© 2000 Oxford University Press

Human Molecular Genetics, 2000, Vol. 9, No. 4

467–475

ARTICLE

Decreased aminoacylation of mutant tRNAs in MELAS
but not in MERRF patients
G. Valentin Börner1,2,+, Massimo Zeviani3, Valeria Tiranti3, Franco Carrara3, Sabine Hoffmann4,
Klaus Dieter Gerbitz4, Hanns Lochmüller5, Dieter Pongratz5, Thomas Klopstock6, Atle Melberg7,
Elisabeth Holme8 and Svante Pääbo1
1Max-Planck-Institute for Evolutionary Anthropology, Inselstrasse 22, D-04103 Leipzig, Germany, 2Department of

Molecular and Cellular Biology, Harvard University, 7 Divinity Avenue, Cambridge, MA 02138, USA, 3Istituto
Nazionale Neurologico ‘Carlo Besta’, Divisione di Biochemica e Genetica, Via Celoria 11, I-20133 Milano, Italy,
4Institutes for Clinical Chemistry and Diabetes Research, Academic Hospital Schwabing, D-80804 Munich, Germany,
5Friedrich Baur Institut, University of Munich, Ziemsenstrasse 1a, D-80336 Munich, Germany, 6Department of
Neurology, Klinikum Großhadern, University of Munich, D-82366 Munich, Germany, 7Department of Neuroscience,
Neurology, University Hospital, S-751 85 Uppsala, Sweden and 8Department of Clinical Chemistry, Sahlgrenska
University Hospital, S-413 45 Göteborg, Sweden
Received 11 October 1999; Revised and Accepted 4 January 2000

Mutations in human mitochondrial tRNA genes are associated with a number of multisystemic disorders. Using an
assay that combines tRNA oxidation and circularization we have determined the relative amounts and states of
aminoacylation of mutant and wild-type tRNAs in tissue samples from patients with MELAS syndrome (mitochondrial myopathy, encephalopathy, lactic acidosis, stroke-like episodes) and MERRF syndrome (myoclonus
epilepsy with ragged red fibers), respectively. In most, but not all, biopsies from MELAS patients carrying the
A3243G substitution in the mitochondrial tRNALeu(UUR) gene, the mutant tRNA is under-represented among
processed and/or aminoacylated tRNAs. In contrast, in biopsies from MERRF patients harboring the A8344G
substitution in the tRNALys gene neither the relative abundance nor the aminoacylation of the mutated tRNA is
affected. Thus, whereas the A3243G mutation may contribute to the pathogenesis of MELAS by reducing the
amount of aminoacylated tRNALeu, the A8344G mutation does not affect tRNALys function in the same way.

INTRODUCTION
Human cells contain hundreds of mitochondrial DNA (mtDNA)
molecules, each of which encodes 13 polypeptides essential for
cell respiration, as well as two ribosomal RNAs (rRNAs) and 22
transfer RNAs (tRNAs), which constitute a complete set of RNAs
necessary for mitochondrial translation (for a review see ref. 1).
To date, >30 mutations in mitochondrial tRNA genes have been
reported to be associated with chronic degenerative diseases
affecting brain, heart, muscle, kidney and endocrine glands (2).
The most studied of these mutations are an A→G transition at
position 8344 in the tRNALys gene, causing the MERRF syndrome
(myoclonus epilepsy with ragged red fibers) (3), and an A→G
transition at position 3243 in the tRNALeu(UUR) gene, which is
frequently associated with the MELAS syndrome (mitochondrial
myopathy, encephalopathy, lactic acidosis, stroke-like episodes)
(4,5), but also with other clinical manifestations such as chronic
progressive external ophthalmoplegia and certain forms of
diabetes mellitus (4–7).

Although characteristic defects in mitochondrial protein
synthesis have been observed in cell lines carrying the MELAS or
the MERRF mutations (8–14), it is still poorly understood how
these mutations interfere with tRNA function. We have taken
advantage of the fact that most disease-associated mutations in
mitochondrial tRNA genes are heteroplasmic in patient tissues,
i.e. they occur as a mixture with wild-type mtDNA (15). Any
effect of a mutation on tRNA processing, tRNA stability or
aminoacylation should therefore be observable as a shift in the
relative abundance of the mutant when mtDNA, total tRNAs and
aminoacylated tRNAs are compared. Using an assay that
combines tRNA oxidation and circularization we show that the
tRNALeu(UUR) carrying the A3243G mutation is under-represented
relative to the tRNA with the wild-type sequence among
aminoacylated tRNAs in patient tissue, indicating a specific defect
of the mutant molecule in stability and/or aminoacylation. In
contrast, no such decrease is observed for the tRNALys carrying
the A8344G mutation. Our results suggest that reduced levels of

+To whom correspondence should be addressed. Tel: +1 617 495 4396; Fax: +1 617 495 0758; Email: boerner@fas.harvard.edu

468

Human Molecular Genetics, 2000, Vol. 9, No. 4

Figure 1. Schematic representation of the OXOCIRC assay. Filled circles,
aminoacylated 3′ ends; double bar, oxidized 3′ ends; dashed arrows, circularization;
black arrowheads, primers for reverse transcription and PCR; shaded arrowheads,
primers for PCR. Letters in parentheses refer to the control experiment in Figure 2d
where a mixture of aminoacylated tRNAAsp with anticodon GCC and nonaminoacylated tRNAAsp with anticodon GUC were analyzed with the OXOCIRC
assay.

functional tRNALeu(UUR) contribute to the pathogenesis of the
MELAS syndrome.
RESULTS
The OXOCIRC assay
To determine the relative abundance of the mutant sequence in
mtDNA, total cellular tRNAs and aminoacylated tRNAs in
heteroplasmic patient tissue, we applied the OXOCIRC assay (16)
(Fig. 1). In this assay, total small cellular RNAs are isolated under
conditions that minimize deacylation (i.e. pH 5 and low
temperature) and incubated with NaIO4 which specifically
oxidizes the cis-diol present in the 3′ terminal ribose of nonaminoacylated RNAs while leaving aminoacylated tRNAs
unaffected. The tRNAs are subsequently deacylated and
incubated with RNA ligase, allowing the formerly aminoacylated
molecules with intact 3′-riboses to undergo intramolecular
ligation to the 5′-phosphate, thereby generating circular tRNA
molecules (16,17). In contrast, formerly non-aminoacylated
tRNAs with oxidized 3′ ends remain in their linear form (16).
tRNAs are subsequently reverse transcribed and amplified by
PCR using primers located such that only circularized (formerly
aminoacylated) tRNAs yield amplification products. To generate
an RT–PCR product representing all tRNAs irrespective of their
aminoacylation state, an aliquot of the RNA preparation is
deacylated and circularized without oxidation. Finally, (RT–)PCR
products generated from mtDNA, total circularized tRNAs and
oxidized circularized tRNAs are analyzed for the relative
abundance of mutant and wild-type sequence by a primer
extension assay in the presence of an informative ddNTP (Figs 2b
and 3b).

Figure 2. OXOCIRC assay on tRNALeu(UUR) from MELAS patients. (a) Detection
of circular tRNALeu by RT–PCR. The black arrowhead indicates the primer
(#1620) used for reverse transcription whereas shaded arrowheads indicate the
primers (#1904 and #1905) used for PCR together with primer #1620.
(b) Detection of the A3243G mutation in (RT–)PCR products by primer extension
with primer #1973 in the presence of ddTTP and the three other
dNTPs.(c) Relative abundance of mutant (MT) and wild-type (WT) in patient
tissues carrying the A3243G mutation analyzed by the OXOCIRC assay followed
by primer extension. Extension products were analyzed by electrophoresis in a
20% polyacrylamide gel. D, mtDNA; T, total small RNAs circularized without
oxidation; A, total small RNAs circularized with previous oxidation. The
quantitation is described in Materials and Methods. (d) Control of the OXOCIRC
assay. Primer extension on RT–PCR products derived from a mixture of nonaminoacylated tRNAAsp(U) and glycyl-tRNAAsp(C) that was circularized without
(lane 2) and with (lane 3) previous oxidation. See Figure 1 for a diagram. The
upper band is derived from tRNAAsp(U), the lower band from tRNAAsp(C). RT–PCR
products derived from pure tRNAAsp(C) and tRNAAsp(U) were analyzed in lanes 1 and
4, respectively. The relative abundance of tRNAAsp(U) is 71% before and 24% after
oxidation in the experiment shown. The band immediately above the primer results
from non-specific incorporation of nucleotides by thermosequenase. Apart from
this band, no extension products are observed in the absence of the specific
template sequence, indicating that the intensity of the two informative bands is not
affected by non-specific incorporation (data not shown).

Human Molecular Genetics, 2000, Vol. 9, No. 4

469

Figure 3. OXOCIRC assay on tRNALys isolated from MERRF patients. (a) Detection of circular tRNALys by RT–PCR. The black arrowhead indicates the primer (#1593)
used for reverse transcription whereas shaded arrowheads indicate the primers (#2050, #2049 and #1747) used for RT–PCR. Note that primer #1747 was used only for
secondary amplification of products generated with primers #2050 and #2049, ensuring the selective amplification of cDNAs derived from circular tRNAs. (b) Detection
of the A8344G mutation in (RT–)PCR products by primer extension with primer #1974. (c) Relative abundance of mutant (MT) and wild-type (WT) tRNALys in MERRF
patient tissues analyzed as described in (a) and (b). Extension products were analyzed on a 20% polyacrylamide gel and quantitated as described in Materials and Methods.
When a PCR product generated from a bacterial plasmid carrying the mutant tRNALys was analyzed in an identical fashion, the resulting ratio of lower to upper band was
∼400:1, much higher than in the case of the patients with the highest ratio of mutant (∼20:1 or 95%, respectively), confirming the heteroplasmic status of the latter (data not
shown). D, mtDNA; T, total small RNAs circularized without oxidation; A, total small RNAs circularized with previous oxidation. (d) pCp 3′ end-labeling (16,27) of total
small RNAs from MERRF patient sample 2B before (lane 1) and after (lane 2) oxidation. The arrow indicates the position of 5S rRNA.

Effects of the A3243G mutation in MELAS patients

Control experiments for MELAS samples

The OXOCIRC assay was performed on seven muscle biopsies
and one lymphoblastoid cell line from patients who carry the
A3243G mutation in the tRNALeu(UUR) gene and display
MELAS or related symptoms (Fig. 2a and b) (see Materials
and Methods for case histories). Representative results are
shown in Figure 2c and the results from all patients are
summarized in Table 1. A decrease of the relative amounts of
the mutant in the processed tRNA pool (Fig. 2c, lanes T)
compared with its abundance in the mtDNA fraction (lanes D)
is observed in four samples (patients 1, 10, 12 and 13) whereas
four other samples exhibit no or only minor deviations from
the ratios in mtDNA (patients 9, 11, 14 and 17). Oxidation of
tRNAs prior to circularization (Fig. 2c, lanes A) results in a
lower relative abundance of the mutant sequence in six
samples (patients 1, 9, 10, 11, 13 and 17), whereas no effect is
apparent in the tRNAs from the lymphoblastoid cell line and
one muscle biopsy (patient 14). Thus, in seven of eight patient
samples the tRNALeu carrying the A3243G mutation is underrepresented in the (RT–)PCR products derived from
aminoacylated tRNA compared with its abundance among
those derived from mtDNA.

Several controls were performed to ensure that the variable effect
of the A3243G mutation on aminoacylation in different patient
samples is not due to variations in the assay system. To monitor
the efficiency of oxidation of non-aminoacylated tRNAs, in vitro
aminoacylated glycyl-tRNAAsp(C) from opossum, which carries
the anticodon GCC, and a non-aminoacylated tRNAAsp(U), which
carries the anticodon GUC but otherwise the same sequence as the
glycyl-tRNAAsp(C) were mixed and divided into two aliquots
(16,18) (Fig. 1). While one aliquot was first deacylated, then
circularized with RNA ligase, and amplified by RT–PCR, the
other aliquot was oxidized with NaIO4 prior to deacylation,
circularization and RT–PCR. Both RT–PCR products were
analyzed by primer extension, and the relative abundance of
non-aminoacylated tRNAAsp(U) was found to be 71% in the nonoxidized and 24% in the oxidized aliquot, confirming that
non-aminoacylated 3′ ends were decreased by oxidation (Fig. 2d).
In two further experiments, the non-aminoacylated tRNAAsp(U)
was decreased from 57 and 62% before to 22 and 21% after
oxidation, respectively (data not shown). It should be noted that 3′
ends of non-aminoacylated species cannot be completely
eliminated by oxidation even if the excess of NaIO4 is increased
(see Materials and Methods) (16). However, oxidation resulted in

470

Human Molecular Genetics, 2000, Vol. 9, No. 4

Table 1. Relative abundance of mutation in mtDNA, total tRNAs and aminoacylated (aa-)tRNAs
Patient

Total tRNA aa-tRNA
(%MT)b
(%MT)c

Expression Aminoacylation
reduced
reduced

Tissue

mtDNA
(%MT)a

1

Muscle

65.4 ± 2.6 45.6 ± 5.6

20.5 ± 3.6 +

+

9

Muscle

55.9 ± 1.0 57.2 ± 2.0

38.6 ± 0.8 –

+

10

Muscle

71.7 ± 0.3 56.3 ± 1.0

38.8 ± 1.1 +

+

11

Muscle

85.3 ± 1.0 78.0 ± 0.3

48.8 ± 1.1 ±

+

12

Cell line

80.6 ± 0.8 67.5 ± 1.8

67.9 ± 1.2 +

–

13

Muscle

38.4 ± 1.2 22.9

9.0

+

+

14d

Muscle

77.4 ± 0.7 78.4 ± 1.2

78.2 ± 2.5 –

–

17d

Muscle

81.6 ± 2.5 78.0

52.0

+

2A

Muscle

93.5 ± 0.7 95.5 ± 0.4

93.3 ± 0.2 –

–

2B

Muscle

93.8 ± 1.5 91.3 ± 0.9

93.4

–

–

MELAS

–

MERRF

5

Muscle

92.6 ± 1.4 89.4 ± 2.2

90.4 ± 2.3 –

–

22d

Muscle

92.0 ± 0.9 88.0 ± 0.2

87.0 ± 1.7 –

–

23d

Muscle

82.8 ± 1.2 80.5 ± 1.5

86.0 ± 2.8 –

–

24

Fibroblasts 91.5 ± 0.9 90.4 ± 1.0

89.7 ± 1.5 –

–

25

Fibroblasts 71.6 ± 1.4 67.0 ± 2.2

75.2

±

–

pT3/pT1

Cybrids

95.3 ± 0.6 69.3 ± 3.3

69.5 ± 5.1 +

–

pT3/LB64 Cybrids

90.0 ± 0.8 82.8 ± 2.1

78.3 ± 1.7 +

±

aMeans ± standard deviation based on primer extension analysis of two or more PCR products derived

from the same extract.
bMeans ± standard deviation based on primer extension analysis of two or more RT–PCR products
derived from the same circularization (9,12, 2A, 22, 23, pT3/pT1, pT3/LB64) or from two different
circularizations (1, 11, 14, 2B, 5, 24) of the same extract. Generation of RT–PCR products succeeded
only once for samples 13 and 17.
cMeans ± standard deviation based on primer extension analysis of two or more RT–PCR products
derived from the same oxidation/circularization (9, 12, 22, 23, pT3/pT1, pT3/LB64) or from two
independent oxidations/circularizations (1, 11, 14, 2A, 5, 24) of the same extract. Generation of RT–
PCR products succeeded only once for samples 13, 17, 2B and 25.
dSamples previously used for thin sectioning.

an at least 6.1 (±1.4)-fold relative decrease of the nonaminoacylated tRNA species in the three samples. [The
tRNAAsp(U)/tRNAAsp(C) ratios were changed from 2.4, 1.3 and 1.6
without oxidation to 0.32, 0.28 and 0.27 with oxidation.] It should
also be noted that the observed variation between the control
experiments is too small to explain the differences in effects
observed among the patient samples. For example, in patient 11,
the mutant tRNALeu was shifted from 78 to 49% (patient 11)
indicating a 3.5-fold decrease in its relative abundance (the
mutant:wild-type ratio was ∼3.5 without oxidation and ∼1.0 with
oxidation) whereas it remained virtually unchanged in patient 14
(∼78%).
The mixture of aminoacylated/non-aminoacylated tRNAAsp
was also used to determine whether the oxidation might be
inhibited in certain patient samples (e.g. sample 14). Mixtures of
glycyl-tRNAAsp(C) and tRNAAsp(U) were added to the small RNA
fractions from patients 11 and 14 and taken through the
OXOCIRC assay. Using the appropriate primer combinations,
both circular tRNAAsp and tRNALeu were amplified by RT–PCR
from the same non-oxidized/circularized or oxidized/circularized
aliquots and the PCR products were analyzed by primer
extension. In the RT–PCR products derived from tRNAAsp the
abundance of the non-aminoacylated tRNA species changed from
41 (without oxidation) to 17% (with oxidation) in patient 14

(∼3.4-fold decrease) and from 58 to 29% in patient 11 (∼3.4-fold
decrease; data not shown). In the same samples, no change in the
relative abundance of mutant tRNALeu was observed in patient 14
whereas in patient 11 the mutant sequence was decreased ∼3.5fold (Table 1 and Fig. 2c). Thus, in a sample where oxidation did
not result in a decrease of the mutant tRNALeu, oxidation had been
as efficient as in a sample where oxidation resulted in a clear
decrease of the mutant tRNA.
These results lead us to conclude that the tRNALeu carrying the
A3243G mutation is under-represented among aminoacylated
tRNAs in some patients whereas no such effect is detected in
others. In fact, among the eight samples analyzed, all possible
combinations of effects are seen. In patients 1, 10, 11 and 13, the
mutant is decreased in the total tRNA fraction compared with the
DNA fraction, and again in the aminoacylated tRNA fraction. In
patients 9 and 17, the relative abundance of mutant molecules in
the processed tRNA fraction is close to that of the DNA fraction,
but drops in the aminoacylated tRNA fraction. In the
lymphoblastoid cell line from patient 12, there is a drop in the
relative abundance of mutant from the DNA to the total tRNA
fraction, but no decline in the aminoacylated tRNA fraction.
Finally, in patient 14, no change is seen either in the total or in the
aminoacylated tRNA fractions.

Human Molecular Genetics, 2000, Vol. 9, No. 4

471

No effects of the A8344G mutation in MERRF patients
In further experiments, five muscle biopsies and two fibroblast
cell lines from MERRF patients containing the A8344G mutation
in the tRNALys gene were analyzed in a way analogous to that of
the MELAS samples (see Fig. 3a and b for details). Figure 3c
shows the extension products generated from (RT–)PCR products
derived from mtDNA, total and aminoacylated tRNAs from three
representative samples. Table 1 summarizes the findings. In all
cases, there are only minor deviations in the abundance of the
mutant sequence from that in mtDNA. Although the mutant is
slightly decreased in some cases from mtDNA to total tRNAs
(e.g. samples 5, 22, 23 and 25), oxidation results in a slight
increase of the mutant sequence. Thus, in the samples from six
MERRF patients, the mutated tRNALys is present in total tRNAs
and in the aminoacylated tRNA fraction at approximately the
same abundance as would be anticipated from the mtDNA. The
mutation per se seems to have no or only minor effects on either
tRNA steady state levels or in vivo aminoacylation.
Control experiments for MERRF samples
As previously described for tRNALeu, the efficiency of oxidation
was monitored by adding a mixture of aminoacylated opossum
glycyl-tRNAAsp(C) and non-aminoacylated tRNAAsp(U) to the tRNA
preparations from patients 2B and 5. When the mixture of these
tRNAs was co-oxidized with tRNAs isolated from patient sample
2B, the relative abundance of RT–PCR products derived from
non-aminoacylated tRNAAsp(U) changed from 47 to 9% whereas
the abundance of the mutant tRNALys remained virtually
unaffected (91 and 93% before and after oxidation, respectively).
Similarly, in patient 5, the non-aminoacylated tRNAAsp(U) was
decreased from 47 to 17% whereas the relative abundance of
mutant tRNALys was 87 and 90% before and after oxidation,
respectively (data not shown). Furthermore, 3′-end-labeling with
[32P]pCp of total small RNAs from patient 2B showed a ≥6-fold
decrease of 3′ ends in the 5S rRNA after oxidation relative to
labeled tRNAs indicating both efficient oxidation and protection
of tRNAs by aminoacylation (Fig. 3d).
Effect of the A8344G mutation in cybrid cell lines
The absence of a detectable effect of the A8344G MERRF
mutation on abundance or aminoacylation of the tRNALys was
surprising since previous studies comparing cybrid cell lines
homoplasmic for the A8344G transition or the wild-type sequence
had reported a reduction of tRNALys in processed (11,12) and in
aminoacylated (12) tRNAs for the mutant cell lines. (Cybrids have
the same nuclear background but carry different mitochondria.)
To determine whether the apparently contradictory results were
due to differences in cell material or in experimental procedures,
we performed the OXOCIRC assay on total small RNAs isolated
from the pT1 and the LB64 cybrid cell lines carrying the A8344G
mutation. Initial analysis of (RT–)PCR products derived from
mtDNA, circularized or circularized/oxidized tRNAs by primer
extension showed that no wild-type sequence was detectable in
these cell lines, indicating that they contain ≥98% of mtDNAs
carrying the A8344G transition (Fig. 4 and data not shown). To
compare the effects of the MERRF mutation in cybrids and
patient tissue, total small RNAs from a cell line homoplasmic for
the wild-type sequence at mitochondrial position 8344 (pT3) were

Figure 4. OXOCIRC assay on (RT–)PCR products generated on mixtures of
DNA/RNA from MERRF and wild-type cybrids, respectively. Lanes 1–3, mixture
from pT1 (MT) and pT3 (WT); lanes 4–6, mixture of deacylated pT1 and nondeacylated pT3; lanes 7–9, mixture of LB64 (MT) and pT3 (WT); lanes 10–12,
mixtures of deacylated LB64 and non-deacylated pT3; lane 13, oxidized and
circularized RNAs from pT1 (MT); lane 14, oxidized and circularized RNAs from
pT3 (WT). D, mtDNA; T, total small RNAs circularized without oxidation; A, total
small RNAs circularized with previous oxidation.

mixed with extracts from mutant cell lines (pT1 or LB64) at ratios
similar to those found in patient tissues. The mixtures were
subsequently analyzed by the OXOCIRC assay.
When DNA/RNA extracts from wild-type (pT3) and mutant
(pT1) cybrids are mixed such that ∼95% of the mtDNAs carry the
MERRF mutation, the abundance of the mutation is decreased to
∼69% in the fraction of tRNAs that can be circularized (Fig. 4).
Oxidation prior to circularization does not change the abundance
of the mutant suggesting that mutant and wild-type tRNALys are
aminoacylated to a similar extent in the two cell lines (Fig. 4, lanes
1–3). To test whether the OXOCIRC assay would have detected
decreased aminoacylation of the mutant tRNALys, isolates from
the pT3 and pT1 cell lines were mixed at a similar ratio as before
(~84% MT) but prior to mixture the aliquot from the pT1 (mutant)
cell line had been deacylated by incubation at 37°C in 20 mM Tris
(pH 8.3) for 30 min. When the abundance of the mutant was
assayed in this aliquot, it was decreased from ∼62% prior to
oxidation to ∼43% (Fig. 4, lanes 3–6). This shows that a decrease
in aminoacylation of the mutant tRNA would have been detected
by the assay. Furthermore, it confirms that a substantial fraction of
the tRNALys was aminoacylated in the extracts from the mutant
cybrids. Our results suggest that, whereas the steady-state level of
tRNALys is decreased in the mutant cybrid cell line pT1 compared
with the respective wild-type cell line, the mutation does not affect
the level of aminoacylation.
The same set of experiments was performed with RNAs
isolated from another 143B-derived cybrid cell line carrying the
A8344G mutation in homoplasmic form (LB64). In a mixture
containing ∼90% of mutation at the mtDNA level (wild-type
pT3), the mutation was present at ∼83% in the (RT–)PCR product
derived from processed tRNALys and at ∼78% in the RT–PCR
product derived from oxidized/circularized tRNALys (Fig. 4, lanes
7–9). Deacylation of the LB64 RNAs prior to mixture with those
from pT3 and subsequent oxidation resulted in a shift from ∼84 to
∼64% confirming that a substantial proportion of the mutant
tRNALys is aminoacylated in this cell line (Fig. 4, lanes 10–12).
These results suggest that cybrid cell lines carrying the mutant
tRNALys gene may indeed contain a lower level of aminoacylated
tRNALys compared with a cell line homoplasmic for the wild-type
sequence (11,12). This decrease is due mainly to a lower amount
of processed tRNAs. No similar effect is detectable in the seven
tissue samples from MERRF patients analyzed in this study.

472

Human Molecular Genetics, 2000, Vol. 9, No. 4

Figure 5. Combined effects of the MELAS (a) and MERRF (b) mutations on
tRNA abundance and aminoacylation in patient tissue. (a) Relative abundance
(MT/WT) of the A3243G MELAS mutation in mtDNA compared with that in
aminoacylated tRNAs. (b) Relative abundance (MT/WT) of the A8344G mutation
in mtDNA compared with that in aminoacylated tRNAs. Numbers indicate patient
samples. The dashed line marks the expected position for ratios unchanged
between the two fractions. MT, mutant; WT, wild-type; mtDNA, mitochondrial
DNA; aa-tRNA, aminoacylated tRNAs.

DISCUSSION
The effects of the MELAS and the MERRF mutations on the
relative abundance of the respective mutations in aminoacylated
tRNAs are summarized in Figure 5. In the case of tissue from
MELAS patients, the tRNALeu(UUR) carrying the A3243G mutation
is decreased on average 4-fold in the aminoacylated tRNA
fraction compared with its abundance in the mtDNA fraction due
to the combined decrease of the mutant tRNA in total and/or
aminoacylated tRNAs (Fig. 5a). In contrast, no similar effect is
detectable in biopsies from MERRF patients where all patient
samples fall close to the line of expected fit (Fig. 5b). This
suggests important differences in the molecular pathogenesis of
the MELAS and MERRF syndromes.
In the case of MELAS, half of the tissue samples examined
show a decrease of tRNAs carrying the A3243G mutation. In six
patients the relative abundance of tRNALeu carrying the A3243G
mutation is (further) decreased in aminoacylated tRNAs
compared with total tRNAs. Assuming that the level of gene
expression of mtDNA molecules carrying the wild-type sequence
is unaffected by the presence of the A3243G mutation in
heteroplasmic tissue, it seems reasonable to extrapolate these
relative decreases in the mutant tRNALeu(UUR) among
aminoacylated tRNAs to the absolute amount of available
aminoacylated tRNALeu(UUR). Under this assumption, the average
mitochondrion in biopsies from all but one patient should contain
<70% of aminoacylated tRNALeu(UUR) (mutant plus wild-type)
compared with a situation where all mtDNA molecules are wildtype. In a case like patient 11, the level of aminoacylated tRNALeu
may be as low as 30% of wild-type. In muscle fibers containing an
accumulation of mutant mtDNA (19), a decreased level of
aminoacylated tRNALeu(UUR) may result in stalling of ribosomes at
leucine codons, mistranslation of UUR codons or translational
frame shifting (20). The possibility that decreased levels of
aminoacylated tRNALeu(UUR) is the primary defect in MELAS is
supported by the recent isolation of a suppressor mutation in the
anticodon of the other mitochondrial tRNALeu(CUN) gene in a
A3243G lung carcinoma cell line (21). Although the suppressor

tRNA is efficiently aminoacylated in the lung carcinoma cell line,
aminoacylation of the tRNALeu(UUR) is severely decreased (22).
The decreased availability of aminoacylated tRNALeu(UUR) may
be enough to explain the pathogenicity of the A3243G mutation.
At the same time, our results suggest that this mutation has a
variable effect on the expression and aminoacylation levels of
tRNALeu(UUR) in the patient population indicating that additional
factors may contribute to the molecular pathogenesis of MELAS.
It is possible that the extent of effects induced by the A3243G
mutation vary between different tissues or parts of tissues in a
given patient, or in a given tissue over time. In addition, nuclear
and mitochondrial background might modulate the effects of the
mutation on tRNA function. More extensive work, including
patients with milder as well as more restricted symptoms, will be
required before insights about the exact correlation between
clinical symptoms and the amount and state of aminoacylation of
tRNALeu can be achieved.
In the case of MERRF, our results, in contrast to previous
studies of homoplasmic cybrid cell lines, indicate that the
A8344G mutation exerts its effects (11–14) in the tissues of
affected individuals by mechanisms other than reduced
accumulation of tRNALys and/or reduced availability of
aminoacylated tRNALys. Such effects could include, for example,
partial misacylation and/or incorrect interactions between the
mutant tRNALys and translation factors.
MATERIALS AND METHODS
Case histories
Patients with the A3243G mutation. Patient 1: male, 35 years,
juvenile onset, seizures, bilateral hearing loss, proximal
myopathy, mental deterioration, polyneuropathy; patient 9:
female, 45 years, typical MELAS syndrome, sister of patient 13;
patient 10: male, juvenile onset, progressing dementia, insulindependent diabetes, episodes of hemianopsia, hemiparesis, motor
aphasia, ataxia, frontal headaches, generalized tonic-clonic
seizures (grand-mal), hypertrophic cardiomyopathy, death at 42
years from cardiac failure; patient 11: female, complete MELAS
syndrome, onset at age 19; patient 12: male, severe MELAS
syndrome from age 1 year, death at age 3 years; patient 13: male,
39 years, typical MELAS syndrome, juvenile onset, brother of
patient 9; patient 14: female, 18 years, mental retardation, short
stature, adult onset of seizures, muscles with ragged red fibers;
patient 17: female, 28 years, epileptic seizures, muscle atrophy of
the legs, moderate dementia, muscles with ragged red fibers.
Patients with the A8344G mutation. All but one sample (patient
22) are from patients exhibiting typical MERRF syndrome.
Sample 22 is from an asymptomatic relative with ragged red
fibers, whose mother and maternal uncle had adult-onset muscle
weakness. Samples 2A and 2B are two different biopsies from the
same patient.
Cell lines and media
The pT1 and pT3 human cell lines, which are homoplasmic for the
mutant G and the wild-type A, respectively, in the mitochondrial
tRNALys gene at position 8344, were derived from a 143B206(ρ°)
cell line and have been described previously (12) (a gift from G.
Attardi, California Institute of Technology). The LB64 human cell

Human Molecular Genetics, 2000, Vol. 9, No. 4

line carrying in homoplasmic form the mutant G at position 8344
was also derived from 143B206 and was generated as described
for other A8344G cell lines by Masucci et al. (11) (a gift from E.
Schon, Columbia University). Cell lines were cultured in
Dulbecco’s modified Eagle’s medium containing 4.5 mg glucose/
ml and 110 µg pyruvate/ml supplemented with 10% fetal bovine
serum.
Oligonucleotides
Nucleotide numbers of human primers are according to Anderson
et al. (23) and of opossum primers (#537, 740, 2169) according to
Janke et al. (24). L denotes primers with L-strand sequences, H
those with H-strand sequences. Primer 1747 spans the ligation site
between the 5′ and 3′ ends of the tRNALys. Post-transcriptionally
added 3′ CCA sequences are underlined.
537, H7043–7063, 5′-AAACCTATAATTTAACTATG-3′;
553, M13 universal, 5′-GTTTCCCAGTCACGAC-3′;
554, M13 reverse, 5′-AGCGGATAACAATTTCACAGGA-3′;
740, L7061–7080, 5′-TTTAACTCCTATATATCTT-3′;
914, L3200–3219, 5′-TATACCCACACCCACCCAAG-3′;
915, H3531–3550, 5′-CGATGGTGAGAGCTAAGGTC-3′;
1593, H8294–8311, 5′-TAAGTTAGCTTTACAGTG-3′;
1620, H3261–3279, 5′-GACTGTAAAGTTTTAAGTT-3′;
1747, H8298–8295–3′ CCA-tail-H8364–8355, 5′-AGTGTGGTCACTGTA-3′;
1904, L3269–3285, 5′-ACTTTACAGTCAGAGG-3′;
1905, L3283–3300, 5′-GGTTCAATTCCTCTTCTT-3′;
1973, H3247–3267 5′-TTAAGTTTTATGCGATTACCG-3′;
1974, H8347–8364 + 3′ CCA-tail, 5′-TGGTCACTGTAAAGAGGTGTT-3′;
2049, L8314–8337, 5′-ATAACCTTTTAAGTTAAGATT-3′;
2050, H8296–8316, 5′-TATGCTAAGTTAGCTTTACAG-3′;
2169, L7023–7040, 5′-AATTCATTACATAACTTT-3′;
SIDS, H8639–8662, 5′-GAATTCTGGGTGGTGATTAGTCGGTTGTTGA-3′.
Isolation of small RNAs
All biopsies had been frozen immediately after removal from
patients and stored at –80°C. Samples 14, 17, 22 and 23 had
been previously surrounded by a plastic matrix to affix them
during thin sectioning in a cryostat. This matrix was removed
prior to homogenization. Thin sectioning requires no chemical
treatment of the samples, so these samples were processed in
the same way as other samples. Frozen muscle samples (20–
60 mg) were cut into small pieces and homogenized in a 5 ml
glass–Teflon homogenizer with 0.5 ml of isolation buffer
(50 mM sodium acetate pH 5.0, 10 mM MgCl2) plus 1% SDS.
Human cell lines (20–40 mg) which had been frozen as pellets
were thawed on ice and resuspended in isolation buffer plus
1% SDS. After an extraction with 1 vol of phenol saturated
with isolation buffer without SDS, and two extractions with
1 vol of chloroform, 0.2 M potassium acetate (pH 5.0) was
added to the aqueous phase followed by isopropanol
precipitation (16). An aliquot of the isopropanol–aqueous
mixture was removed for analysis of mtDNA. The precipitate
was responded in isolation buffer and applied to a DEAE–
cellulose matrix (DE52; Whatman, Maidstone, UK). After
three washes with isolation buffer plus 100 mM NaCl, small
RNAs were eluted with isolation buffer plus 900 mM NaCl.

473

OXOCIRC assay
Oxidation, deacylation and circularization of tRNAs from patient
samples were essentially as described (16). Eluates from the
DEAE matrix were divided into two aliquots and precipitated with
3 vol of ethanol. One pellet was resuspended in 32 µl of ddH2O
containing 10 µl of 4× circularization buffer (200 mM Tris–HCl
pH 7.7, 40 mM MgCl2, 0.4 µg/µl bovine serum albumin),
deacylated by heating for 5 min to 70°C and cooling for 30 min at
37°C and subsequently circularized for 2 h at 37°C by adding
0.6 µl of 10 mM ATP, 6 µl of DMSO and 40 U of T4 RNA ligase
(New England Biolabs, Beverly, MA). The other pellet was
resuspended in 20 µl of oxidation buffer (10 mM sodium acetate,
10 mM MgCl2, 1 mM EDTA, 15 mM β-mercaptoethanol) (25),
and 20 µl of 65 mM NaIO4 in oxidation buffer was added and
incubated for 20 min at room temperature in the dark. Since <5 µg
of small cellular RNAs could be isolated from muscle biopsies,
this equals a ≥1000-fold excess of NaIO4 over 3′ ends. Higher
NaIO4 concentrations do not increase the oxidation efficiency yet
interfere with subsequent enzymatic manipulations (data not
shown). Oxidation was stopped by the addition of glucose in
equimolar amounts to NaIO4 followed by two subsequent ethanol
precipitations in the presence of 20 µg of glycogen. The pellet of
oxidized small RNAs was resuspended in 32 µl of ddH2O
containing 10 µl of 4× circularization buffer, deacylated and
circularized as described above. To estimate the efficiency
of circularization, total cellular tRNAs from HeLa cells were
circularized, separated by electrophoresis on a 10%
polyacrylamide gel and analyzed by RT–minisequencing (16) of
RNAs eluted from gel slices. Approximately 25% of the human
tRNALeu(UUR) and tRNALys, respectively, were detected in a
circular fraction migrating above the respective linear tRNAs
(data not shown). When a 5′ 32P-labeled in vitro transcript of a
mitochondrial tRNAAsp (16) was included in the circularization of
5 µg of total tRNAs from HeLa cells and analyzed by gel
electrophoresis, 90% of the in vitro transcript was shifted to a
single additional band above the linear tRNA, indicating that
circularization is favored over tRNA multimerization under these
conditions (data not shown).
cDNA synthesis
One microliter from the circularization reactions with oxidized/
non-oxidized tRNAs was transferred into a 20 µl cDNA synthesis
assay containing 0.5 µM of the respective gel-purified primer,
50 µM dNTPs and the appropriate buffer. After annealing (92°C,
2 min; 24°C, 10 min; 0°C, 10 min) AMV reverse transcriptase
(Pharmacia, Uppsala, Sweden) was added and the reaction was
incubated at 42°C for 30 min. cDNA synthesis on circular human
tRNALeu(UUR), circular human tRNALys and circular marsupial
tRNAAsp was performed with primers #1620, #1593 and #537,
respectively.
PCR of cDNAs
One microliter from the cDNA synthesis was transferred into a
30 µl PCR reaction. PCR reactions contained 67 mM Tris–HCl pH
8.9, 4 mM MgCl2, 16 mM (NH4)2SO4, 10 mM β-mercaptoethanol,
0.5 mg/ml bovine serum albumin (26) and 0.3 µM of the respective
primers. cDNAs derived from circular human tRNALeu(UUR) were
amplified as follows. cDNAs from patients 1 and 11 were amplified
with primers #1620 and #1905 (92°C, 2 min; and 40 cycles of 94°C,

474

Human Molecular Genetics, 2000, Vol. 9, No. 4

30 s; 48°C, 30 s; 72°C, 30 s). The remaining MELAS samples were
amplified for 20 cycles with primers #1620 and #1904 (92°C,
2 min; 94°C, 30 s; 48°C, 30 s; 72°C, 30 s), and 1 µl of the reaction
was transferred into a new PCR reaction and cycled for another 20
cycles with the nested primer pair #1620 and #1905 (same
conditions as before). cDNAs derived from circular human tRNALys
were amplified as follows. cDNAs from patient 2A and from cybrid
cell lines (pT1, pT3, LB64) were amplified by PCR with primers
#2049 and #2050 (92°C, 2 min; and 40 cycles of 94°C, 30 s; 48°C,
30 s; 72°C, 30 s). cDNAs from other patients were amplified by
running 20 cycles with primers #2049 and #2050 (92°C, 2 min;
94°C, 30 s; 48°C, 30 s; 72°C, 30 s). After transferring 1 µl into a
new PCR reaction, another 20 cycles with primers #1747 and #2049
followed. cDNAs generated from opossum tRNAAsp were
amplified for 40 cycles with primers #537 and #740 (92°C, 2 min;
94°C, 30 s; 42°C, 30 s; 72°C, 30 s). In all cases, the appropriate
controls (circularization reaction to which no tRNAs had been
added followed by cDNA synthesis) were included. When
circularization was omitted or the circularization reaction was
treated with RNase A prior to cDNA synthesis, no PCR product of
the expected length was generated (data not shown).
PCR of mtDNA and cloned tRNA genes
PCR products representing the tRNALeu(UUR) gene were generated
with primers #914 and #915, recognizing sequences in the 16S
rRNA gene and in the ND1 gene, respectively (92°C, 2 min; and
35 cycles of 94°C, 30 s; 54°C, 30 s; 72°C, 30 s). PCR products
representing the tRNALys gene were generated with primers SIDS
and #2049, recognizing sequences in the ATPase 6 gene and the
tRNALys gene, respectively (92°C, 2 min; and 35 cycles of 94°C,
30 s; 50°C, 30 s; 72°C, 30 s). PCR products representing tRNALys
with the A8344G mutation cloned into pUC18 were generated
with primers #553 and #554 which recognize sequences adjacent
to the multiple cloning site (92°C, 2 min; and 35 cycles of 94°C,
30 s; 54°C, 30 s; 72°C, 30 s).
Primer extension
Non-incorporated primers and nucleotides were removed from
(RT–)PCR products by three washes with 200 µl of ddH2O during
ultracentrifugation (Ultrafree 30; Millipore, Bedford, MA).
Subsequently, the PCR products were resuspended in 20 µl of
ddH2O, 5 µl of which were included in the detection reaction. The
extension was done in 13 µl containing 4 mM MgCl2, 50 mM
Tris–HCl pH 8.6, 5 U of Thermosequenase (Amersham, Little
Chalfont, UK) and 0.1 pmol of the appropriate gel-purified 5′ 32Plabeled primer. The concentration of the informative ddNTP was
175 µM and those of the other three dNTPs were 70 µM. PCR
products derived from tRNALeu(UUR) were analyzed with primer
#1973 with ddTTP and the three other dNTPs, and those derived
from tRNALys were analyzed with primer #1974 plus ddCTP and
the three other dNTPs. Primer #2169 plus ddCTP and the three
other dNTPs were used to analyze PCR products derived from the
tRNAAsp from opossum. After three to five cycles between 92 and
56°C, 1 min each, reactions were stopped with loading buffer,
analyzed on a 20% polyacrylamide gel and quantitated with a
Fuji-Phosphoimager. Primer extension analysis of RT–PCR
products derived from pure wild-type tRNALys and pure mutant
tRNALeu reveals a weak ladder of bands that decrease in intensity
with increasing length leading up to the major extension product
(data not shown and Figs 2c and 3c). To account for these

unspecific extension products (presumably generated by
premature termination), in each lane the extension signals
immediately below the wild-type and mutant bands were
subtracted as background from the major extension signals,
respectively. Only tissue samples where both mutant and wildtype sequences could be detected in the mtDNA fraction were
further analyzed.
ACKNOWLEDGEMENTS
We thank G. Attardi and E.A. Schon for providing cell lines, W.
Schartau for synthesis of oligonucleotides, and N. Kleckner for
providing laboratory space during the completion of experiments.
Thanks for help and discussion go to H. Zischler, T. Witt, T.F.
Wang, S. Burgess, A. Guhathakurta, H. Poinar, M. Mörl, A.
Reichard, A. Greenwood, M. Dörner and W. Schaffner. This work
was supported by the DFG and by a Telethon grant (no. 767) to
M.Z.
REFERENCES
1. Grossman, L.I. and Shoubridge, E.A. (1995) Mitochondrial genetics and
human disease. Bioessays, 18, 983–991.
2. Wallace, D.C., Lott, M.T., Brown, M.D., Huoponen, K. and Torroni, A.
(1995) Report of the committee on human mitochondrial DNA. In
Cuticchia, A.J. (ed.), Human Gene Mapping 1995: A Compendium. Johns
Hopkins University Press, Baltimore, MD, pp. 910–945.
3. Shoffner, J.M., Lott, M.T., Lezza, A.M., Seibel, P., Ballinger, S.W. and
Wallace, D.C. (1990) Myoclonic epilepsy and ragged-red fiber disease
(MERRF) is associated with a mitochondrial DNA transfer RNALys
mutation. Cell, 61, 931–937.
4. Goto, Y., Nonaka, I. and Horai, S. (1990) A mutation in the tRNALeu(UUR)
gene associated with the MELAS subgroup in mitochondrial
encephalomyopathies. Nature, 348, 651–653.
5. Kobayashi, Y., Momoi, M.Y., Tominaga, K., Momoi, T., Nihei, K.,
Yanagisawa, M. and Ohta, S. (1990) A point mutation in the mitochondrial
tRNALeu(UUR) gene in MELAS (mitochondrial myopathy, encephalopathy,
lactic acidosis and stroke-like episodes). Biochim. Biophys. Res. Commun.,
173, 816–822.
6. Ciafaloni, E., Ricci, E., Shanske, S., Moraes, C.T., Silvestri, G., Hirano, M.,
Simonetti, S., Angelini, C., Donati, A., Garcia, C. et al. (1992) MELAS:
clinical features, biochemistry and molecular genetics. Ann. Neurol., 31,
391–398.
7. Gerbitz, K.-D., van den Ouweland, J.M.W., Maassen, J.A. and Jaksch, M.
(1995) Mitochondrial diabetes mellitus: a review. Biochim. Biophys. Acta,
1271, 253–260.
8. Chomyn, A., Martinuzzi, A., Yoneda, M., Daga, A., Hurko, O., Johns, D.,
Lai, S.T., Nonaka, I., Angelini, C. and Attardi, G. (1992) MELAS mutation
in mtDNA binding site for transcription termination factor causes defects in
protein synthesis and in respiration but no change in levels of upstream and
downstream mature transcripts. Proc. Natl Acad. Sci. USA, 89, 4221–4225.
9. King, M.P., Koga, Y., Davidson, M. and Schon, E.A. (1992) Defects in
mitochondrial protein synthesis and respiratory chain activity segregate
with the tRNALeu(UUR) mutation associated with mitochondrial myopathy,
encephalopathy, lactic acidosis, and strokelike episodes. Mol. Cell. Biol.,
12, 480–490.
10. Flierl, A., Reichmann, H. and Seibel, P. (1997) Pathophysiology of the
MELAS 3243 transition mutation. J. Biol. Chem., 272, 27189–27196.
11. Masucci, J.P., Davidson, M., Koga, Y., Schon, E.A. and King, M.P. (1995)
In vitro analysis of mutations causing myoclonus epilepsy with ragged-red
fibers in the mitochondrial tRNALys gene: two genotypes produce similar
phenotypes. Mol. Cell. Biol., 15, 2872–2881.
12. Enriquez, J.A., Chomyn, A. and Attardi, G. (1995) MtDNA mutation in
MERRF syndrome causes defective aminoacylation of tRNALys and
premature translation termination. Nature Genet., 10, 47–55.
13. Chomyn, A., Meola, G., Bresolin, N., Lai, S.T., Scarlato, G. and Attardi, G.
(1991) In vitro genetic transfer of protein synthesis and respiration defects
to mitochondrial DNA-less cells with myopathy mitochondria. Mol. Cell.
Biol., 11, 2236–2244.

Human Molecular Genetics, 2000, Vol. 9, No. 4

14. Boulet, L., Karpati, G. and Shoubridge, E.A. (1992) Distribution and
threshold expression of the tRNALys mutation in skeletal muscle of patients
with myoclonic epilepsy and ragged-red fibers (MERRF). Am. J. Hum.
Genet., 51, 1187–1200.
15. Wallace, D.C. (1992) Diseases of the mitochondrial DNA. Annu. Rev.
Biochem., 61, 1175–1212.
16. Börner, G.V., Mörl, M., Janke, A. and Pääbo, S. (1996) RNA editing
changes the identity of a mitochondrial tRNA in marsupials. EMBO J., 15,
5949–5957.
17. Pan, T., Gutell, R.G. and Uhlenbeck, O.C. (1991) Folding of circularly
permuted transfer RNAs. Science, 254, 1361–1364.
18. Börner, G.V. and Pääbo, S. (1996) Evolutionary fixation of RNA editing.
Nature, 383, 225.
19. Petruzella, V., Moraes, C.T., Sano, M.C., Bonilla, E., DiMauro, S. and
Schon, E.A. (1994) Extreme high levels of mutant mtDNAs co-localize
with cytochrome c oxidase-negative ragged fibers in patients harboring a
point mutation at nt 3243. Hum. Mol. Genet., 3, 449–454.
20. Kurland, C.G. and Gallant, J.A. (1986) The secret life of the ribosome. In
Kirkwood, T.B.L., Rosenberger, R.F. and Galas, D.J. (eds), Accuracy in
Molecular Processes. Chapman and Hall, London, UK, pp. 127–157.

475

21. Meziane, A.E., Lehtinen, S.K., Hance, N., Nijtmans, L.G.J., Dunbar, D.,
Holt, I.J. and Jacobs, H.T. (1998) A tRNA suppressor mutation in human
mitochondria. Nature Genet., 18, 350–353.
22. Meziane, A.E., Lehtinen, S.K., Holt, I.J. and Jacobs, H.T. (1998)
Mitochondrial tRNALeu isoforms in lung carcinoma cells containing the np
3243 mtDNA mutation. Hum. Mol. Genet., 7, 2141–1747.
23. Anderson, S., Bankier, A.T., Barrell, B.G., de Bruijn, M.H.L., Coulson,
A.R., Drouin, J., Eperon, I.C., Nierlich, D.P., Roe, B.A., Sanger, F. et al.
(1981) Sequence and organization of the human mitochondrial genome.
Nature, 290, 457–465.
24. Janke, A., Feldmaier-Fuchs, G., Thomas, W.K., von Haeseler, A. and
Pääbo, S. (1994) The marsupial mitochondrial genome and the evolution of
placental mammals. Genetics, 137, 243–256.
25. Andrulis, I.L. and Arfin, S.M. (1979) Methods for determining the extent of
tRNA aminoacylation in vivo in cultured mammalian cells. Methods
Enzymol., 59, 268–271.
26. Lisitsyn, N. and Wigler, M. (1995) Representational difference analysis in
detection of genetic lesions in cancer. Methods Enzymol., 254, 291–304.
27. England, T.E. and Uhlenbeck, O.C. (1978) 3′ terminal labeling of RNA with
T4 RNA ligase. Nature, 275, 560–561.

476

Human Molecular Genetics, 2000, Vol. 9, No. 4