Abortive translation caused by peptidyl-tRNA drop-off
at NGG codons in the early coding region of mRNA
Ernesto I. Gonzalez de Valdivia and Leif A. Isaksson
Department of Genetics, Microbiology and Toxicology, Stockholm University, Sweden
As an early step during translation initiation in bac-
teria the mRNA is anchored to the 30S ribosomal sub-
unit by base pairing between a sequence close to the
end of the 16S ribosomal RNA and the Shine–Dal-
garno (SD) sequence, a few bases upstream of the initi-
ation codon in the mRNA. Even though the SD
sequence increases initiation efficiency, mRNAs that
lack this sequence can be translated [1] albeit at a
lower efficiency [2]. In such case the sequence in the
downstream region (DR) immediately following the
initiation codon has a large influence on gene expres-
sion at the translation level [3]. This effect was origin-
ally suggested to be the result of an additional
anchoring by base pairing between the DR sequence in
mRNA and 16S rRNA but this model has been refu-
ted experimentally [4,5]. Many rare codons are used
within the first 25 codons in Escherichia coli [6].
Clearly, the early coding region in mRNA is important
for gene expression, presumably at the translational
level.
A closer study of the DR region has revealed that
the nature of the +2 codon can affect expression by
up to a factor of 20. A number of codons, but in par-
ticular G-rich codons, at this position give low gene
expression. The codons that follow in the DR have rel-
atively similar effects giving normal gene expression
with the notable exception of NGG codons (AGG,

same codons, if placed at position +7, did not give this effect. Other
codons, such as CGU and AGA, at location +2 to +5, did not give any
growth inhibition of either the wild-type or the mutant strain. The inhibi-
tory effect on the pth mutant strain by NGG codons at location +5 was
suppressed by overexpression of the Pth enzyme from a plasmid. However,
the overexpression of cognate tRNAs for AGG or GGG did not rescue
from the growth inhibition associated with these codons early in the
induced model gene. The data suggest that the NGG codons trigger pep-
tidyl-tRNA drop-off if located at early coding positions in mRNA, thereby
strongly reducing gene expression. This does not happen if these codons
are located further down in the mRNA at position +7, or later.
Abbreviations
SD, Shine–Dalgarno; DR, downstream region; Pth, peptidyl-tRNA hydrolase; IPTG, isopropyl thio-b-
D-galactoside; Ts, temperature sensitive.
5306 FEBS Journal 272 (2005) 5306–5316 ª 2005 FEBS
mRNA levels or secondary structure formation.
Rather, some abortive event seems likely. One possibil-
ity would be shifting of the translational reading frame
giving premature termination at some out-of-phase ter-
mination codon. However, this explanation does not
seem to be valid. Another possibility would be drop-
off of peptidyl-tRNA from the translating ribosome
during early elongation [11]. In this case the released
peptidyl-tRNA would be cleaved by the peptidyl-
tRNA hydrolase (Pth) [12], thereby initiating turnover
and re-use of the amino acid residues and the tRNA
moiety.
A mutant E. coli strain is available that has a heat
sensitive Pth. At high temperature this mutant does not
grow because the accumulated peptidyl-tRNA cannot

G-rich codons GGN and GNG (where N is non-G),
do not give such low gene expression. The low expres-
sion associated with NGG is not seen if the codon is
located further down in the mRNA coding region at
+7 [10].
One possible reason for the low gene expression
caused by the early NGG codons could be abortive
translation as a result of peptidyl-tRNA drop-off at
these codons [13]. Such released peptidyl-tRNA is nor-
mally hydrolysed by a Pth, thus allowing recycling of
both the amino acids and the tRNA. A mutant strain
MB01 is available with a pth(Ts) (Table 1). In this
mutant strain accumulation of peptidyl-tRNA, that
cannot be degraded and recycled, leads to inability to
grow at high temperature [14]. Introduction of plasmid
pVH1 with its pth
+
gene gives partial complementa-
tion of the temperature sensitivity of MB01 (Table 2).
This complementation is not complete, probably
because of the low copy number of pVH1 [20].
Table 1. Bacterial strains and plasmids.
Relevant characteristics References
Strains
MC1061 araD139, D(lacI POZYl) 74, galU,
galK, rpsL, D(ara A BC-leu)7697, hsdR,
mcrB,Sm
S
,
[44]

Arg4
), argW
+
(tRNA
Arg5
)Kan
R
[40]
pVH1 pth
+
,Kan
R
[20]
pJMM19 lysV
+
(tRNA
Lys
), Kan
R
[20]
pMO22 glyU
+
(tRNA
Gly1
), Tet
R
[39]
Table 2. Effect of temperature on growth of strain MB01 with a
heat sensitive peptidyl-tRNA hydrolase. Strains were streaked on
Luria–Bertani agar plates with IPTG (1 m

expression associated with NGG codons in the down-
stream region following the initiation codon [9,10], is
the result of peptidyl-tRNA drop-off that is excessive
enough to inhibit growth of a pth(Ts) mutant strain.
Pth enzyme rescues from growth inhibition by
+5 NGG codons
As described above, it appeared likely that the inhibi-
tory growth effect by early NGG codons in the pth(Ts)
mutant MB01 was due to accumulation of peptidyl-
tRNA, as the result of an abortive drop-off event.
MB01 did not grow at 43 °C and it showed disturbed
growth at 37.5 °C (Table 2). To confirm that the
A B
Fig. 1 Plasmid constructs are derivatives of pDA3480 [9,10]. The 5¢ end of the transcript is 5¢-AAUUGUGAGCGGAUAACAAUUUCA-
CCAGGUAAUAAAUU
AAAUAAAAUUUAAAUAUG-3¢ for the gene variants that lack a functional Shine–Dalgarno sequence (SD, underlined)
[21]. (A) LacZ reporter gene construct pCMS71 was used as a cloning vector for the insertion of different AUG downstream context
sequences using the restriction sites SwaI ⁄ SalIandSwaI ⁄ Csp45I. The lacZ gene is under control of the trc promoter [9]. (B) Protein A¢
reporter gene construct pEG998, earlier denoted pEG1000 [10], was used as cloning vector to subclone different AUG downstream
context sequences using the restriction sites SwaI ⁄ SalI and SwaI ⁄ Csp45I.
Abortive translation due to drop-off at NGG codons E. I. Gonzalez de Valdivia and L. A. Isaksson
5308 FEBS Journal 272 (2005) 5306–5316 ª 2005 FEBS
disturbed growth was the result of peptidyl-tRNA
accumulation a plasmid (pVH1) containing the pth
+
gene was introduced into the pth(Ts) strain, harbour-
ing another compatible plasmid that carries the lacZ
reporter gene with AGG or GGG at position +5.
Transformants were streaked on broth plates, with or
without 1 mm IPTG, and incubated at 30 °Cor

strain, in
the absence of IPTG induction, are presented as inserts
in Fig. 1. These values confirm the previous findings of
low gene expression being associated with early NGG
codons [10,21]. They also demonstrate the correlation
between a negative effect on gene expression and a
strong negative effect on growth of a pth
–
mutant
strain.
Overexpression of tRNA has no effect on gene
expression or growth inhibition associated with
early NGG codons
The MB01 mutant strain is supposed to be growth
inhibited at 37.5 °C because of accumulation of pep-
tidyl-tRNA, thus possibly causing decreased turnover
and starvation for the tRNA moiety [22]. One would
therefore expect that overexpression of the correspond-
ing tRNA would protect from growth inhibition at
37.5 °C by the +5 NGG codons, in a similar manner
as does introduction of a plasmid encoding the pth
+
gene. To test for this possibility a plasmid with +5
AGG in the test gene was combined in the MB01
strain with the plasmid pUBS520 that carries the
tRNA
Arg4
gene, or plasmid pArgUW with the
tRNA
Arg4

M) (+) or without (–), and incubated at 37.5 °C.
Plasmids Features
IPTG
–+
MG1655 MB01 MG1655 MB01
– – ++++ +++ ++++ +++
pUBS250 tRNA
Arg4
++++ +++ ++++ – ⁄ +
pArgUW tRNA
Arg4; 5
++++ +++ ++++ – ⁄ +
pMO22 tRNA
Gly 1
++++ +++ ++++ +++
pEG216 AGG +5 ++++ +++ ++++ – ⁄ +
pEG216 ⁄ pVH1 AGG +5 ⁄ pth
+
++++ ++++ ++++ ++++
pEG216 ⁄ pUBS250 AGG +5 ⁄ tRNA
Arg4
++++ +++ ++++ – ⁄ +
pEG216 ⁄ pArgUW AGG +5 ⁄ tRNA
Arg4; 5
++++ +++ ++++ – ⁄ +
pEG207 GGG +5 ++++ +++ ++++ – ⁄ +
pEG207 ⁄ pVH1 GGG +5 ⁄ pth
+
++++ ++++ ++++ ++++
pEG207 ⁄ pMO22 GGG +5 ⁄ tRNA

test gene is used to analyse the effects on expression
by base alterations whereas the 2A¢ is kept constant as
an internal control. The protein products can be affin-
ity purified in a single step using an IgG–Sepharose
column. Gene expression can be estimated by separ-
ation of the two A¢ proteins on gel electrophoresis.
Scanning of the protein bands give the relative gene
expression (3A¢⁄2A¢ ratio).
The question was asked if excess tRNA cognate
to the codon preceding an NGG codon has any
influence on the low gene expression associated with
such codons. The lysine codon AAA, that gives high
gene expression if located at position +2, was cho-
sen. Plasmid constructs with AAA at +2 and either
one of the NGG codons at +3 (Fig. 2B) were coex-
pressed in strain XAC together with another plasmid
(pJMM19) encoding tRNA
Lys
. The transformants
with their two compatible plasmids were grown
under appropriate double antibiotic selection pressure
and the 3A¢ and 2A¢ protein products were isolated
and quantified by SDS ⁄ PAGE. Figure 3A shows that
tRNA
Lys
overexpression, being cognate to the +2
AAA, did not significantly suppress the low expres-
sion caused by the +3 NGG codons in the 3A¢ test
gene.
The question was also asked if excess tRNA, cog-

Arg4
, AGA at positions +2, +3, +4, +5 or +7
is associated with high gene expression, thus being very
different from the related codon AGG. The other
NGG codons CGG and UGG were not analysed for
the influence of increased concentration of cognate
tRNAs.
The question remains whether the plasmid associ-
ated tRNA genes are expressed in the analysed bac-
teria. For this reason, the levels of gene expression and
tRNA concentration in a wild-type strain was com-
pared using a plasmid encoding tRNA
Arg4
(decoding
AGA ⁄ AGG) and tRNA
Arg5
(decoding AGG) and
another plasmid with the 3A¢ test gene, using double
antibiotic selection. As shown in Fig. 4, the protein
expression value for +5 AGG is  10 times lower than
that for +5 AGA, or compared to either codon at
position +7. The low expression value for +5 AGG
is not increased despite the fact that the concentration
of the cognate decoding tRNAs is significantly
increased in the strain.
In summary, the NGG codons CGG, AGG, UGG
and GGG are associated with very low gene expression
if placed at positions +2 to +5. This phenomenon is
not observed for GGN and GNG (where N is non-G)
or for the other arginine codons CGU, CGC, CGA

mRNA–ribosome complexes shows that at least 13
and possibly as many as 20 codons of mRNA are cov-
ered by a single ribosome [31,32]. A second ribosome
cannot start translation initiation as long as the first
ribosome is still translating any codon in the early
region downstream of the initiation codon. If the
NGG codons in the downstream region are translated
slowly down to, and including position +5, this would
give extended pausing and thus low gene expression.
However, pausing at position +7 should also interfere
with ribosome loading at the translational start site
even if the effect could possibly be smaller. Alternat-
ively, if the codon is rapidly translated at position +7,
a high gene expression should be the result. In this
A
B
CD
Fig. 2. Inhibition of growth of Pth(Ts)
mutant strain MB01 by peptidyl-tRNA drop-
off at 37.5 °C. Growth of MB01 with differ-
ent plasmid constructions. Open symbols;
noninduced cultures; closed green symbols,
cultures induced with IPTG (1 m
M). Growing
cultures were monitored by measurements
of D
590
as a function of time after the induc-
tion with IPTG at time zero. The codons
analysed (CGU, AGA, CGG, AGG, UGG or

[19,36].
The codon dependent growth inhibition of the pth
mutant strain MB01 described here requires that the
test gene is induced by IPTG. The resulting growth
inhibition can be suppressed by an introduced pth
+
gene on a plasmid. This finding further supports the
model that NGG at early positions gives drop-off and
accumulation of peptidyl-tRNA thereby inhibiting
growth of MB01, with its Pth deficient enzyme
(Fig. 1). As the negative effect by NGG codons is not
found for position +7, this suggests that a heptamin-
opeptidyl-tRNA is stable on the translating ribosome,
thus being prevented from drop-off, even if it decodes
an NGG codon (Fig. 1D).
Contrary to the observed suppression of AGG
dependent growth inhibition by a pth
+
gene in strain
MB01 we did not obtain such suppression by supply-
ing the AGG cognate tRNA
Arg4
or tRNA
Arg5
or both.
These results are consistent and are obtained using
two different plasmids with the tRNA
Arg4
and
tRNA

x
100) [46]. The standard error of all experiments is ± 0.15
or smaller. (A) pJMM19 (encoding tRNA
Lys
) influences on expres-
sion of the 3A¢ model gene with +3 NGG constructs. (B) Influence
on 3A¢ gene expression by pUBS250 (encoding tRNA
Arg4
), pArgUW
(encoding tRNA
Arg4
and tRNA
Arg5
) and pMO22 (encoding tRNA
Gly1
)
as combined with +2, +3, +4, +5 or +7 AGG, GGG, AGA or AAA at
each indicated position in the 3A¢ gene.
Abortive translation due to drop-off at NGG codons E. I. Gonzalez de Valdivia and L. A. Isaksson
5312 FEBS Journal 272 (2005) 5306–5316 ª 2005 FEBS
Overexpression of certain cognate tRNAs can give
an increased gene expression [20,34,35,39–41]. Here, we
have used the 3A¢ reporter gene system to analyse whe-
ther the low gene expression associated with some early
codons could be rescued by overexpression of the cog-
nate tRNA. Co-expression of 3A¢ gene constructs with
the lysine codon AAA at +2 together with a tRNA
Lys
encoding plasmid (pJMM19) in the strain, did not sup-
press the low gene expression caused by any NGG

sequence of the nascent peptide is the same in all of
these cases. This fact eliminates any amino acid contri-
bution by the amino acid residues in the nascent pep-
tide to the observed low expression values observed for
the NGG arginine codons. Similar arguments apply
to the glycin codon family with GGG giving low while
the others give high expression [10]. We are left with a
model implying that the codon ⁄ anticodon interaction
involving NGG codons is intrinsically weak in the
early coding region, thus frequently leading to an
abortive event like peptidyl-tRNA drop-off.
Experimental procedures
Chemicals and kits
Chemicals used were of the highest available purity from
Sigma-Aldrich Chemie Gmbh (Steinheim, Germany).
Restriction enzymes, T4 DNA ligase and T4 kinase were
either from New England BioLabs (Ipswich, USA) Invitro-
gen (Invitrogen AB, Sweden). Plasmids were prepared with
GFX
TM
Micro Plasmid Prep Kit and gel band extractions
were prepared with GFX
TM
PCR and Gel Band Purifica-
tion Kit (Amersham Bioscience, Bucks, UK). DNA tem-
plates were sequenced by MWG (The Genomic Company,
Germany). Total RNA extraction was performed with
Fig. 4. tRNA levels and influence of pUBS250 (encoding tRNA
Arg4
)

gene), pArgUW (with tRNA
Arg4
and tRNA
Arg5
genes) pMO22 (with tRNA
Gly1
gene) or
pJMM19 (with tRNA
Lys
gene) (Table 1) together with the
protein 3 A¢ gene system (Fig. 1) was used to analyse for
tRNA influences on gene expression. All tRNA encoding
plasmids have different origins of replication and are com-
patible with the lacZ or the 3 A¢ carrying plasmid.
Plasmid constructions and DNA sequencing
Plasmids were constructed using standard recombinant
DNA techniques [44]. Some of the constructs (Table 1)
have been reported previously [10].
b-Galactosidase assays and growth conditions
Transformants were grown overnight at 37 °C in minimal
medium [45] supplemented with all amino acids at recom-
mended concentrations [8] and 100 lgÆmL
)1
ampicillin.
These cultures were used to inoculate fresh medium at
37 °C. Exponentially growing cells (optical density at
590 nm 0.4–0.5) were harvested without IPTG induction.
b-Galactosidase activity of the lysed uninduced cells were
determined [9]. All measurements were carried out using a
Titer tech iEMS Reader MF (multiscan microplate photom-

the mid-log phase of growth (D
590
¼ 0.2–0.25). Exponenti-
ally growing cells (D
590
¼ 0.5) were cooled and harvested
by centrifugation, followed by re-suspension in 1 mL
10 · TST buffer [46]. Cells were lysed by incubation at
95 °C for 10 min, and cell debris was eliminated by centri-
fugation. Protein A¢ was purified from the supernatant frac-
tion using IgG–Sepharose (Pharmacia, Fairfield, CT, USA)
mini-columns and a vacuum mini-fold system (Promega,
Madison, WI, USA). A¢ proteins were eluted with 0.1 mL
0.5 m HAc at pH 3.2. The eluted protein was dried in a
Savant SpeedVac
Ò
plus SC110A (Telechem International,
Inc. Sunnyvale, USA). Protein samples were dissolved in
sample loading buffer, after denaturation at 95 °C for
5 min. Separation of the A¢ proteins was achieved by
SDS ⁄ PAGE (12% acrylamide) [44]. The bands were quanti-
fied by scanning using FujiFilm Image Reader 1000 V1.2
(FujiFilm Life Science, Japan). The protein ratios were
obtained by using image gauge 4.0 quantification pro-
file ⁄ MW, microsoft excel program and extrapolation of
plots. Protein A¢ values were normalized and are presented
as relative expression where 1.0 stands for 34 ± 0.015 units
(3 A¢⁄2A¢
x
100) [46]. Each value represents the mean value

Arg4
the oligo 5¢-GAACCTGCGGCC-
CACGACTTAGAA-3¢ was used for hybridization [34].
Probes were purified using MicroSpin
TM
G-25 columns
(Amersham Pharmacia Biotech). The transferred RNAs
were hybridized overnight (12–14 h, at 42 °C) to the
[
32
P]ATP[cP] deoxyoligonucleotide probe, which is comple-
mentary to the encoding tRNA sequence [44]. Filters were
Abortive translation due to drop-off at NGG codons E. I. Gonzalez de Valdivia and L. A. Isaksson
5314 FEBS Journal 272 (2005) 5306–5316 ª 2005 FEBS
exposed to a phosphorimaging screen, scanned by image
reader V1.8E software (FujiFilm FLA 3000) and saved
using the image gauge V3.45 software (FujiFilm FLA
3000). The tRNA ratios were obtained by using Image
Gauge 4.0 quantification profile ⁄ MW, microsoft excel
program and extrapolation of plots, where 1.0 stands for
tRNA
Arg4
for the wild-type strain value. Each value repre-
sents the mean of at least three independent measurements.
Acknowledgements
We thank Dr Margarete Bucheli-Witschel for her
input at the very beginning of this project. We thank
Dr Isabella Moll, Dr Takayoshi Wakagi, Dr Valerie
Heurgue-Hamard and Dr Michael O’Connor for gifts
of plasmids and helpful advice. This work was suppor-

7 Looman AC, Bodlaender J, Comstock LJ, Eaton D,
Jhurani P, deBoer HA & van Knippenberg PH (1987)
Influence of the codon following the AUG initiation
codon on the expression of a modified lacZ gene in
Escherichia coli. EMBO J 6, 2489–2492.
8 Faxe
´
n M, Plumbridge J & Isaksson LA (1991) Codon
choice and potential complementarity between mRNA
downstream of the initiation codon and bases 1471–
1480 in the 16S ribosomal RNA affects expression of
gln S. Nucleic Acids Res 19, 5247–5251.
9 Stenstro
¨
m CM, Jin H, Major LL, Tate WP & Isaksson
LA (2001a) Codon bias at the 3¢-side of the initiation
codon is correlated with translation initiation efficiency
in Escherichia coli. Gene 263, 273–284.
10 Gonzalez de Valdivia EI & Isaksson LA (2004) A codon
window in mRNA downstream of the initiation codon
where NGG codons give strongly reduced gene expres-
sion in Escherichia coli. Nucleic Acids Res 32, 5198–
5205.
11 Kurland CG, Hughes D & Ehrenberg M (1996) Limita-
tions of translational accuracy. In Escherichia coli and
Salmonella thyphimurium: Cellular and Molecular Bio-
logy (Neidhardt FC, ed.), pp. 979–999. ASM Press,
Washington D.C.
12 Atherly AG & Menninger JR (1972) Mutant E. coli
strain with temperature sensitive peptidyl-transfer RNA

esis by accumulating peptidyl-tRNAArg4. Mol Micro-
biol 49, 1043–1049.
20 Menez J, Heurgue-Hamard V & Buckingham RH
(2000) Sequestration of specific tRNA species cognate
to the last sense codon of an overproduced gratuitous
protein. Nucleic Acids Res 28, 4725–4732.
E. I. Gonzalez de Valdivia and L. A. Isaksson Abortive translation due to drop-off at NGG codons
FEBS Journal 272 (2005) 5306–5316 ª 2005 FEBS 5315
21 Croitoru VV, Bucheli-Witschel M & Isaksson LA (2005)
In vivo involvement of mutated initiation factor IF1 in
gene expression control at the translational level. FEBS
Lett 579, 995–1000.
22 Menninger JR (1978) The accumulation as peptidyl-
transfer RNA of isoaccepting transfer RNA families in
Escherichia coli with temperature-sensitive peptidyl-
transfer RNA hydrolase. J Biol Chem 253, 6808–6813.
23 Stenstro
¨
m CM & Isaksson LA (2002) Influences on
translation initiation and early elongation by the mes-
senger RNA region flanking the initiation codon at the
3¢ side. Gene 288, 1–8.
24 Spanjaard RA & Duin JV (1988) Translation of the
sequence AGG-AGG yields 50% ribosomal frameshift.
Proc Natl Acad Sci USA 85, 7967–7971.
25 Zahn K & Landy A (1996) Modulation of lambda inte-
grase synthesis by rare arginine tRNA. Mol Microbiol
21, 69–76.
26 Varenne S, Baty D, Verheij H, Shire D & Lazdunski C
(1989) The maximum rate of gene expression is depen-

J Mol Biol 291, 745–759.
34 Brinkmann U, Mattes RE & Buckel P (1989) High-level
expression of recombinant genes in Escherichia coli is
dependent on the availability of the dnaY gene product.
Gene 85, 109–114.
35 Zahn K (1996) Overexpression of an mRNA dependent
on rare codons inhibits protein synthesis and cell
growth. J Bacteriol 178, 2926–2933.
36 Cruz-Vera LR, Magos-Castro MA, Zamora-Romo E &
Guarneros G (2004) Ribosome stalling and peptidyl-
tRNA drop-off during translational delay at AGA
codons. Nucleic Acids Res 32, 4462–4468.
37 Bjo
¨
rk GR (1996) Escherichia coli and Salmonella
typhimurium. In Escherichia coli and Salmonella
thyphimurium: Cellular and Molecular Biology
(Neidhardt FC, ed.), pp. 861–880. ASM Press,
Washington D.C.
38 Gustafsson C, Govindarajan S & Minshull J (2004)
Codon bias and heterologous protein expression. Trends
Biotechnol 22, 346–353.
39 O’Connor M (1998) tRNA imbalance promotes -1
frameshifting via near-cognate decoding. J Mol Biol
279, 727–736.
40 Imamura H, Jeon B, Wakagi T & Matsuzawa H (1999)
High level expression of Thermococcus litoralis 4-alpha-
glucanotransferase in a soluble form in Escherichia coli
with a novel expression system involving minor arginine
tRNAs and GroELS. FEBS Lett 457, 393–396.