NCCR RNA & DiseAse CHIMIA 2019, 73, No. 5 391
doi:10.2533/chimia.2019.391 Chimia 73 (2019) 391–394 © Swiss Chemical Society
*Correspondence: PD Dr. S. Pascoloab
aUniversity Hospital of Zürich; Department of Dermatology, Gloriastrasse 31, CH-8091 Zürich; bFaculty of Medicine, University of Zurich, Zurich; cCenter for Integrative
Genomics, University of Lausanne, Genopode, CH-1015 Lausanne; dDepartment of Dermatology and Allergy, University Hospital, LMU Munich, Munich, Germany
Design of in vitro Transcribed mRNA
Vectors for Research and Therapy
Marina Tusupab, Lars E. Frenchabd, Mara De Matosc, David Gatfieldc, Thomas Kundigab,
and Steve Pascolo*ab
Abstract: The use of in vitro transcribed messenger RNA (ivt mRNA) for vaccination, gene therapy and cell repro-
graming has become increasingly popular in research and medicine. This method can be used in vitro (transfect-
ed in cells) or administered naked or formulated (lipoplexes, polyplexes, and lipopolyplexes that deliver the RNA
to specific organs, such as immune structures, the lung or liver) and is designed to be an immunostimulatory or
immunosilent agent. This vector contains several functional regions (Cap, 5' untranslated region, open reading
frame, 3' untranslated region and poly-A tail) that can all be optimised to generate a highly efficacious ivt mRNA.
In this study, we review these aspects and report on the effect of the ivt mRNA purification method on the func-
tionality of this synthetic transient genetic vector.
Keywords: Cap · Gene therapy · Globin UTR · in vitro transcribed mRNA · ivt RNA · Vaccination
Steve Pascolo studied biology/biochemistry at Ecole Normale
Supérieure (Paris, France) and obtained his PhD in 1998 from
the Pasteur Institute / University Paris 6, where he created an im-
munologically humanised mouse model called HHD (mice defi-
cient in the endogenous H-2 class I molecules and transgenic for
a human HLA class I molecule). As a post-doctoral fellow at the
University of Tübingen, Germany, he focused on immunisation
methods using WT and HHD mice, testing different vaccination
formats (e.g. peptides, plasmids, and ivt mRNAs), and discov-
ered that RNA is a ‘danger signal’ that primes innate and adaptive
immune responses. With the goal of translating this vaccination
method to humans, Steve Pascolo co-founded CureVac GmbH in
2000, implemented the first large scale and GMP production of
ivt mRNAs worldwide and initiated the first clinical studies in
which patients with cancer received direct ivt mRNA injections.
In 2006, Steve Pascolo moved to the University Hospital of Zurich
(Switzerland) and continued to develop innovative immunochem-
otherapies against cancer (vaccination, immunomodulation, and
chemotherapies). In 2017, the University of Zurich supported the
creation of an mRNA platform. This platform allows researchers
and clinicians to have access to high quality ivt mRNAs produced
on demand for their research and pre-clinical projects.
Vaccination (induction of an immune response against a pro-
tein) and gene therapy (expression of a therapeutic protein) can be
performed using genetic vectors, such as plasmid DNA, in vitro
transcribed mRNA (ivt mRNA) or recombinant microorganisms
(modified viruses or bacteria). Because of this technology’s su-
perlative features (safety, efficacy, steerable immunostimulatory
properties, and ease of production and preservation), ivt mRNA
is emerging as the optimal vector format.[1,2] We have initiated
the clinical use of directly injected ivt mRNAs through academic
(University of Tübingen, Germany) and industrial (CureVac AG,
Germany) co-developments. As a key milestone, we implemented
the first large scale GMP (good manufacturing practice) produc-
tion of an ivt mRNA (CureVac AG, Germany) worldwide and
led the first clinical studies in which patients with cancer were
injected with ivt mRNAs.[3–6] This pioneering work has revealed
for the first time that an ivt mRNA can be produced at a large scale
in injectable quality, and that its direct administration in humans is
safe and leads to the expected pharmacodynamics (expression of
the encoded protein and development of an immune response).
The most recently developed ivt mRNA-based approaches were
reported to induce very strong immune responses in patients with
cancer.[8,9] Key to this notably high efficacy are the formulation
of the ivt mRNA, the site of injection, as well as the structure of
ivt mRNA. The present review will focus on this latter aspect.
2. Structure of the ivt mRNA
2.1 The 5' Cap
Eukaryotic mRNAs contain a 5' Cap called Cap 0 that con-
sists of a methyl-7 guanine nucleotide connected to the RNA via
a 5' to 5' triphosphate linkage. If the first base of the mRNA is
methylated at the 2' position, the Cap is called Cap 1. It is the
optimal structure to stabilise the mRNA and recruit the eukaryotic
initiation factor 4E (eIF4E) that binds to eIF4G and facilitates the
subsequent assembly of the eIF4F complex. Alternatives to the
5' Cap for the recruitment of the translation machinery are Cap-
independent viral sequences, such as the 5' untranslated region of
the tobacco etch virus, or internal ribosomal entry site (IRES)
sequences. However, such sequences at the 5' end of an ivt mR-
NA are not sufficient to achieve high levels of protein expression
after transfection of the RNA in cells. Thus, the production of a
5' capped ivt mRNA is recommended. Two methods for capping
have been reported: (i) co-transcriptional capping, where a Cap
analogue (or an anti-reverse Cap analogue ‘ARCA’ eventually
optimised or a Cap dinucleotide ‘CleanCap®’ from Trilink) is
present in the transcription reaction, and (ii) post-transcriptional
capping, where the transcribed mRNA is capped using enzymes
such as the vaccinia virus capping enzyme. In all cases, a Cap
392 CHIMIA 2019, 73, No. 5 NCCR RNA & DiseAse
able for codon optimisation that incorporate several features (e.g.
predicted secondary RNA structure and avoiding consecutive rep-
etition of identical codons). Thus, we advise codon optimisation
for the production of mRNAs encoding non-mammalian proteins
(optionally using two independent companies/software packag-
es and comparing the functionality of the resulting ivt mRNAs),
while, a priori, the wild-type coding sequence should be used to
produce mRNAs coding mammalian proteins (a subsequent test
of codon-optimised forms can be envisioned).
2.4 The 3'UTR and Poly-A Tail
In ivt mRNAs, the most frequently used 3'UTR is the globin
3'UTR.[14,17–20] An optimised version of this 3'UTR is a tandem
repeat of the sequence, which contains then approximately 300
bases. Because the presence of repeated sequences may hamper
the synthesis of the synthetic DNA that is used as a matrix for tran-
scription, Orlandini von Niessen et al. recently reported the iden-
tification of new stabilising 3'UTRs. The researchers started
from a library of long-lived mRNAs (derived from dendritic cells
treated with actinomycin D for a few hours to block transcription)
and cloned fragments of those reverse transcribed molecules at
the 3' end of an ORF encoding destabilised eGFP. This library
was then transcribed into mRNAs, transfected into dendritic cells,
RNA was extracted at different time points, and reverse transcrip-
tion followed by PCR amplification was performed. Using an
iterative process of transfection/RNA extraction/amplification/
transcription, optimal 3'UTR sequences were identified. The best
sequences corresponded to UTRs from mRNAs with long half-
lives (this result validated the approach). The combination of two
of those sequences in the 3'UTR, one from mitochondrially en-
coded 12S rRNA ‘mtRNR1’ and one from the amino-terminal
enhancer of split mRNA ‘AES’, produced an ivt mRNA with the
highest expression (the combined sequences produced a UTR of
less than 300 bases). This mRNA’s expression in vitro (several cell
types) and in vivo (mice) was greater than two-fold higher than
the expression of the usual ivt mRNA with a tandem repeat of the
globin 3'UTR. The poly-A tail can be encoded by the DNA matrix
directly such that the ivt mRNA contains a 3' stretch of at least
90 A bases. Importantly, extra C, G or U bases at the 3' end of the
poly-A sequence should be avoided. If the DNA template does
not encode this 3' poly-A sequence, an incubation of the ivt mR-
NA in a reaction with a poly-A polymerase post-transcriptionally
generates the poly-A tail.
2.5 Purification Methods
The RNA polymerases that are usually used to produce ivt
mRNAs (T3, T7, and SP6) are very processive and specific for their
promoters. However, ivt mRNAs visualised on a gel or with an
Agilent analysis may show some aberrant shorter or longer tran-
scripts. Those transcripts might be derived from cryptic promoters
or cryptic terminators of transcription (e.g. highly structured RNA
regions). Therefore, we have implemented a dedicated HPLC-
based purification method for large-scale GMP production. This
method allows identifying and isolating the mRNA with the cor-
rect size, hence removing potential by-products that are longer or
shorter. In addition, the RNA polymerase may generate hairpins
at the 3' end by synthesizing a complementary RNA sequence. A
newly available T7 RNA polymerase that is optimised to generate
RNA at 50 °C (Hi-T7™ RNA Polymerase from NEB Biolabs)
was reported to reduce dsRNA by-product formation. We also
noticed that this Hi-T7™ can produce full-length mRNA from
hard-to-transcribe DNA templates (e.g. containing a GC rich 5'
region, Fig. 2). Meanwhile, for research-grade ivt mRNAs, usual-
ly the classical production using wild type polymerases and sim-
ple purification of ivt mRNAs using selective precipitation (LiCl
that precipitates only long RNA molecules, not oligonucleotides,
DNA or proteins) or columns (silica membranes) is sufficient. A
0 or Cap 1 can be generated (the latter requires the utilisation
of a Cap 2'-O-methyltransferase in addition to the vaccinia virus
capping enzyme in the context of enzymatic capping). Based on
our experience, the most functional ivt mRNA was produced by
co-transcriptional capping using the CleanCap® Reagent AG (3'
OMe) from Trilink (m7(3'OMeG)(5')ppp(5')(2'OMeA)pG). It has
the dual advantages of being an anti-reverse Cap 1 and of not
competing with GTP to initiate transcription. Thus, the GTP con-
centration in the transcription reaction can be maintained at high
levels without reducing the percentage of capped mRNA (using
usual Cap mononucleotide analogues, the GTP concentration
must be reduced to favour initiation by Cap and not by G). The
method’s disadvantage is that the DNA template must contain a
modified T7 promoter, where the C that is used to base pair with
the first nucleotide of the ivt mRNA (usually a G) is a T (allowing
incorporation of the A of the CleanCap® Reagent AG). Thus, new
dedicated DNA constructs must be produced to use the CleanCap®
2.2 The 5'UTR
Located between the 5' Cap and the first ATG codon (Fig.
1), the 5'UTR sequence must provide a good Kozak consensus
sequence that favours initiation of the translation process at the
expected ATG. An 80 base sequence from the globin 5'UTR is
frequently used. We recently reported the superior efficacy of
a short sequence of 40 nucleotides corresponding to an aptamer
that binds eIF4G. The rationale was that in some cell types, the
availability of eIF4E is a limiting factor for translation; therefore,
the direct recruitment of eIF4G to the 5'UTR would likely pro-
mote translation. We now typically produce ivt mRNAs with this
5' aptamer sequence, as follows: CapAG - G A C T C A C T A T T
T G T T T T C G C G C C C A G T T G C A A A A A G TG TC G
C C A C C A T G (italicized bases represent the eIF4G aptamer
and underlined bases represent the start codon).
2.3 Coding Sequence
Depending on the species, codon usage can vary and reflects
the relative abundance of tRNAs; efficient translation relies on
optimal codons that correspond to abundant tRNA species.
Rare codons in an open reading frame hinder the production of
full-length proteins. These codons may also decrease the stability
of the mRNA, as slow-moving ribosomes may trigger mRNA de-
cay. Therefore, ivt mRNAs are typically produced from synthetic
DNA templates that have been ‘codon optimised’. However, the
natural sequence of a gene may have evolved to favour proper
protein production, particularly folding, by controlling the speed
of translation; rare codons are used to slow translation between
structural domains allowing time for each domain to fold proper-
ly. Therefore, the replacement of suboptimal codons with optimal
codons may increase protein misfolding. Indeed, we observed
that some codon-optimised ivt mRNAs encoding secreted pro-
teins were less functional than ivt mRNAs with the WT sequences
(data not shown). Notably, multiple software packages are avail-
Fig. 1. Schematic representation of optimised ivt mRNA. A 5' Cap (op-
timally a methyl-7 guanine nucleotide connected via a 5' to 5' triphos-
phate linkage to a 2'-O-methyl modied residue: Cap1) is followed by a
5'UTR (optimally an aptamer that binds eIF4G: ‘Apt’) and an open read-
ing frame (optimally a codon optimised sequence depicted in yellow) be-
tween a start (AUG) and a stop codon. The 3' end of the mRNA can con-
tain a stabilising 3'UTR (for example a tandem repeat of the beta-globin
3'UTR ‘β’) and a poly-A tail (ideally of more than 90 residues).
NCCR RNA & DiseAse CHIMIA 2019, 73, No. 5 393
genicity (for a completely 1-methyl-pseudoUridine -substituted
mRNA, Fig. 3B). Thus, LiCl does not exert any deleterious effects
on ivt mRNAs that impair their subsequent applications, and it is
a simple, inexpensive and scalable method for purifying mRNAs
from in vitro transcription reactions.
The production of ivt mRNAs for research and clinical in-
terventions is easy and inexpensive. However, because this tech-
nology has many facets, including the structure of the ivt mRNA
and its formulation, substantial expertise is helpful for producing
optimal ivt mRNAs. To this end, the implementation of a central-
ised ivt mRNA platform, which we have used at the University of
html), represents a significant asset for researchers and clinicians
aiming to use ivt mRNAs for research or clinical interventions.
Investigator-driven clinical trials have been initiated based on
findings from academic studies, in particular for vaccinations
that we have pioneered[3–5,19,20,23,24] and for modified T-cells.[25,26]
Meanwhile, several dedicated biotechnology companies have
initiated human studies, including BioNTech AG, CureVac AG,
eTheRNA, Moderna Inc. Ongoing company-sponsored clinical
trials include vaccinations (e.g. vaccines against solid cancers and
viruses, such as rabies, HPV16 and HIV), and non-immunogenic
protein expression (e.g. ivt mRNAs encoding cytokines that are
injected into the tumour). Furthermore, the versatility of the ivt
mRNA format allows the design of individualized therapeutics;
mRNA vaccines encoding mutated peptides corresponding to pa-
tient-specific cancer mutations have been designed, produced and
administered. In addition, ivt mRNAs can be combined with
comparison of both purification methods (Fig. 3) revealed that
they are equivalent and present similar efficacy in terms of the
amount of recovered RNA, OD260/OD280 ratio (ca. 2), of the
observed transfection efficiencies (Fig. 3A) and reduced immuno-
Fig. 3. Effects of the purication method on the functionality of ivt mRNAs. A completely 1-methyl-pseudoUridine-substituted synthetic mRNA
encoding luciferase was puried either with LiCl precipitation (2.5 M LiCl nal concentration at –20 °C for one hour, followed by centrifugation and
washing of the pellet with 75% ethanol), ‘Pseudo LiCl’ or MinElute columns (‘Pseudo MinElute’) from Qiagen. A, The ivt mRNA was formulated with
MessengerMax (ThermoFisher) according to the manufacturer’s recommendations and added to 100 µL of cells (in each well: 200 ng of mRNA per
1 million fresh human peripheral blood mononuclear cells (PBMCs'') or 10 ng of mRNA per 100,000 human HEK cells or mouse CT26 tumour cells)
in RPMI containing 10% foetal bovine serum and antibiotics. Cells that were not treated with the RNA were used as a negative control (‘No RNA’).
Twenty hours later, the luciferase activity in each well was recorded following the addition of 25 µL of BrightGlo (Promega) and measurement on a
GloMax instrument (Promega). The results in A represent the average luciferase activity ± the standard deviation (from triplicates). They show that
the two purication methods provide mRNAs with comparable expression capacities. B, The ivt mRNA was formulated with protamine to gener-
ate 130 nm nanoparticles[29,30] and 10 µg of the preparation (5 µg mRNA plus 5 µg Protamine) were added to 200 µL of cells (1 million fresh human
peripheral blood mononuclear cells) in RPMI containing 10% foetal bovine serum and antibiotics. Twenty hours later, the interferon-alpha and TNF-
alpha levels in the supernatants were measured using ELISAs (from Mabtech and Bioscience, respectively). A luciferase mRNA generated with the
four canonical bases ACGU and precipitated with LiCl (‘U LiCl’) was used as a positive control. As negative controls, cells without added RNA (‘No
RNA’) or treated with only 5 µg Protamine (‘Protamine’) were used. The results in B represent the average cytokine concentrations ± standard devia-
tions (from triplicates). They show that the two purications method give 1-methyl-pseudoUridine substituted RNA with very reduced immunogenic
capacities even at the high concentration used here in the cell culture (25 µg/mL nal).
Fig. 2. Transcription of difcult DNA matrices. Transcriptions were
performed using three PCR matrices with a GC content of 78% in the
rst 161 bases. The produced mRNAs were analysed on a denaturing
Formaldehyde MOPS agarose gel. The expected size was over ca.
2500 bases. The canonical T7 RNA polymerase (‘T7’) at 37 °C produces
abundant shorter transcripts while the Hi-T7™ RNA Polymerase (‘Hi-T7’)
at 50 °C generates mostly full-length transcripts.
394 CHIMIA 2019, 73, No. 5 NCCR RNA & DiseAse
 A. G. Bert, R. Grepin, M. A. Vadas, G. J. Goodall, RNA 2006, 12, 1074,
 A. N. Kuhn, M. Diken, S. Kreiter, A. Selmi, J. Kowaska, J. Jemielty, E.
Darzynkiewicz, C. Huber, Ö. Türeci, U. Sahin, Gene Ther. 2010, 17, 961,
 J.-P. Carralot, J. Probst, I. Hoerr, B. Scheel, R. Teufel, G. Jung, H.-G.
Rammensee, S. Pascolo, Cell Mol. Life Sci. 2004, 61, 2418.
 M. K. Tusup, T. Kundig, S. Pascolo, EPH Int. J. Med. Health Sci. 2018, 4,
 G. Hanson, J. Coller, Nat. Rev. Mol. Cell Biol. 2018, 19, 20, doi:10.1038/
 J.-P. Carralot, B. Weide, O. Schoor, J. Probst, B. Scheel, R. Teufel, I. Hoerr,
C. Garbe, H.-G. Rammensee, S. Pacolo, Genet. Vaccines Ther. 2005, 3, 6.
 I. Hoerr, R. Obst, H.-G. Rammensee, G. Jung, Eur. J. Immunol. 2000, 30, 1.
 B. Weide, J.-P. Carralot, A. Reese, B. Scheel, T. K. Eigentler, I. Hoerr, H.-
G. Rammensee, C. Garbe, S. Pascolo, J. Immunother. 2008, 31, 180, doi:
 B. Weide, S. Pascolo, B. Scheel, E. Derhovanessian, A. Pugfelder, T. K.
Eigentler, G. Pawelec, I. Hoerr, H.-G. Rammensee, C. Garbe, J. Immunother.
2009, 32, 498, doi: 10.1097/CJI.0b013e3181a00068.
 S. Holtkamp, S. Kreiter, A. Selmi, P. Simon, M. Koslowski, C. Huber, Ö.
Türeci, U. Sahin, Blood 2006, 108, 4009, doi:10.1182/blood-2006-04-015024.
 A. G. Orlandini von Niessen, M. A. Poleganov, C. Rechner, A. Plaschke, L.
M. Kranz, S. Fresser, M. Diken, M. Löwer, B. Vallazza, T. Beissert, V. Bukur,
A. N. Kuhn, Ö. Türeci, U. Sahin, Mol. Ther. 2019, 27, 824, doi:10.1016/j.
 S. M. Rittig, M. Haentschel, K. J. Weimer, A. Heine, M. R. Muller,
W. Brugger, M. S. Horger, O. Maksimovic, A. Stenzl, I. Hoerr, H.-G.
Rammensee, T. A. W. Holderried, L. Kanz, S. Pascolo, P. Brossart, Mol.
Ther. 2011, 19, 990, doi:10.1038/mt.2010.289.
 S. M. Rittig, M. Haentschel, K. J. Weimer, A. Heine, M. R. Müller, W. Brugger,
M. S. Horger, O. Maksimovic, A. Stenzl, I. Hoerr, H.-G. Rammensee, T. A.
W. Holderried, L. Kanz, S. Pascolo, P. Brossart, Oncoimmunology 2016, 5,
 G. L. Beatty, A. R. Haas, M. V. Maus, D. A. Torigian, M. C. Soulen, G. Plesa,
A. Chew, Y. Zhao, B. L. Levine, S. M. Albelda, M. Kalos, C. H. June, Cancer
Immunol. Res. 2014, 2, 112, doi:10.1158/2326-6066.CIR-13-0170.
 J. Svoboda, S. R. Rheingold, S. I. Gill, S. A. Grupp, S. F. Lacey, I.
Kulikovskaya, M. M. Suhoski, J. J. Melenhorst, B. Loudon, A. R. Mato,
S. D. Nasta, D. J. Landsburg, M. R. Youngman, B. L. Levine, D. L.
Porter, C. H. June, S. J. Schuster, Blood 2018, 132, 1022, doi:10.1182/
 T. Mutzke, G. Schubkegel, R. Teufel, T. Ketter, J. Probst, B. Scheel, J.-P-
Carralot, S. Pascolo, K. Ghoreschi, C. Weigert, Nucleosides Nucleotides
Nucleic Acids 2005, 24, 147.
 M. Tusup, L. E. French, E. Guenova, T. Kundig, S. Pascolo, J. Pharmacol.
Clin. Toxicol. 2018, 6, 1117.
 L. Rettig, S. P. Haen, A. G. Bittermann, L. von Boehmer, A. Curioni, S.
D. Krämer, A. Knuth, S. Pacolo, Blood 2010, 115, 4533, doi:10.1182/
 M. Tusup, S. Pascolo, Meth. Mol. Biol. 2017, 1499, 155, doi:10.1007/978-1-
siRNAs or aptamers to “bring a protein” and at the same time
“block a protein or pathway”. Meanwhile, drugs have been
used to enhance the efficacy of ivt mRNA-based therapies.
Thus, ivt mRNAs are a highly versatile tool for research and clin-
ical studies and are predicted to be the format used in many ther-
apies in the future.
Conict of Interest
The authors have no conicts of interest to disclose.
This work was supported by the University Research Priority
Project (URPP) ‘Translational Cancer Research’ and by the Monique
Dormonville de la Cour Stiftung.
Received: March 1, 2019
 S. Pascolo, J. Multidiscipl. Engin. Sci. Technol. 2017, 4, 6937.
 U. Sahin, K. Kariko, O. Tureci, Nat. Rev. 2014, 13, 759, doi:10.1038/
 S. Pascolo, Expert Opin. Biol. Ther. 2004, 4, 1285,
 S. Pascolo, Methods Mol. Med. 2006, 127, 23, doi:1-59745-168-1:23
 S. Pascolo, Handb. Exp. Pharmacol. 2008, 221, doi:10.1007/978-3-540-
 J. Probst, B. Weide, B. Scheel, B. J. Pichler, I. Hoerr, H. G. Rammensee, S.
Pascolo, Gene Ther. 2007, 14, 1175.
 S. Pascolo, Expert Rev. Vaccines 2015, 14, 153, doi:10.1586/14760584.2015
 L. M. Kranz, M. Diken, H. Haas, S. Kreiter, C. Loquai, K. C. Reuter, M.
Meng, D. Fritz, F. Vascotto, H. Hefesha, C. Grunwitz, M. Vormehr, Y.
Hüsemann, A. Selmi, A. N. Kuhn, J. Buck, E. Derhovanessian, R. Rae, S.
Attig, J. Diekmann, R. A. Jabulowsky, S. Heesch, J. Hassel, P. Langguth, S.
Grabbe, C. Huber, Ö. Tüeci, U. Sahin, Nature 2016, 534, 396, doi:10.1038/
 U. Sahin, E. Derhovavessian, M. Miller, B.-P. Kloke, P. Simon, M. Löwer,
V. Bukur, A. D. Tadmor, U. Luxemburger, B. Schrörs, T. Omokoko, M.
Vormehr, C. Albrecht, A. Paruzynski, A. N. Kuhn, J. Buck, S. Heesch, K.
H. Schreeb, F. Müller, I. Ortseifer, I. Vogler, E. Godehardt, S. Attig, R. Rae,
A. Breitkreuz, C. Tolliver, M. Suchan, G. Martic, A. Hohberger, P. Sorn,
J. Diekmann, J. Ciesla, O. Waksann, A.-K. Brück, M. Witt, M. Zillgen, A.
Rothermel, B. Kasemann, D. Langer, S. Bolte, M. Diken, S. Kreiter, R.
Nemecek, C. Gebhardt, S. Grabbe, C. Höller, J. Utikal, C. Huber, C. Loquai,
Ö. Türeci, Nature 2017, 547, 222, doi:10.1038/nature23003.
 M. Tusup, L. E. French, E. Guenova, T. M. Kundig, S. Pascolo, Biomed. J.
Sci. Tech. Res. 2018, doi: 10.26717/BJSTR.2018.07.001487
 D. R. Gallie, J. Virol. 2001, 75, 12141, doi:10.1128/JVI.75.24.12141-