious and not lethal in
vivo in mice.13–16 Under these conditions, only Cytochalasin B significantly inhibited the expression of luciferase
from injected RNActive (Figure 1h). Thus, it appears that
the exogenous RNA goes through moving vesicles before
being delivered into the cytosol. However, the lack of a
statistically significant effect of Amiloride, Chlorpromazine and Nystatin may be due to too low dosage, poor
availability of the drug at the site of RNA injection or
unknown side effects. Thus, these experiments do not
definitively rule out the involvement of macropinocytose, clathrin-coated pits and caveolae, respectively in
the uptake of exogenous mRNA. We conclude from these
experiments that the local uptake of exogenous RNA by
cells in the dermis can be enhanced by the addition of
calcium, is mediated by a saturable mechanism that
is specific for nucleic acids and involves movements
of vesicles. The expression of the transgene in the form
of RNActive is transient: using this method, a sustained
topical expression of a foreign molecule would require
reinjection of RNActive approximately every 3 days. In
contrast to pDNA or recombinant viruses that can show
uncontrolled persistence, no accumulation or overexpression of the transgene-encoded protein could occur
even with repeated mRNA applications.
In order to identify the cell type(s) that take up
the exogenous RNA at the site of injection, we used
b-galactosidase-coding RNActive. Approximately 16 h
after injection, ears were frozen in embedding medium
and cut into sections using a cryomicrotom. Sections
were incubated with X-gal containing solution (for the
optimization of the staining conditions, see Supplementary Figures S3–S8). Cells that had taken up exogenous
RNA were identified by blue staining in sections (Figure 2a).
Starting from the basal to the apical part of the ear, we
visualized sections and documented the presence of blue
cells. As shown in the diagram of Figure 2b, from one to
ten blue cells could be observed in consecutive sections
spanning more than 1 mm at the site of injection (starting
at 3.00 mm from the base of the ear to 4.2 mm from the
base of the ear). The partial section shown in panel a is
the one located at 3720 mm from the top of the ear, it is
labeled a. The characterization of blue cells by antibody
staining was difficult because the blue color interferes
1177
Figure 1 Transfer of luciferase-coding mRNA in vivo in mouse skin. For the experiments depicted in (a–h), mouse ear pinnae were injected
with a fixed volume of 100 ml containing globin UTR-stabilized mRNA (RNActive) that encodes luciferase. The luciferase activity recovered
from complete mouse ear lysate is indicated in amount of molecules on the y axis. The detection limit is 4–6 million luciferase molecules
depending on the set of experiments and is symbolized by a line. Each dot indicates a single ear; the number of tested ears is written under
the x axis after ‘N ¼ ’. Statistical significance between the experimental groups and the control group is shown in the upper part of each figure
by a line ending on the P-value calculated according to a Mann–Whitney rank sum test. In (a, b, f, g and h), ears were removed 18 h after
injection. In (a), different injection solutions were tested: PBS, Hepes/NaCl and Ringer lactate. In (b), the lack of lactate or calcium in Ringer
was tested. In (c), the amount of luciferase molecules produced by human DCs electroporated in vitro in standard conditions (ca. 4 million of
cells in 200 ml OptiMEM pulse at 300 V and 150 mF in a 0.4 cm cuvette) with 10 mg of luciferase-coding RNActive is shown: Luc stands for DCs
transfected with luciferase-encoding RNActive and lacZ stands for DCs transfected with b-galactosidase-encoding RNActive. In (d), the
expression of pDNA or RNActive vector (mRNA) was compared after injection of these nucleic acids dissolved in PBS or Ringer lactate,
respectively. In (e), the localization of luciferase activity and the duration of the expression after pDNA (D) or RNActive (R) injection is
reported using an in vivo imaging system. In (f), the effect of titrating the concentration of RNActive on luciferase activity was tested. In (g),
an excess of several biomolecules (‘Comp’) was added to the RNActive to assess the capacity of these compounds to interfere with RNA
uptake: irR stands for influenza matrix 1-coding RNActive, soR stands for luciferase-coding mRNA that is degraded by sonication, dsR
stands for poly IC double-stranded RNA, dsD stands for double-stranded pDNA, ssD stands for single-stranded CpG phosphorothioate
DNA oligonucleotide and Pep stands for peptide SYFPEITHI. In (h), several drugs were injected in mice before injection of RNActive: Amil
stands for amiloride, Chpr stands for chloropromazine, Nyst stands for nystatin and CytB stands for cytochalasin B. DC, dendritic cells;
mRNA, messenger RNA; PBS, phosphate-buffered saline; pDNA, plasmid DNA.
Gene Therapy
Uptake of naked mRNA
J Probst et al
1178
Figure 2 Characterization of cells taking up b-galactosidase-coding mRNA in vivo. Mouse ear pinnae were injected with 100 ml of Ringer
lactate solution containing 20 mg of RNActive-encoding b-galactosidase. Eighteen hours after injection, ears were frozen in embedding
medium and cut perpendicularly to the mouse ear symmetry axis. Individual sections were stained overnight with a solution containing
X-gal (a) or magenta red (c). The diagram in (b) shows the number of b-galactosidase-expressing cells visible on the stained sections prepared
from the injected area located between 3 and 4.3 mm from the base of the ear. The sections shown in (a and c) are indicated by a, b and w,
respectively, in diagram (b). (c) shows two different sections (in three series) from the same ear stained with magenta red and further
incubated sequentially with biotinylated anti-MHC class II molecules and streptavidin-Alexa 546. All images were acquired on a fluorescence
microscope. The left panels were made using visible light. The middle panels with the fluorescence channel (a purple trace reports the
position of b-galactosidase-expressing cells). The right panels are superpositions of both magenta red and antibody staining using artificial
colors. Most b-galactosidase-expressing cells appear MHC class II negative, but some such as b(2/8) or b(7/8) are possibly MHC class II
positive. MHC, major histocompatibility complex; mRNA, messenger RNA.
strongly with both immunohistochemistry staining and
fluorescence in different confocal microscopy channels
(see Supplementary Figures S5–S7). Thus we tested other
dyes metabolized by b-galactosidase and identified
magenta red as a sensitive method that is compatible
with fluorescence emission at 570 nm (Alexa red, see
Supplementary Figure S6). Using an anti-mouse major
histocompatibility complex (MHC) class II molecule
antibody, we could demonstrate that most b-galactosidase-positive cells are not MHC class II positive. Two
such stained sections (shown in parts) with the corresponding cells labeled b and w are shown in Figure 2b
and c. From the location of the cells, their shape and their
MHC class II-negative phenotype, we conclude that cells
that take up naked RNA at the site of injection are muscle
cells, fibroblasts or keratinocytes (see also Supplementary Figures S7 and S8). For some cells expressing
b-galactosidase, it could not be excluded that they are
Gene Therapy
MHC class II positive as shown for cells b(2/8) and
b(7/8) in Figure 2c, but most MHC class II-positive
cells were negative for b-galactosidase expression.
However, since antigen-presenting cells (APCs) such
as Langerhans cells are very efficient in the degradation
of proteins, our results do not exclude that APCs take up
the foreign naked RNA and present peptides derived
from the neotranslated antigen to the immune system.
To check that injected RNActive molecules are also
taken up and expressed in human skin and in order to
exploit this feature for therapy, we created the facilities,
methods and quality management system that allow the
production of high amounts of RNActive under good
manufacturing practice (GMP) conditions. A purification
step called PUREmessenger was implemented whereby
the long RNA molecules are purified according to their
length. This technology allows the elimination of
contaminating shorter or longer transcripts. The final
Uptake of naked mRNA
J Probst et al
Figure 3 Transfer of luciferase-coding RNActive in vivo in human
skin. GMP quality RNActive-coding luciferase was produced
and dissolved at 0.8 mg/ml in Ringer lactate solution. As shown in
(a), the injection of 100 ml of this GMP quality solution in mouse ears
resulted in the production of more luciferase in comparison to
laboratory-grade RNActive (‘Lab’). The same solution injected
intradermally in human resulted also in the in vivo production
of luciferase as detected in a 3-mm diameter punch biopsy made
on the middle of the injection site (forming a bubble of ca. 2 cm
in diameter) 16 h after injection (b). A punch biopsy made outside
the injected zone did not contain detectable luciferase activity. GMP,
good manufacturing practice.
globin UTR-stabilized molecule is efficient and strongly
depends on the presence of calcium in the injection
solution. We show evidence that this uptake is mediated
by an active and saturable mechanism specific for
nucleic acids. Several cell types in the mouse dermis
are capable of taking up the foreign protein-coding
RNA. We show that the uptake of exogenous mRNA
can also be demonstrated in human skin. The capacity
of cells to actively and qualitatively take up exogenous
coding RNA is a surprising feature. It may be part of
an understudied natural process involved in cell-to-cell
communication. This pathway could be mediated
by secretion and recapture of RNA by neighboring
cells, as originally suggested by Benner.17 Such a
process may be further regulated by the secreted
RNases located between cells of the body. This hypothesis would explain why some extracellular growth
factors such as angiogenins contain RNase activity.
Whether RNA molecules can be short distance cell
growth factors or differentiation signals and whether
this capacity depends on the potential of RNA molecules
to code for proteins is an intriguing concept that
needs further investigation. Meanwhile, this process
can be exploited for local and transient protein
expression as described in the present work. We
anticipate that our results, the infrastructure developed
for the production of GMP-grade in vitro transcribed
mRNA and the demonstration of in vivo mRNA transfer
in human tissues, open the way to mRNA-based
therapies.
1179
Acknowledgements
RNA is a highly pure mRNA transcript. We produced in
this infrastructure pharmaceutical-grade RNActive coding for luciferase. The GMP quality RNA resuspended in
Ringer lactate and injected in mice gave a significantly
higher production of luciferase compared to the laboratory-grade mRNA (Figure 3a). A qualified healthy
individual volunteered for intradermal injections of
150 ml of a Ringer lactate solution containing the GMP
quality RNActive coding for luciferase. Before performing the experiment, a letter of consent was signed by the
volunteer. Sixteen hours after injection, 3 mm diameter
punch biopsies were performed under local anesthesia
with 1% lidocainhydrochlorid (Xylocain). A biopsy made
on the middle of the injection site (‘injected’ in Figure 3b)
and one made outside the injection area (‘not injected’ in
Figure 3b) were snap frozen, crushed and resuspended
in a lysis buffer. As shown in Figure 3b, luciferase
activity was found in the biopsy taken from the injection
site. This result demonstrates the uptake of RNActive in
vivo in the human skin. Since only a 3 mm punch biopsy
– approximately 5 mm2 of skin – was performed in the
injected area of approximately 300 mm2 (the diameter of
the bubble formed by intradermal injection being
approximately 20 mm), only part of the luciferase
activity was collected. This may explain the lower
apparent amount of luciferase obtained in the human
situation compared to the mouse situation where the
whole injection site (the whole ear) was used to measure
luciferase activity.
Our experiments show for the first time that, in mice,
the uptake of injected naked mRNA in the form of a
This work was supported by the Fritz-Bender Stiftung.
JP was supported by the DFG Graduiertenkolleg
‘infektionsbiologie’ 685 of Tübingen.
References
1 Wolff JA, Malone RW, Williams P, Chong W, Acsadi G, Jani A et
al. Direct gene transfer into mouse muscle in vivo. Science 1990;
247: 1465–1468.
2 Robinson HL, Hunt LA, Webster RG. Protection against a
lethal influenza virus challenge by immunization with a
haemagglutinin-expressing plasmid DNA. Vaccine 1993; 11:
957–960.
3 Ulmer JB, Donnelly JJ, Parker SE, Rhodes GH, Felgner PL,
Dwarki VJ et al. Heterologous protection against influenza by
injection of DNA encoding a viral protein. Science 1993; 259:
1745–1749.
4 Boczkowski D, Nair SK, Snyder D, Gilboa E. Dendritic cells
pulsed with RNA are potent antigen-presenting cells in vitro and
in vivo. J Exp Med 1996; 184: 465–472.
5 Carralot JP, Probst J, Hoerr I, Scheel B, Teufel R, Jung G et al.
Polarization of the immunity induced by direct injection of
naked globin UTR-stabilized mRNA vaccines. Cell Mol Life Sci
2004; 61: 2418–2424.
6 Conry RM, LoBuglio AF, Wright M, Sumerel L, Pike MJ,
Johanning F et al. Characterization of a messenger
RNA polynucleotide vaccine vector. Cancer Res 1995; 55:
1397–1400.
7 Granstein RD, Ding W, Ozawa H. Induction of anti-tumor
immunity with epidermal cells pulsed with tumor-derived RNA
Gene Therapy
Uptake of naked mRNA
J Probst et al
1180
8
9
10
11
12
or intradermal administration of RNA. J Invest Dermatol 2000;
114: 632–636.
Hoerr I, Obst R, Rammensee HG, Jung G. In vivo application of
RNA leads to induction of specific cytotoxic T lymphocytes and
antibodies. Eur J Immunol 2000; 30: 1–7.
Pascolo S. Messenger RNA-based vaccines. Expert Opin Biol Ther
2004; 4: 1285–1294.
Donnelly J, Berry K, Ulmer JB. Technical and regulatory hurdles
for DNA vaccines. Int J Parasitol 2003; 33: 457–467.
Klinman DM, Takeno M, Ichino M, Gu M, Yamshchikov G, Mor
G et al. DNA vaccines: safety and efficacy issues. Springer Semin
Immunopathol 1997; 19: 245–256.
Forg P, von Hoegen P, Dalemans W, Schirrmacher V. Superiority
of the ear pinna over muscle tissue as site for DNA vaccination.
Gene Therapy 1998; 5: 789–797.
13 Lyons JC, Ross BD, Song CW. Enhancement of hyperthermia
effect in vivo by amiloride and DIDS. Int J Radiat Oncol Biol Phys
1993; 25: 95–103.
14 Messiha FS. Neurotoxicity of chlorpromazine and modulation
by amantadine as a function of mouse strain. Neurotoxicology
1991; 12: 571–581.
15 Uma DP, Satish Rao BS, Kamath R. A method to score
micronuclei in vivo using cytochalasin B-induced cytokinesis
block. Mut Res 1998; 401: 33–37.
16 Mehta RT, Hopfer RL, McQueen T, Juliano RL, Lopez-Berestein
G. Toxicity and therapeutic effects in mice of liposomeencapsulated nystatin for systemic fungal infections. Antimicrob
Agents Chemother 1987; 31: 1901–1903.
17 Benner SA. Extracellular ‘communicator RNA’. FEBS Lett 1988;
233: 225–228.
Supplementary Information accompanies the paper on Gene Therapy website (http://www.nature.com/gt)
Gene Therapy