ArticlePDF Available

Abstract and Figures

In human cells, ATP is generated using oxidative phosphorylation machinery, which is inoperable without proteins encoded by mitochondrial DNA (mtDNA). The DNA polymerase gamma (Polγ) repairs and replicates the multicopy mtDNA genome in concert with additional factors. The Polγ catalytic subunit is encoded by the POLG gene, and mutations in this gene cause mtDNA genome instability and disease. Barriers to studying the molecular effects of disease mutations include scarcity of patient samples and a lack of available mutant models; therefore, we developed a human SJCRH30 myoblast cell line model with the most common autosomal dominant POLG mutation, c.2864A>G/p.Y955C, as individuals with this mutation can present with progressive skeletal muscle weakness. Using on-target sequencing, we detected a 50% conversion frequency of the mutation, confirming heterozygous Y955C substitution. We found mutated cells grew slowly in a glucose-containing medium and had reduced mitochondrial bioenergetics compared to the parental cell line. Furthermore, growing Y955C cells in a galactose-containing medium to obligate mitochondrial function enhanced these bioenergetic deficits. Also, we show complex I NDUFB8 and ND3 protein levels were decreased in the mutant cell line, and the maintenance of mtDNA was severely impaired (i.e., lower copy number, fewer nucleoids, and an accumulation of Y955C-specific replication intermediates). Finally, we show the mutant cells have increased sensitivity to the mitochondrial toxicant 2'-3'-dideoxycytidine. We expect this POLG Y955C cell line to be a robust system to identify new mitochondrial toxicants and therapeutics to treat mitochondrial dysfunction.
Restriction endonuclease mapping of SJCRH30 POLG WT and Y955C mtDNA. A, detection of mtDNA restriction fragments with either an ND1 heavy (H)-strand (left) or ND4 H-strand (right) single-stranded (ss) DIG-labeled probe. B, maps of the expected mtDNA restriction fragment lengths when visualized with either the ND1 or ND4 probe. Panel 1 shows expected fragment lengths produced with BamHI/NheI double digestion, and panel 2 shows expected lengths produced with XbaI digestion. In both panels, the mtDNA light (L) strand is represented as a thin intact inner circle and the H-strand origin of replication (O H ), the L-strand origin of replication (O L ), and the ND1 (*) and ND4 (#) probe sequences are also highlighted on this strand. Note that the numbering of base pairs is counterclockwise and is based on mtDNA NC_012920. Nascent continuous leading H-strand synthesis, using the L-strand template and beginning at O H (position 191), is represented as a thick dashed line with replication proceeding clockwise around the circle. The parental/ template H-strand is shown as a displaced intact, thick circle that is used as the template strand during continuous nascent lagging L-strand synthesis, which is represented as a thin dashed line beginning at O L (position 5770) and proceeding counterclockwise. The location of two predicted G-quadruplexes (G4s) are shown in panels 1 and 2 as black circles above the parental H-strand, 2GQH located at positions 4260 to 4229, and 3GQH located at 15545 to 15516. C, summary of the expected and estimated molecular weights (MW) of mtDNA bands generated by restriction endonuclease (RE) digestions. PvuII-digested mtDNA fragment lengths greater than 10 kb were estimated using linear regression of the log10 base pair values of Lambda DNA/HindIII Marker 2 fragments loaded onto the same gel versus distance traveled in millimeters (R 2 values were ≥ 0.94). For the remaining restriction digests that generate fragment lengths less than 10 kb, linear regression of the log10 base pair values of exACTGene 1 kb Plus DNA ladder fragments versus distance traveled in millimeters was utilized to estimate molecular weights (R 2 values were ≥ 0.99).
… 
Content may be subject to copyright.
Heterozygous p.Y955C mutation in DNA polymerase γleads
to alterations in bioenergetics, complex I subunit expression,
and mtDNA replication
Received for publication, January 23, 2022, and in revised form, June 16, 2022 Published, Papers in Press, June 24, 2022,
https://doi.org/10.1016/j.jbc.2022.102196
Md. Mostajur Rahman
1,
, Carolyn K. J. Young
1,
, StefGoffart
2
, Jaakko L. O. Pohjoismäki
2
, and
Matthew J. Young
1,
*
From the
1
Department of Biochemistry and Molecular Biology, Southern Illinois University School of Medicine, Carbondale,
Illinois, USA;
2
Department of Environmental and Biological Sciences, University of Eastern Finland, Joensuu, Finland
Edited by Craig Cameron
In human cells, ATP is generated using oxidative phos-
phorylation machinery, which is inoperable without proteins
encoded by mitochondrial DNA (mtDNA). The DNA poly-
merase gamma (Polγ) repairs and replicates the multicopy
mtDNA genome in concert with additional factors. The Polγ
catalytic subunit is encoded by the POLG gene, and mutations
in this gene cause mtDNA genome instability and disease.
Barriers to studying the molecular effects of disease mutations
include scarcity of patient samples and a lack of available
mutant models; therefore, we developed a human SJCRH30
myoblast cell line model with the most common autosomal
dominant POLG mutation, c.2864A>G/p.Y955C, as individuals
with this mutation can present with progressive skeletal muscle
weakness. Using on-target sequencing, we detected a 50%
conversion frequency of the mutation, conrming heterozy-
gous Y955C substitution. We found mutated cells grew slowly
in a glucose-containing medium and had reduced mitochon-
drial bioenergetics compared with the parental cell line.
Furthermore, growing Y955C cells in a galactose-containing
medium to obligate mitochondrial function enhanced these
bioenergetic decits. Also, we show complex I NDUFB8 and
ND3 protein levels were decreased in the mutant cell line, and
the maintenance of mtDNA was severely impaired (i.e., lower
copy number, fewer nucleoids, and an accumulation of Y955C-
specic replication intermediates). Finally, we show the mutant
cells have increased sensitivity to the mitochondrial toxicant
20-30-dideoxycytidine. We expect this POLG Y955C cell line to
be a robust system to identify new mitochondrial toxicants and
therapeutics to treat mitochondrial dysfunction.
Cells synthesize most of their ATP using the mitochondrial
oxidative phosphorylation (OXPHOS) machinery, and this
machinery requires 13 mitochondrial DNA (mtDNA)-enco-
ded proteins to function. Thus, mtDNA maintenance is
essential to meeting the basic energy demands within our
cells. The human mtDNA genome is a covalently closed cir-
cular double-stranded 16,569-bp molecule that harbors the
above-mentioned 13 OXPHOS genes in addition to 2 genes
encoding rRNAs and 22 genes coding for tRNAs. The 24 RNA
genes are required to translate the 13 mtDNA-encoded
polypeptides. Disease mutations are associated with all 37
mtDNA genes, and 1 in 200 healthy individuals harbors a
pathogenic mtDNA mutation that could cause disease in a
child born to a female carrier (1,2), underscoring the
importance of maintaining the mtDNA genome.
The multicopy mtDNA genome is replicated and repaired
throughout the cell cycle by the mtDNA polymerase gamma
(Polγ) in concert with additional replisome factors, for
example, Twinkle mtDNA helicase, topoisomerases, mito-
chondrial single-stranded DNA-binding protein, and others
(3). According to the strand displacement model of mtDNA
replication, replisomes containing Polγsynthesize both the
nascent heavy (H) and light (L) strands continuously without
the formation of Okazaki-fragment-like replication products
(4). The two mtDNA strands are named H and L based on the
ability to separate them on denaturing cesium chloride gradi-
ents. The H-strand is richer in G + T content, making it
heavier on density centrifugation (5,6). The Polγholoenzyme
is composed of three subunits encoded by two nuclear DNA
(nDNA) genes: (1) POLG codes for the 140-kDa catalytic
subunit, p140 (or Polγα), and (2) POLG2 encodes the 110-
kDa homodimeric accessory subunit, p55 (or Polγβ). Muta-
tions in POLG and POLG2 are associated with primary
mitochondrial disorders that cause mtDNA instability, such as
mtDNA depletion and deletions. POLG mutations are the most
common cause of inherited mitochondrial disease, and the
number of individuals harboring a POLG mutation is estimated
to be 2% of the population (7).
The most common autosomal dominant POLG mutation is
the c.2864A>G substitution, which encodes a substitution of
the p140 tyrosine (Tyr/Y) 955 amino acid residue for cysteine
(Cys/C), p140 Y955C (8). In a groundbreaking 2001 paper, the
heterozygous POLG c.2864A>G/p.Y955C mutation (hereafter
POLG Y955C) cosegregated with autosomal dominant pro-
gressive external ophthalmoplegia (adPEO) in a family,
establishing that a POLG mutation causes the disorder (9).
Manifestation of adPEO is characterized by adult-onset
These authors contributed equally to this work.
*For correspondence: Matthew J. Young, matthew.young@siu.edu.
RESEARCH ARTICLE
J. Biol. Chem. (2022) 298(8) 102196 1
© 2022 THE AUTHORS. Published by Elsevier Inc on behalf of American Society for Biochemistry and Molecular Biology. This is an open access article under the CC
BY license (http://creativecommons.org/licenses/by/4.0/).
progressive weakness of the extraocular eye muscles resulting
in strabismus and ptosis (7,8). In addition, muscle weakness
can extend to the limb-girdle skeletal muscles (manifesting as
generalized myopathy or proximal myopathy). Other symp-
toms include parkinsonism, ataxia, sensorineural hearing loss,
depression, cataracts, and premature ovarian failure (8). Un-
related patients with POLG Y955C mutations harbor mtDNA
deletions in skeletal muscle samples and 20% to 60% re-
ductions in mtDNA content relative to control subjects
(912). In a study where p140 Y955C was overexpressed in
293 Flp-In TRex cells, mtDNA levels were reduced by >60%
(13).
Several biochemical and biological models have been
developed to understand the molecular mechanisms contrib-
uting to mitochondrial dysfunction associated with adPEO.
However, due to the scarcity of patient samples, the molecular
and physiological details regarding the effects of adPEO mu-
tations on human cell mtDNA homeostasis, bioenergetics, and
OXPHOS machinery remain poorly understood. Based on
previous biochemical, crystal structure, and molecular model
work, the Y955C p140 substitution has been demonstrated to
localize at the PolγDNA polymerase active site and cause
signicantly reduced dNTP incorporation. Still, the Y955C
variant enzyme maintains wild-type (WT)-like DNA-binding
(1418). Additionally, the 30-50exonuclease (exo+), or proof-
reading procient, Y955C Polγholoenzyme is 2-fold less ac-
curate than WT Polγdespite having a functional exo+ domain
(14,19). Intriguingly, in another study, when provided a high
concentration of dATP (200 μM) and 1 μM other dNTPs,
Y955C PolγDNA synthesis activity could be restored. The
authors elegantly showed that when WT and Y955C Polγ
holoenzymes are mixed using a constant concentration of
dNTPs (1 μM), Y955C Polγis dominant-negative, and via its
exo+ activity, chews back nascently synthesized DNA and
likely stalls on the template in an idling mode, which blocks
the activity of WT Polγ(13). In agreement with an autosomal
dominant mode of inheritance, an in vitro rolling circle
replication assay demonstrated that the Y955C Polγholoen-
zyme has a dominant-negative effect on DNA synthesis in the
presence of the WT Polγholoenzyme at dNTP concentrations
of 1 and 10 μM(10).
In yeast models harboring the orthologous POLG Y955C
mutation (p.Y757C), haploid yeast suffered a complete loss of
the mtDNA genome and subsequently the ability to respire
and maintain cellular viability. Diploid yeast harboring the
p.Y757C heterozygous variant had severely reduced mtDNA
copy number (about half of WT levels) and reduced respiratory
competence with a signicant increase in mtDNA damage (20).
Another study used a heteroallelic yeast strain harboring both
the WT POLG (MIP1) and p.Y757C mutant alleles to show
that increasing dNTP pools (by overexpression of the RNR1
ribonucleotide reductase gene) suppresses mtDNA damage
and depletion (21). In Drosophila melanogaster lines, homo-
zygous for the orthologous POLG Y955C mutation (p.Y873C),
larval lethality at the third instar stage was seen, but hetero-
zygous p.Y873C ies did not show phenotypic abnormalities
or mtDNA deletions. Although heterozygous p.Y873C ies had
normal lifespans after 15 generations of intercrossing, the ies
were developmentally delayed and presented with mtDNA
depletion (20% less mtDNA compared with WT ies at the
F1 generation) and 40% reduced from WT at the F15 gen-
eration (10). Also, POLG Y955C heart-specic transgenic
overexpression in a mouse model caused cardiomyopathy with
decreased mtDNA (about half of WT levels) and mitochondrial
ultrastructural defects (22).
Barriers to studying mitochondrial dysfunction include
scarcity of patient-derived broblasts, scarcity of patient tissue
samples, and a lack of readily available human cell line models
that harbor disease-causing mutations. Many disorders have a
mitochondrial phenotypebut lack a mtDNA or nDNA
mutation suggesting many unidentied or hard to detect
exogenous factors play a role in secondary mitochondrial
dysfunction (SMD). Also, SMD is associated with many other
diseases, including fatty acid oxidation disorders, limb-girdle
muscular dystrophy, myopathy, drug-induced peripheral neu-
ropathies, spinal muscular atrophy, cancer, and many others
(23). We suspect that a human cell line model harboring a
mitochondrial disease mutation will have enhanced sensitivity
to mitochondrial stressors, thereby allowing the detection of
agents that otherwise may be undetectable using standard WT
cell lines.
Here we engineered a knock-in of the POLG Y955C het-
erozygous substitution into the nuclear genome of the human
SJCRH30 cell line using CRISPR-Cas9. The SJCRH30 cell line
is derived from the tumor of a 17-year-old male with rhab-
domyosarcoma. SJCRH30 cells harbor attenuated sarcomere
structures resembling those found in primitive rhabdomyo-
blasts (24). SJCRH30 has been used to evaluate the cytotoxicity
of chemotherapeutic drugs such as cisplatin, doxorubicin, and
topotecan (25,26). Also, SJCRH30 has been used as a model of
human myoblasts to study the regulation of mitochondrial
biogenesis and cellular oxygen consumption rates (27,28). The
effects of the POLG Y955C mutation on cell growth, mtDNA
maintenance, OXPHOS machinery subunits, cellular bio-
energetics, and sensitivity to the known mitochondrial toxicant
20-30-dideoxycytidine (ddC) were determined.
Results
Generation of the SJCRH30 POLG Y955C cell line
Before gene editing, studies of genetic disease mutations
relied on primary cells, and assessment of patient pathophys-
iology can require post mortem or in vivo measurement such
as MRI. Therefore, our goal was to engineer a cell line model
harboring the POLG Y955C variant to have an unrestricted
number of cells to understand better the cellular and molec-
ular impacts of this mitochondrial disease mutation. The
SJCRH30 cell line was transfected with an in vitro transcrip-
tion (IVT) guide RNA (gRNA), the GeneArt TrueCut Cas9 V2
nuclease, and a donor single-stranded DNA (ssDNA) oligo-
nucleotide harboring the POLG Y955C mutation as described
under Experimental Procedures. Isolated clones were screened
A human cell line model of mitochondrial disease
2J. Biol. Chem. (2022) 298(8) 102196
by cell lysis plus PCR amplication of a specicanking POLG
Y955C region, and the PCR products were sequenced by
Sanger sequencing. A candidate clone from the Sanger
sequencing analysis was subjected to on-target next-genera-
tion sequencing (NGS) analysis. NGS conrmed a single clone
with 50% homology-directed repair (HDR), indicating that the
clone harbored a heterozygous POLG Y955C mutation.
Following the expansion of the clone, the mutation was
reveried by on-target NGS and Sanger sequencing, Figure 1A.
Maintenance of the Y955C heterozygous locus was checked
and veried again at the end of the study using Sanger
sequencing.
Flow cytometry of propidium iodidestained cells helps
measure cellular DNA and segregates cells into their cell cycle
phases (29,30). Therefore, we performed cell cycle ow
cytometry analysis on POLG Y955C cells and determined that
the cell cycle stages and ploidy were like that of the SJCRH30
parental cell line, Fig. S1. Our results support that a POLG
Y955C heterozygous mutation was sufciently inserted into
the SJCRH30 nuclear genome.
Figure 1. Substitution of the p140 tyrosine 955 amino acid residue for cysteine (Y955C) likely disrupts a key intramolecular interaction between
the palm and ngers subdomains. A,left, model of the p140 Y955C variant. The p140 DNA polymerase domain folds to resemble a right handcomposed
of the palm (red), ngers (blue), and thumb (magenta) subdomains. Other regions include the aminoterminal domain (NTD; cyan), the spacer domain
(yellow), and the exonuclease domain (orange). Note that the Y955C substitution localizes to the ngers subdomain. The model was generated using the
PDB ID 4ZTZ structure as a template in Missense3D (100). Top right, Sanger sequencing traces of both strands of a POLG exon 18 PCR product conrming the
heterozygous POLG c.2864A>G/p.Y955C mutation. Middle right and bottom right, wildtype (WT) and p140 Y955C active sites, respectively. D890 and D1135
are key catalytic residues that interact with the dCTP (dC, colored pink)-magnesium (green spheres) complex. R943, K947, and Y951 of the O-helix (residues
943955) coordinate dC at the active site. Y955 is vital for effective dNTP binding and the pronounced template bending in the single-/double-strand
junction. Y955 is predicted to make an intramolecular hydrogen bond with E895, and this interaction is lost in Y955C. Nitrogens are colored blue, oxygens
light red,sulfur yellow, and phosphates orange.B, linear representation of the p140 catalytic subunit amino acid residues. Colors are as described in A.
A human cell line model of mitochondrial disease
J. Biol. Chem. (2022) 298(8) 102196 3
The p140 p.Y955C variant is predicted to disrupt a key
intramolecular interaction
Based on the previously published crystal structure of Polγ
bound to a primer-template, Y955 is located at the end of an
alpha helix, which also contains the crucial Y951 residue that
stacks with the incoming nucleotide, Figure 1A. Y955, along
with R943 and K947, assist with facilitating the effective
binding of an incoming nucleotide into the p140 active site.
Furthermore, the p140 Y955 residue is needed for the bending
of DNA located in a single double strand junction. This
bending is mediated by Y955 and a short connecting loop that
contains amino acid residues A957 and G958 (15). The Y955
and E895 side chains are in a close 2.3 Å proximity to each
other. We predict the Y955 side chain hydroxyl group forms a
critical intramolecular hydrogen bond with the E895 side chain
carbonyl group. Y955 is in the ngers subdomain of p140
(Fig. 1B), while E895 is located in the palm subdomain. Y955C
is predicted to disrupt the close 2.3 Å interaction between
residues at 955 and 895, resulting in the distance between
C955 and E895 increasing to 6.9 Å in the model. We expect
that disruption of this hydrogen bond is detrimental to Polγ
and results in a destabilization of the active site conformation,
which prevents the efcient incorporation of nucleotides into
nascent mtDNA.
The SJCRH30 POLG Y955C cell line has a slow growth
phenotype
The doubling times of the SJCRH30 POLG WT and
Y955C cell lines were calculated from the cell counts moni-
tored over 5 days to determine whether the POLG Y955C
mutation affects cell growth (Fig. 2A). The mean doubling
times of the WT and Y955C mutant were 35.5 ± 3.6 and 45.7 ±
5.6 h, respectively, and the decreased growth rate for Y955C is
signicantly different, Figure 2B. By day 7, the average Y955C
viable cell densities remaining on the tissue culture dishes
were 3.1-fold less than WT (Fig. 2C). Also, both cell types
maintained >95% cellular viability over the 7-day timeline as
judged by the trypan blue exclusion method.
The SJCRH30 POLG Y955C cell line has compromised
bioenergetic parameters
To determine the downstream effects of the POLG Y955C
mutation on SJCRH30 mitochondrial bioenergetic proles
POLG WT and Y955C cells were exposed to known pharma-
cological stressors of the OXPHOS machinery. Mitochondrial
OXPHOS is the oxygen-dependent process of coupling sub-
strate oxidation to produce the energy-rich molecule adeno-
sine triphosphate (ATP). The 37 mtDNA genes are required
for OXPHOS, and mitochondrial translation of the 13
mtDNA-encoded polypeptides generates essential OXPHOS
machinery subunits. During OXPHOS, molecular oxygen (O
2
)
is reduced to water (H
2
O). Using the Seahorse XFp extracel-
lular ux analyzer, O
2
biosensors measure the real-time rate at
which cells convert O
2
to H
2
O, the O
2
consumption rate
(OCR). The second set of XFp biosensors measure the extra-
cellular acidication rate (ECAR) resulting from the cyto-
plasmic breakdown of glucose-derived pyruvate to lactate and
the respiratory evolution of carbon dioxide (CO
2
). Glycolysis is
the major cytosolic O
2
-independent metabolic pathway that
converts one glucose molecule into two molecules of pyruvate,
ATP, and NADH. When pyruvate is shunted through the
mitochondrion to the pyruvate dehydrogenase complex, and
subsequently through the tricarboxylic acid cycle, CO
2
is
generated. In solution, a molecule of CO
2
can combine with a
molecule of H
2
O forming carbonic acid that dissociates at
physiological pH into the bicarbonate anion and a proton that
contributes to medium acidication. In a Mito Stress test,
oligomycin, carbonyl cyanide p-triuoromethoxyphenylhydra
zone (FCCP), and rotenone + antimycin A are sequentially
injected to inhibit ATP synthase (OXPHOS complex V),
dissipate the mitochondrial proton motive force, and inhibit
OXPHOS complex I + III, respectively. Six Mito Stress test
bioenergetic parameters are determined using these four
stressors. Rotenone plus antimycin A is added last during the
experiment to terminate electron ow through the electron
transport chain and enable calculation of the OCR from
nonmitochondrial oxidases (nonmitochondrial respiration).
This nonmitochondrial OCR is subtracted to accurately
Figure 2. SJCRH30 POLG Y955C cells grow more slowly than wildtype cells under standard tissue culture conditions. A,POLG Wildtype (WT) and
Y955C growth curves. Exponential growth was observed for both cell types. After 5 days of growth, the number of viable WT cells was 2.5-fold higher than
the mutant. Mean viable cells/cm
2
values (n = 4) and standard deviation (SD) errors for a representative experiment (P17 for WT and P8 for Y955C) are
shown in the graph. B, doubling time of POLG WT and Y955C cells. The mean doubling times (DT) based on three independent experiments utilizing WT at
passages 13, 15, and 17, and Y955C at 8, 10, and 11 are reported in hours with error as SDs. The mean doubling time of WT cells is signicantly less than
Y955C cells, ****p<0.0001, as judged by a Studentsttest. DT values were calculated using the least-squares t of the exponential growth equation in
Graph Pad Prism. C,POLG WT and mutant cell densities (cells/cm
2
) up to 7 days of growth. Data are from the same representative experiment shown in A.
A human cell line model of mitochondrial disease
4J. Biol. Chem. (2022) 298(8) 102196
determine mitochondrial contributions to basal respiration
(baseline OCR), proton leaklinked respiration (the remaining
respiration in the presence of oligomycin), and maximal res-
piratory capacity (a measure of the ability of a protonophore,
FCCP, to uncouple proton movement and ATP synthesis, and
restore ux through complexes IIV). ATP-linked respiration
measures the OCR coupled to ATP production, and spare
respiratory capacity is the difference in OCRs between
maximal respiratory capacity and basal respiration. Spare res-
piratory capacity is dened as the extramitochondrial capacity
available to produce ATP during increased work or stress
conditions (31,32). Compared with WT, Y955C has signi-
cantly decreased mitochondrial function as indicated by de-
creases in basal respiration (16% reduced), ATP-linked
respiration (21% reduced), and maximal respiratory capacity
(15% reduced), Figure 3A. Also, Y955C nonmitochondrial
respiration was signicantly decreased by 23%.
Interestingly, enhanced POLG Y955C ECAR relative to
WT was observed during all stages of the Mito Stress test
(see the bottom panel of Fig. 3C). ECARs increased in both
cell types following oligomycin injection, suggesting the
glycolytic pathway compensates for the block at OXPHOS
complex V. Also, at later time points, the slight but signi-
cant increase in ECAR values in both cell types following
injection of rotenone plus antimycin A suggests the
enhanced ECAR results from the glycolytic pathway and not
from the tricarboxylic acid cycle activity (bicarbonate-asso-
ciated acid production), i.e., compare the differences between
the 12th and 9th time points in both cell types at 75 and
55 min, p<0.0001 for WT (11.0 ± 1.3 and 9.0 ± 0.9 mpH/
minute/μg cellular protein, respectively) and p<0.0014 for
Y955C (15.4 ± 1.5 and 13.5 ± 1.5 mpH/minute/μg cellular
protein, respectively) (31,32). Therefore, the glycolytic
pathway is likely active in both cell types, and we hypothe-
sized that glycolysis is upregulated in POLG Y955C to
compensate for compromised OXPHOS.
To further investigate the difference in ECARs observed
between the two cell types, we conducted ECAR Stress tests.
ECAR Stress tests sequentially inject glucose, oligomycin, and
2-deoxyglucose (2-DG) into a glucose-free medium, bathing
WT or Y955C cells. Following glucose injection, oligomycin
inhibits the OXPHOS machinery and allows estimation of
ECAR parameters. As a competitive hexokinase inhibitor, 2-
DG inhibits the cells ability to utilize free glucose to
generate pyruvate via glycolysis, and ECAR drastically de-
creases (33). Basal ECAR is the last of three ECARs measured
immediately before sugar injection (15 min into the ECAR
Stress test). Glucose-stimulated ECAR is the maximal rate
following glucose injection (but before oligomycin injection)
minus basal ECAR. Glucose-stimulated ECAR represents the
total extracellular acidication from cellular pathways metab-
olizing free glucose, including the production of cytoplasmic
lactate and mitochondrial protons produced via CO
2
hydra-
tion and dissociation. However, as we observed an increase in
ECAR values in both SJCRH30 cell types following oligomycin
and rotenone + antimycin A injections in Mito Stress tests,
glucose-stimulated ECAR can serve as a proxy for glycolysis
ECAR.ECAR capacity (glycolytic capacity) is a measurement
of extracellular acidication, including glucose-stimulated
ECAR and oligomycin-stimulated ECAR production. The
apparent glycolytic reserve parameter is obtained by subtract-
ing glucose-stimulated ECAR from ECAR capacity. The
apparent glycolytic reserve is the estimated amount of unused
glycolytic capability of the cell that could be utilized if cellular
ATP demand was increased. In the absence of glucose, POLG
Y955C cell basal ECAR was 30% lower than WT. Following
glucose injection, Y955C glucose-stimulated (glycolytic) ECAR
increased 1.9-fold compared with WT, suggesting that mutant
cells rely more heavily on the glycolytic pathway to generate
ATP. In addition, Y955C ECAR capacity was 1.2-fold higher
than WT when OXPHOS is shut down by oligomycin. Still the
apparent glycolytic reserve capacity was only about half of WT
levels indicating a reduced glycolytic capability with increased
energy demand, Figure 3,Band D.
As skeletal muscle cells heavily rely on anaerobic glycolysis
to generate ATP during muscle contraction (34), we per-
formed Mito Stress tests following cell growth in a galactose-
based medium to force the cells to rely on mitochondrial
OXPHOS (35) and to more clearly understand the effect of the
POLG Y955C mutation on mitochondrial bioenergetics.
Indeed, when compared with cells grown in glucose (Fig. 3),
galactose-grown POLG Y955C cells have >2-fold reductions in
basal respiration (41% reduced compared with WT), maximal
respiratory capacity (33% reduced), and ATP-linked respira-
tion (43% reduced), Figure 4. The nonmitochondrial respira-
tion changes were similar in the two experiments (19%
reduced in galactose-grown Y955C cells versus 23% reduced in
glucose-grown cells). Still, a 30% decrease in Y955C proton
leak is apparent under galactose conditions. Furthermore, the
signicantly higher Y955C Mito Stress test ECARs observed in
the glucose-grown cells (Fig. 3) are not seen in the galactose-
grown cells (Fig. 4), supporting the concept that growth in
galactose obligates OXPHOS function. Based on these results,
experiments performed in glucose-grown SJCRH30 cells can
omit the effects of the POLG Y955C mutation on proton leak
and weaken the fold-change results of mitochondrial
bioenergetics.
OXPHOS complex I subunit protein expression levels are
decreased in POLG Y955C cells
As mentioned earlier, POLG Y955C is associated with
mtDNA depletion and deletions in patient skeletal muscle
samples (912). Therefore, we hypothesized that, in a POLG
Y955C cell, the mitochondrial OXPHOS machinery will be
disrupted or depleted as 13 essential OXPHOS subunits are
encoded by the mtDNA. Indeed, using samples from post-
mortem patients with POLG-related disorders, substantia
nigra neurons have been demonstrated to have decreased
expression of OXPHOS complex I or IV subunits or both, e.g.,
1. POLG p.Gly848Ser and p.Ser1104Cys, 2. p.Trp748Ser and
p.Arg1096Cys, 3. p.Ala467Thr and p.Trp748Ser, and 4.
p.Thr251Ile and p.Ala467Thr (36). Interestingly, a nuclear-
encoded complex I subunit was used to measure the
A human cell line model of mitochondrial disease
J. Biol. Chem. (2022) 298(8) 102196 5
Figure 3. SJCRH30 POLG Y955C cells have altered bioenergetics. A, results from Mito Stress tests comparing SJCRH30 POLG wildtype (WT) and Y955C
mutant cells side-by-side on XFp cell culture miniplates. Top, Description of mitochondrial bioenergetic parameters: basal respiration (Basal resp.), ATP-
linked respiration (ATP-linked resp.), maximal respiratory capacity (Max. resp. cap.), spare respiratory capacity (SRC), proton leak, and nonmitochondrial
respiration (Non-mito resp.). Metabolic stressors are injected sequentially from Ports A (oligomycin), B (FCCP), and C (antimycin A + rotenone). Bottom, Mito
Stress test bioenergetic parameters. B, results from ECAR Stress tests comparing WT and Y955C mutant cells side-by-side on XFp cell culture miniplates. Top,
Description of ECAR bioenergetic parameters: basal ECAR, glucose-stimulated ECAR (Glu. Stim.), ECAR capacity (ECAR Cap.), and apparent glycolytic reserve
(Glycolytic Reserve). First, glucose was injected from port A, followed by oligomycin (port B), and 2-deoxyglucose (port C). Bottom, ECAR Stress Test bio-
energetic parameters. Complete OCR and ECAR proles for WT and Y955C C, Mito Stress and D, ECAR Stress Tests. Data in the graphs and scatter plots are
presented as mean ± SD, n = 15 (in triplicate or sextuplicate from three independent experiments using different passages). For several data points in the
A human cell line model of mitochondrial disease
6J. Biol. Chem. (2022) 298(8) 102196
decreased expression (NDUFB8) of the complex, suggesting
that disruption of expression of one or more of the mtDNA-
encoded complex I subunits in POLG-related disease (ND1,
ND2, ND3, ND4L, ND4, ND5, ND6) destabilize the complex
leading to destabilization or proteolysis and degradation of the
respirasome and supercomplexes. The respirasome and
supercomplexes are higher-order supramolecular structures
composed of OXPHOS machinery enzyme complexes (37,38).
Therefore, to determine whether the OXPHOS machinery is
disrupted in POLG Y955C cells, we isolated mutant and POLG
WT mitochondria and then analyzed a subunit from each
OXPHOS complex via Western blotting. Using this approach,
an 2-fold decrease in expression of the NADH dehydroge-
nase complex I NDUFB8 subunit was detected in the Y955C
mutant, but the other complex subunits remained at WT
levels, Figure 5. Although the expression level of the Y955C
mtDNA-encoded cytochrome c oxidase complex IV COX2
subunit was not different from WT (Fig. 5A), we speculated
that disruption of expression of a mtDNA-encoded complex I
subunit(s) might lead to destabilization and degradation of the
nuclear-encoded NDUFB8 subunit. The protein expression
level of the ND3 subunit was investigated to test this idea.
top panels in A and B, errors are not shown as the error bars are shorter than the height of the symbol. Oligo., oligomycin; Rot., rotenone; A.A., antimycin A;
Gluc., glucose; 2-DG, 2-deoxyglucose. ****p0.0001, ***p0.001, **p0.01, and *p0.05. Statistical signicance between two parametric groups was
determined using a Students or a Welchsttest, while signicance between two nonparametric groups was determined using a MannWhitney U test.
Figure 4. SJCRH30 POLG Y955C mitochondrial bioenergetic deciencies are enhanced when galactose is substituted for glucose in the growth
medium. Results from Mito Stress tests comparing POLG wildtype (WT) and Y955C mutant cells are shown. Data in the graphs and scatter plots are
presented as mean ± SD, n 10 (from three independent experiments using different passages). For several data points in the top left panel, errors are not
shown as the error bars are shorter than the height of the symbol. Abbreviations are as described in Figure 3. ****p0.0001. Statistical signicance was
determined using Studentsttests.
A human cell line model of mitochondrial disease
J. Biol. Chem. (2022) 298(8) 102196 7
Indeed, Y955C mitochondrial ND3 was 1.6-fold decreased
compared with WT, Figure 5,B, and C.
Restriction enzyme mapping shows POLG Y955C mtDNA
harbors expected fragment sizes
Southern blotting and mtDNA-specic probes to visualize
restriction endonuclease digestion products are powerful tools
for identifying mtDNA deletions and replication intermediates
(39,40). The two mtDNA strands are named heavy (H) and
light (L) based on the ability to separate them on denaturing
cesium chloride gradients (5,6). The origin of H-strand DNA
replication (O
H
) is located in the noncoding control region,
and the origin of L-strand replication (O
L
) is situated at
11,000 base pairs downstream of O
H
(4). O
H
and O
L
divide
the circular mtDNA genome into the major arc, which harbors
the NADH dehydrogenase subunit 4 gene (ND4), and the
minor arc, which contains the ND1 gene. Many mtDNA de-
letions localize within the mtDNA major arc (4143). To
determine whether the SJCRH30 POLG Y955C cell line har-
bors mtDNA deletions, WT and Y955C mtDNA restriction
maps were analyzed using key restriction enzymes and two
single-stranded (ss) oligonucleotide probes, ND4 and ND1.
PvuII cuts once in the mtDNA genome and linearizes cova-
lently closed circular mtDNA to a discrete band of 16.6 kb,
Figure 6A. Both BamHI and NheI cut once within the cova-
lently closed circular ds-mtDNA, and a double digest reaction
with these enzymes generates 9.7- and 6.9-kb linear frag-
ments. Also, XbaI was used and cuts 5 times in the mito-
chondrial genome creating ve linear fragments that are 1.8,
4.5, 0.9, 2.0, and 7.5 kb in size (note, only four XbaI cut
sites are shown in the map in Fig. 6Bto emphasize the bands
detected with the two mtDNA-specic probes).
As mtDNA molecules harboring deletions still usually har-
bor the ND1 gene (42), an ND1 minor arcspecic probe
should detect molecules harboring deletions when linearized
with a restriction endonuclease. The expected WT fragment
lengths were detected in SJCRH30 POLG WT and Y955C
using the aforementioned enzymes and the digoxigenin (DIG)-
labeled ss-oligonucleotide H-strand probe targeted to ND1.A
single high-molecular-weight (HMW) mtDNA band was
observed for PvuII-digested DNA samples, which agrees well
with previous results for the human cell line HepaRG (44).
When POLG WT and Y955C whole-cell extracted (WCE)
DNA samples were separately digested with BamHI plus NheI,
a major band with a calculated molecular weight (MW) of 7.1
kb was observed in both, which is in good agreement with the
expected length of the BamHI/NheI fragment containing the
region for the ND1 probe, 6.9 kb (Fig. 6,Aand C). Interest-
ingly, using the ND1 probe, we detected a couple of minor
intensity bands at 6.3 and 4.9 kb in the WT BamHI/NheI
double digest (expected lengths of 6.5 and 4.4 kb, respectively).
The minor intensity bands may represent replication in-
termediates resulting from stalling of the replisome during
strand-displacement mtDNA. Stalling of Polγduring coun-
terclockwise lagging-strand synthesis (nascent L-strand) at the
predicted G-quadruplex (G4) located at 4229 to 4260 bp, fol-
lowed by a strand break in the template H-strand, could result
in the minor intensity BamHI/NheI digest band with the ex-
pected 6.5 kb length, Figure 6B. A previous study showed that
mtDNA G4 forming sequences are associated with mtDNA
deletion breakpoints suggesting a role for G4s in perturbation
of mtDNA replication by fork stalling (45,46). In addition, G4s
are associated with nuclear genome instability and gene
expression defects (47). G-quadruplexes are noncanonical
secondary structures formed by planar stacking of four
Figure 5. Mitochondrial NADH dehydrogenase complex I NDUFB8, and
ND3 levels are decreased in POLG Y955C. A, Y955C NDUFB8 expression is
decreased compared with WT. POLG wildtype (WT) and Y955C mitochon-
drial extracts were run on SDS-PAGE, followed by Western blot analysis. An
antibody cocktail containing monoclonal antibodies specic for a subunit of
each OXPHOS complex (I, II, III, IV, and V) was used. Four of the subunits are
encoded by the nuclear genome (ATP5A, UQCRC2, SDHB, and NDUFB8), and
one is encoded by the mtDNA genome (COX2). A representative blot is
shown. B, in comparison with WT, Y955C ND3 expression is decreased. POLG
WT and Y955C mitochondrial (Mito.) and whole-cell (Cell) extracts were
subjected to Western blot analysis as described in A, but a mtDNA encoded
ND3-specic polyclonal primary antibody was used. The trichloroethanol
(TCE)stained whole-cell or whole-mitochondrial protein in each lane of the
blot was used to normalize the data for the chemiluminescent bands, and a
representative blot is shown. C, relative expression levels of NDUFB8 and
ND3 in mitochondrial protein extracts. The chemiluminescent band areas in
each lane were normalized to their respective total protein signal on the
blot. The WT values were set to 100%, and the data are mean ± SDs. For the
NDUFB8 subunit, the experiment was repeated twice on different days (n =
8 lanes from two blots), and for the ND3 subunit, the experiment was
repeated in triplicate on different days (n = 12 from six blots; two blots per
experiment). ****p0.0001.
A human cell line model of mitochondrial disease
8J. Biol. Chem. (2022) 298(8) 102196
Figure 6. Restriction endonuclease mapping of SJCRH30 POLG WT and Y955C mtDNA. A, detection of mtDNA restriction fragments with either an ND1
heavy (H)-strand (left)orND4 H-strand (right) single-stranded (ss) DIG-labeled probe. B, maps of the expected mtDNA restriction fragment lengths when
visualized with either the ND1 or ND4 probe. Panel 1 shows expected fragment lengths produced with BamHI/NheI double digestion, and panel 2 shows
expected lengths produced with XbaI digestion. In both panels, the mtDNA light (L) strand is represented as a thin intact inner circle and the H-strand origin
of replication (O
H
), the L-strand origin of replication (O
L
), and the ND1 (*) and ND4 (#) probe sequences are also highlighted on this strand. Note that the
numbering of base pairs is counterclockwise and is based on mtDNA NC_012920. Nascent continuous leading H-strand synthesis, using the L-strand
template and beginning at O
H
(position 191), is represented as a thick dashed line with replication proceeding clockwise around the circle. The parental/
template H-strand is shown as a displaced intact, thick circle that is used as the template strand during continuous nascent lagging L-strand synthesis, which
is represented as a thin dashed line beginning at O
L
(position 5770) and proceeding counterclockwise. The location of two predicted G-quadruplexes (G4s)
are shown in panels 1 and 2 as black circles above the parental H-strand, 2GQH located at positions 4260 to 4229, and 3GQH located at 15545 to 15516.
C, summary of the expected and estimated molecular weights (MW) of mtDNA bands generated by restriction endonuclease (RE) digestions. PvuII-digested
mtDNA fragment lengths greater than 10 kb were estimated using linear regression of the log10 base pair values of Lambda DNA/HindIII Marker 2
fragments loaded onto the same gel versus distance traveled in millimeters (R
2
values were 0.94). For the remaining restriction digests that generate
fragment lengths less than 10 kb, linear regression of the log10 base pair values of exACTGene 1 kb Plus DNA ladder fragments versus distance traveled in
millimeters was utilized to estimate molecular weights (R
2
values were 0.99).
A human cell line model of mitochondrial disease
J. Biol. Chem. (2022) 298(8) 102196 9
guanines called a G-tetrad, and two or more G-tetrads can
stack to form a thermodynamically stable quadruplex (46,47).
Furthermore, mtDNA base pairs 4229 to 4260 localize to a
previously identied region of replication pausing found in
human tissue (brain, heart, skeletal muscle, placenta, kidney)
and cell (HEK293T, HeLa, 143B, Jurkat) mtDNA (48).
The expected BamHI/NheI 4.4- (4.9 observed) kb frag-
ment detected with ND1 could be produced if the clockwise
leading strand (nascent H-strand) Polγstalled (or just
nished H-strand synthesis) around O
H
and an L-strand/
template strand break occurs in this region. Replication
initiation of nascent leading H-strand synthesis at O
H
in-
volves a complex hybrid G4 between mitochondrial RNA
generated by mtDNA transcription and the nontemplate
H-strand (4). We predict this hybrid structure can stall or
signal the completion of Polγnascent H-strand synthesis at
O
H
resulting in the production of the minor 4.4-kb species
on BamHI/NheI digest when a template strand break occurs.
Curiously, in Y955C compared with WT, the ND1-probed
4.4-kb band is diminished and the 6.5-kb band is not
detectable. Finally, in agreement with an expected mtDNA
XbaI cleavage product of 4.5 kb, harboring the comple-
mentary ND1 probe sequence, a band with a calculated MW
of 4.8 kb was detected in both cell types.
Next, an H-strand ss-DIG-labeled oligonucleotide comple-
mentary to the major arc ND4 gene was used. A single HMW
band was observed for PvuII-digested WT and Y955C DNA
samples, which is in good agreement with what was observed
with the ND1 probe described above. When POLG WT and
Y955C WCE DNA samples were separately double digested
with BamHI and NheI and then probed with ND4, a major
band with a calculated MW of 9.4 kb was observed in both
samples and agreed with the expected length of the 9.7-kb
BamHI/NheI fragment, Figure 6. In addition, using the ND4
probe with BamHI/NheI digests, a faint band at 8.7 kb (ex-
pected MW 8.5 kb) was observed that could represent a
truncated counterclockwise nascent L-strand fragment that
lacks the NheI site due to initiation from O
L
and a break in the
template H-strand near this origin. Following XbaI digestion of
WCE DNA samples, Southern blotting, and ND4 probe hy-
bridization, we detected a major band at 7.9 kb in both the cell
types, which is in good agreement with the expected length of
the ds-mtDNA 7.5 kb probed XbaI fragment. Also, two less
intense bands were observed below the 7.9-kb band in both the
cell lines, 6.8- (6.5 expected) and 4.8- (5.3 expected) kb bands,
although the two bands were less prominent in Y955C. In the
strand-displacement model of mtDNA replication, the 6.5-kb
band can be rationalized by clockwise initiation of nascent
H-strand mtDNA synthesis at O
H
with a break in the template
L-strand in this region and cutting by XbaI at position 10,257
in the mtDNA genome. On the other hand, the expected
5.3-kb band could represent a species initiated from counter-
clockwise mtDNA synthesis at O
L
with Polγstalling at the
H-strand/template strand G4 located at position 15,545 to
15,516. In this scenario, the 5.3-kb band requires a single-
strand break near the H-strand G4 sequence to be visualized
upon XbaI digestion. We obtained similar results with
ss-DIG-labeled oligonucleotide L-strand probes localizing to
ND1 and ND4 (Fig. S2).
To further rule out the possibility that the minor bands
detected with the ND1 and ND4 probes (bands 3, 4, 7, 9, and
10) resulted from mtDNA deletions, we used long-range PCR
to screen for truncated deletion products. We did not detect
signicant deletions in the WT or Y955C cell lines. Overall,
these data support that the minor intensity bands of lower-
than-expected molecular weights predominantly result from
single-strand breaks of replication intermediates initiated at
O
H
or O
L
and replisome stalling events at G-quadruplexes.
SJCRH30 POLG Y955C cells have fewer nucleoids compared
with wildtype cells
PicoGreen is a simple and effective dsDNA-specic probe
used to image the tightly packed mtDNA nucleoid in living
cells (49,50). These nucleoprotein complexes are visualized as
cytoplasmic foci or puncta that colocalize with mitochondria
using uorescence microscopy (51). Live cell mitochondria are
effectively detected using the MitoTracker Red uorescent
probe (51). PicoGreen was employed to label total cellular
dsDNA (mtDNA + nDNA) with green uorescence, and
MitoTracker Red was used to label mitochondria. PicoGreen-
stained mtDNA nucleoids (puncta) localizing to MitoTracker
Redstained mitochondria were counted in the POLG WT and
mutant cells. Relative to focal planes containing SJCRH30
POLG WT cells, POLG Y955C cells have half the PicoGreen
stained nucleoids (p<0.0001), Figure 7,Aand B.
Mitochondrial morphology was assessed in 111 cells of
each type using the Mitochondrial Analyzer plugin for ImageJ/
Fiji (52). Compared with POLG Y955C mutant cells, WT cells
have 1.5- to 2-fold increases in mitochondrial area, counts,
perimeter, branches, branch lengths, branch endpoints, and
branch junctions, Table 1. This image analysis agrees with the
1.7-fold increase in mitochondrial protein level per cell
detected in our WT mitochondrial extract compared with the
Y955C mutant (see Isolation of mitochondria under
Experimental procedures). In addition, an increased level of
WT mitochondrial branch endpoints (where branches end
without connecting to another branch) helps explain the
notable WT tubular-shaped mitochondria compared with the
more clustered appearance of the Y955C organelles, Figure 7A.
The POLG Y955C mutation causes decreased mtDNA copy
number
Compared with the parental POLG WT cells, the reduced
number of POLG Y955C mtDNA nucleoids and stalled repli-
cation intermediates revealed by restriction endonuclease
mapping supports that the mtDNA copy number is lower in
the mutant. Therefore, to determine whether the POLG Y955C
mutation causes mtDNA depletion, the mtDNA copy number
was investigated using our relative copy number method (53).
The WCE DNA samples were prepared from cells obtained on
days 3, 5, and 7 post seeding as described under Experimental
procedures. The levels of mtDNA in each lane of the blot were
detected with the mtDNA-specic probe and were normalized
A human cell line model of mitochondrial disease
10 J. Biol. Chem. (2022) 298(8) 102196
Figure 7. SJCRH30 POLG Y955C cells contain less mtDNA and fewer mtDNA nucleoids than POLG wildtype cells. A,POLG wildtype (WT) and
Y955C cells were separately dual stained with MitoTracker Red and PicoGreen dsDNA Reagent, and live-cell images were collected on a uorescent mi-
croscope. Three POLG WT and four Y955C PicoGreen-stained nuclei are labeled N,and three PicoGreen-stained mtDNA nucleoids are emphasized with
white arrows in both cell types. For clarity, the PicoGreen (green) and MitoTracker Red (red) channels are shown separately in grayscale and together in color
in the merged images. Arepresentative image for each cell type is shown. Scale bars represent 10 μm. B, mitochondrial nucleoids from n = 64 total cells
were counted for each cell line and from two different experiments (n = 32 cells for each) using different passages; ****p<0.0001. C, BamHI-digested whole-
cell extracted DNA samples were analyzed via Southern blot and nonradioactive probe hybridization. The blots were simultaneously probed with the DIG-
labeled 18S nDNA probe (N, lower panels; nucleotide positions 101600) and the mtDNA-specic probe (MT, upper panel; nucleotide positions 168606).
Bands were quantitated using the open-source image-processing package Fiji, as described (44,53). A representative blot is shown. On each blot, the
average normalized band intensity values of WT mtDNA relative to nDNA on day 7 were set to 100%, and all other samples were compared with it.
Statistical signicance was determined by a two-way ANOVA, n = 12 (quadruplicates from three experiments using different passages, a total of six blots
were analyzed, two per experiment); ****, ####, and $$$$ p<0.0001; ***p<0.001; ## and **p<0.01.
Table 1
Comparison of SJCRH30 POLG wildtype and Y955C mitochondria
Measurements normalized per cell
a
SJCRH30 POLG wildtype
b
SJCRH30 POLG Y955C
b
Fold-change P-value
c
(fold-change)
Mitochondria (count) 54.2 ± 5.1 37.0 ± 3.7 1.5 <0.002
Mitochondrial total area (μm
2
) 54.7 ± 4.9 33.9 ± 6.1 1.6 <0.002
Total perimeter of mitochondria (μm) 257.0 ± 23.5 156.7 ± 17.8 1.6 <0.0005
Mitochondrial branches 80.8 ± 7.6 50.1 ± 4.6 1.6 0.0005
Total mitochondrial branch lengths (μm) 86.2 ± 9.3 48.5 ± 10.7 1.8 <0.002
Mitochondrial branch endpoints 109.9 ± 11.1 72.5 ± 5.6 1.5 <0.001
Mitochondrial branch junctions 14.3 ± 1.9 7.2 ± 3.3 2.0 <0.01
a
Live-cell MitoTracker Red CMXRos staining, uorescence microscopy, and imaging were carried out as described under Experimental procedures. Measurements were made
using The Mitochondrial Analyzer plugin for ImageJ/Fiji and data are normalized to values per cell (see MitoTracker Red CMXRos-stained mitochondria measurements).
b
Data are mean ± SD, n = 4 images with 111 total cells (two digital photographs from independent experiments performed on different days with different preparations of cells).
c
p-Values <0.05 were accepted as signicantly different.
A human cell line model of mitochondrial disease
J. Biol. Chem. (2022) 298(8) 102196 11
to nDNA levels seen with the 18S probe in the same lane,
Figure 7C. In the POLG WT cell line, we observed an increase
in mtDNA copy number over the 7-day experiment; therefore,
WT mtDNA on day 7 was normalized to 100%. On days 3, 5,
and 7 of growth, the SJCRH30 WT mtDNA copy number
values were 92.4 ± 5.1, 92.9 ± 5.9, and 100.0 ± 3.1%, respec-
tively. At the same time points, the SJCRH30 POLG Y955C
mutant mtDNA values were 79.5 ± 9.6, 70.3 ± 8.8, and 76.8 ±
11.9%, respectively. A two-way ANOVA was done to assess the
interactions at different time points and between the two
different genotypes (POLG WT and Y955C). Throughout the
experiment, we did not observe signicant changes in mtDNA
copy number within each genotype/group; however, signicant
differences between WT and Y955C were detected at all the
time points analyzed. When comparing mtDNA content be-
tween WT and Y955C on days 3, 5, and 7, Y955C mitochon-
drial genomes were reduced by 14.0%, 24.3%, and 23.2%
relative to WT on the same day. Therefore, mtDNA content in
Y955C is signicantly reduced relative to the WT cell line
during standard cell culture growth conditions. We suspect
that the more severe Y955C mtDNA depletion detected with
PicoGreen (50%) as compared with our Southern blotting
technique (20%) is due to the specicity of PicoGreen for
dsDNA over ssDNA as stated by the manufacturer (Thermo
Fisher Scientic) and others (50).
SJCRH30 POLG Y955C cells accumulate mtDNA replication
intermediates
One-dimensional (1D) and two-dimensional (2D) agarose
gel electrophoresis (AGE) are powerful tools used to analyze
the numerous complex mtDNA genome topological structures
that exist within a cell such as supercoiled, linear, relaxed
circles, and various catenanes (40,5456). An elegant study by
Kolesar et al.(40) demonstrated that, among different cell types
and tissues derived from humans and mice, there exist similar
major mtDNA topoisomers (e.g., catenanes, relaxed circles,
linear molecules), but the mtDNAs can be distributed differ-
ently, and additional structures can be seen depending on the
cell type or tissue. We speculated that the POLG Y955C
mutant cell line could harbor different distributions of mtDNA
topoisomers relative to the WT. To investigate mtDNA to-
pological structures, we separately isolated total cellular DNA
samples (WCE DNA) from POLG WT and Y955C cells and
subjected them to RNase A to remove RNA and reveal regions
of ssDNA. Next, the WCE DNA samples were subjected to
1D- or 2D-AGE, Southern blotting, and nonradioactive probe
hybridization to visualize mtDNA topoisomers.
First, 1D-AGE was used to resolve HMW mtDNAs, and the
species that formed discrete bands were quantitated. Cate-
nanes and relaxed circular (RC) mtDNAs were the major
topoisomer species observed in the parental POLG WT WCE
DNA. Catenanes included HMW mtDNA structures found in
the gel wells (well species), mid-range-molecular-weight
(MMW) catenanes, and low-molecular-weight (LMW) cate-
nanes, Figure 8. A smear of signal from the WT well species
down to the MMW catenanes is indicative of additional HMW
complex catenated species that exist in vivo. Two other minor
species were quantitated, a barely detectable replication in-
termediate (RI) band and a faint linear mtDNA band (Fig. S3
shows overexposed replicate blots to highlight the minor
species). We treated the WCE DNA samples with S1 nuclease,
an ssDNA-specic nuclease, to determine whether the mtDNA
structures contain ssDNA. Compared with untreated WT
samples, a 7.1-fold reduction in well species and a decrease in
the smear from the well down to the MMW catenanes were
detected following S1 treatment. These observations are in
agreement with Kolesar et al. (40). Also, following S1 treat-
ment, WT MMW and LMW catenanes decreased 3.8- and
3.4-fold, respectively, while RC and linear mtDNA increased
2.1- and 8.1-fold, respectively. These results suggest that RC
and linear species are released from the HMW molecules
following S1 treatment. To support that the S1-sensitive spe-
cies harbor ssDNA, we omitted the alkaline denaturation step
before Southern transfer (nondenaturing conditions). Alkaline
denaturation of DNA is necessary to hybridize a dsDNA with a
single-stranded probe. Therefore, omission of the denaturation
step can detect regions of ssDNA (40). Under nondenaturing
conditions, we observed potential regions of ssDNA in the WT
that include mtDNA well species, a weak smear of signal from
the well species down to the MMW catenanes, MMW cate-
nanes, LMW catenanes, and RC mtDNAs. After S1 treatment,
the nondenatured WT well species, the smear of signal from
the well species down to the MMW catenanes, and the MMW
catenanes were undetectable; the LMW catenane species were
barely detectable; and the RCs increased 1.8-fold relative to the
untreated samples but were not signicantly different from
untreated. Linear nondenatured WT mtDNAs were unde-
tectable in untreated samples but appeared after S1 nuclease
treatment suggesting these molecules originated from the
untreated nondenatured HMW species.
Strikingly, under alkaline denaturing conditions, the POLG
Y955C mtDNA RI found between the LMW catenanes and the
relaxed circles was the most abundant species on the blots and
thus was set to 100% for normalization of the data, Figure 8.
The Y955C RI was increased 143-fold relative to untreated
WT under denaturing conditions. In addition to the RI, Y955C
DNA extracts contained HMW well species, a weak smear of
signal from the well species down to the MMW catenanes,
MMW catenanes, LMW catenanes, and relaxed circles. The
HMW well species, the MMW catenanes, the LMW cate-
nanes, and the RCs were signicantly less than those found in
WT, being 8.4-, 2.6-, 7.2-, and 10.9-fold decreased, respec-
tively. Also, relative to untreated WT WCE DNA samples, a
22-fold increase in Y955C linear mtDNAs was observed, and a
unique band was detected under the linear band. Following S1
treatment of Y955C DNA extracts, the mtDNA well species,
the smear from the well down to MMW catenanes, and the
unique band under the linear mtDNA band were completely
digested, suggesting these molecules harbor signicant quan-
tities of ss-mtDNA. In addition, the MMW and LMW cate-
nated species were nearly completely digested, about half of
the RI molecules were degraded, and the linear mtDNAs
increased 2.5-fold. The decrease in the HMW well species, the
A human cell line model of mitochondrial disease
12 J. Biol. Chem. (2022) 298(8) 102196
smearing from the well down to MMW catenanes, and the
increased abundance of linear mtDNAs agree with the results
observed for WT; however, Y955C did not display an increase
in the RC molecules following S1 digestion. Relative to WT
and under nondenaturing conditions, we observed a signicant
signal for potential regions of ssDNA in the Y955C RI species
and lower levels of LMW catenane and RC mtDNAs. Another
difference from WT is that linear mtDNAs harboring potential
regions of ssDNA were detected in untreated Y955C WCE
DNA samples. Following S1 treatment, the nondenatured
Y955C MMW and LMW catenane species were undetectable,
RIs decreased 3.1-fold, and linear mtDNAs increased 6.4-fold.
These results suggest that linear Y955C mtDNAs harboring
regions of ssDNA were released from the MMW catenanes,
LMW catenanes, and RIs following digestion with S1 nuclease.
Surprisingly, the unique Y955C S1 nuclease-sensitive band
found below the linear mtDNA band under denaturing con-
ditions was not detected under nondenaturing conditions.
To further investigate the differences observed between
POLG WT and Y955C mtDNAs on 1D-AGE, we used
2D-AGE as an additional method to examine mtDNA top-
oisomers. First, WCE DNA was separated by 1D-AGE as
described above, and we cut lanes containing DNA samples
of interest out of the gels. Second, individual lanes were
Figure 8. The SJCRH30 POLG Y955C cell line harbors additional mtDNA topoisomers sensitive to S1 nuclease. WCE DNA samples were treated with
RNase A and digested with BglII to fragment nDNA (but not mtDNA). Where indicated, samples were treated with S1 nuclease (S1). Alkaline denaturation
(Denaturing) before Southern blotting is necessary for hybridization of dsDNA with a single-stranded (ss) probe but can be omitted to assess potential
single-stranded DNA species. For nondenaturing analysis (Non-denaturing), the gels were excluded from the denaturing solution step, but the other steps
were performed. The blots were probed using the DIG-labeled mtDNA-specic probe (nucleotide positions 168606). The untreated denatured samples
(without S1 nuclease) contain major mtDNA topoisomers such as catenanes (Cat.), relaxed circles (RC), linears, and supercoiled (S.C.) molecules. Catenanes
include HMW mtDNA well species (Well sp), mid-range-MW (MMW) catenanes, and low-MW (LMW) catenanes. Y955C cells accumulate a large amount of a
replication intermediate (Rep. intermed. or RI) that localizes between the LMW catenanes and RCs, is sensitive to S1 nuclease, and is barely detectable in the
WT. Under denaturing conditions, a unique species was seen below the linear mtDNAs in Y955C untreated samples and is sensitive to S1 nuclease digestion
(S1 sens, highlighted with an arrow on the blot). A BamHI-digested WT sample was run in parallel as a control for linear 16.6-kb mtDNA. Data below the
blots are mean values ± SD. Mean and SD values were calculated from n = 6 data points from two sets of blots, four in total (two denatured and two
nondenatured, all photographed on the same image and probed with the same batch/concentration of probe, a representative blot for each is shown). The
three replicates for each treatment shown are separate WCE DNA preparations from different passages of cells. The denatured and nondenatured blots
were separately analyzed, and topoisomer differences among WT untreated, WT S1 nuclease treated, Y955C untreated, and Y955C S1 nuclease treated were
determined (the rows of the table below the blots were compared). Three or more data sets per row were compared using a one-way ANOVA or a Welchs
ANOVA, while data sets of two (e.g., nondenatures MMW catenanes) were analyzed using a ttest. Identical lowercase letters within a row ((a) versus (a) or (b)
versus (b)) indicate no signicant difference in the mean level of topoisomers, while different letters represent statistically signicant differences. p<0.05 is
considered signicant, and all p-values were <0.031.
A human cell line model of mitochondrial disease
J. Biol. Chem. (2022) 298(8) 102196 13
separately caste into another gel with identical agarose con-
centration but now containing ethidium bromide (EtBr). EtBr
was added to the running buffer to enhance the rigidity of
DNA and exaggerate the separation of mtDNAs with differ-
ences in extended shape. In this scenario, mtDNA RIs move
more slowly than linear mtDNAs (57). Also, EtBr induces
further supercoiling in covalently closed or topologically
constrained mtDNAs (40). The six SJCRH30 WT mtDNA
species identied in 1D-AGE were seen by 2D-AGE, plus an
additional catenane band above the MMW catenane band
was resolved: 1. HMW mtDNA well species, 2. HMW cate-
nanes, 3. MMW catenanes, 4. LMW catenanes, 5. a barely
detectable RI, 6. RC, and 7. linear mtDNAs, Figure 9A.We
treated the WCE DNA sample with S1 nuclease to examine
mtDNA structures with ssDNA. Following treatment of DNA
extracts with S1 nuclease, the HMW mtDNA structures in
the well, the HMW catenanes, and the RI were undetectable,
while the MMW and LMW catenanes decreased in abun-
dance but were still detectable and RCs and linears increased,
Figure 9C.
Figure 9. Examination of mtDNA topoisomers by two-dimensional agarose gel electrophoresis (2D-AGE) reveals additional POLG Y955C-specic
replication intermediates. A,POLG WT, B,POLG Y955C, C,POLG WT S1 nuclease treated, and D,POLG Y955C S1 nuclease treated WCE DNA samples. The
WCE DNA samples were subjected to 2D-AGE, Southern blotting, and probe hybridization using the DIG-labeled mtDNA-specic probe (nucleotide positions
168606). For comparison, above each 2D-AGE blot is a representative 1D blot lane in the orientation needed for the 2D separation. We identied 11
different mtDNA topoisomers in the two cell types and six are shared between WT and Y955C. Numbers from 1 to 11 are 1. mtDNA well species, 2. High-
molecular-weight (HMW) catenanes, 3. mid-range-MW (MMW) catenanes, 4. Low-MW (LMW) catenanes, 5. a replication intermediate (RI) found in Y955C and
WT, 6. relaxed circles (RC), 7. linear mtDNAs, 8. a minor Y955C-specicRI(rst minor RI), 9. a Y955C-specic S1-sensitive species, 10. a second minor Y955C-
specic RI (second minor RI), and 11. a second Y955C-specic S1-sensitive species running near the linear band (second S1 sensitive species). See Fig. S4 for
another set of example blots.
A human cell line model of mitochondrial disease
14 J. Biol. Chem. (2022) 298(8) 102196
Compared with POLG WT cell mtDNAs, 2D-AGE revealed
four new minor Y955C-specic structures. The new structures
include 1. a slower-moving, less abundant Y955C-specicRI
(rst minor RI; band 8), 2. a faint Y955C-specic S1-sensitive
species (band 9), 3. a fast-moving second RI (second minor RI;
band 10), and 4. a second faint Y955C-specic LMW S1-
sensitive species that migrated near the linear band (band
11), Figures 9Band S4. In addition, a subtle horizontal range
(smear) of Y955C molecules spanning the region between the
second minor RI (band 10) and the linear band (band 7) was
revealed. The HMW catenane band 2 seen in WT cells was
undetectable in Y955C, and the well species (band 1) was
weak. Following S1 nuclease treatment, the well species and
the S1-sensitive species (bands 9 and 11) were undetectable on
the blot, the RC band 6 was about the same as the untreated
sample, the linear mtDNA band 7 became sharper, and the
other bands (35, 8, and 10) had decreased intensities.
SJCRH30 POLG Y955C DNA extracts lack detectable four-way
mtDNA junctions
Our initial mtDNA restriction endonuclease mapping sug-
gested that some WT RIs involve L/template strand breaks
near O
H
and that Y955C cells have reduced amounts of these
molecules, indicating problems with mtDNA maintenance at
O
H
,Figure 6. The end of mtDNA replication has been
demonstrated to occur via a hemicatenane formed at O
H
, and
topoisomerase 3αis essential for resolving this structure (39).
In comparison with SJCRH30 POLG WT, Y955C cells suffer
from depletion of monomeric RCs (Fig. 8), which according to
the vertebrate strand displacement model, are generated
following decatenation (58). Therefore, we hypothesized that
following completion of mtDNA synthesis topoisomerase 3α
decatenation of daughter mtDNAs may be altered in POLG
Y955C cells. To dissect this further, we used two-dimensional
neutral agarose gel electrophoresis (2DNAGE) and Southern
blotting to visualize X-form mtDNA structures that are ex-
pected to result from hemicatenated molecules generated at
O
H
. Indeed, in comparison with POLG WT, X-form dimeric
fragments joined by four-way junctions (i.e., hemicatenanes)
were undetectable in Y955C and the ascending mtDNA
replication fork arc (Y-arc) was very weak supporting our
hypothesis that decatenation at O
H
is altered in POLG Y955C,
Figure 10. Also, a strong descending Y-arc signal combined
with depleted X-forms is seen in other cases of mtDNA
replication stalling, including ddC treatment, TFAM over-
expression, or expression of catalytically defective Polγor
Twinkle. An interpretation is that the slow progression of
replication causes an accumulation of Y-forms while the
replication termination intermediates are depleted due to their
constant resolution (54,5961).
SJCRH30 POLG Y955C has increased sensitivity to
20-30-dideoxycytidine
Further complicating mitochondrial dysfunction, genetic
mutations encoding variants of DNA polymerase subunits,
which localize to the mitochondrion, could predispose patients
to mitochondrial toxicity, e.g., p140 (R964C, R953C, and
E1143D/G) and PrimPol D114N (6267). For example, in two
lymphoblastoid cell lines harboring POLG p.R964C, 10 μMof
the nucleoside reverse transcriptase inhibitor stavudine (20,30-
didehydro-20,30-dideoxythymidine) reduced mtDNA levels in
the mutant cell lines but not in WT lymphoblastoid cell lines.
Furthermore, compared with recombinant WT p140, p140
R964C had only 14% DNA polymerase activity (62). Therefore,
if the POLG Y955C mutation enhances mitochondrial toxicity,
then the mutant cell line can serve as a sensitive system to
identify unidentied or hard-to-detect mitochondrial stressors.
To test whether Y955C has enhanced mitochondrial toxicity,
we exposed SJCRH30 POLG WT and Y955C cells to a known
mitochondrial toxicant, the nucleoside reverse transcriptase
inhibitor ddC. Using various human cell lines grown in tissue
culture, ddC has been shown to induce mtDNA replication
stress and, consequently, mtDNA depletion (49,53,6873);
therefore, we expect Y955C cells will be more sensitive to ddC.
The half-maximal inhibitory concentration (IC
50
) value or
concentration of ddC that reduces the number of viable
treated cells by 50% relative to untreated control cells was
Figure 10. POLG Y955C mtDNA lacks detectable four-way DNA junc-
tions (i.e., hemicatenanes) and has a reduced replication fork signal.
WCE DNA was digested with HincII, subjected to 2DNAGE and Southern
blotting and probed with a mtDNA-specic probe (positions 37611). The
mtDNA probe is specictoan3.9-kb subgenomic mtDNA HincII restriction
fragment harboring the heavy strand origin of replication (O
H
). The sche-
matic below the wildtype and Y955C blots emphasize the 1n, 3.9-kb non-
replicating HincII fragment (large black circle); the ascending and
descending parts of the Y arc (mtDNA replication fork) represented as Y and
Y0, respectively; and X-form molecules (X), dimeric fragments joined by four-
way junctions.
A human cell line model of mitochondrial disease
J. Biol. Chem. (2022) 298(8) 102196 15
determined. Indeed, POLG Y955C cells are more sensitive to
ddC, as indicated by a 5.4-fold reduction in the mutant IC
50
value relative to WT, Figure 11,AC. However, after exposing
cells to 1 μM ddC and monitoring mtDNA depletion over
6 days, the depletion rates were not signicantly different in
the two cell types, Figure 11D. Also, we monitored mtDNA
replication following a 24-h treatment of cells with ddC.
Interestingly, when the WT and mutant cells were seeded at a
high density, mtDNA copy number continued to decline after
the removal of ddC. The mtDNA genome levels slightly
increased from 96 to 144 h (3 to 5 days post treatment,
respectively) in both cell types, Figure 11E. At 144 h, the WT
and Y955C mtDNA copy numbers were 1.6- and 1.5-fold
higher than at 96 h, respectively. However, only the 1.6-fold
increase in WT mtDNA was signicantly increased (p<
0.002 WT, p= 0.07 Y955C), suggesting recovery of mutant
mtDNA replication is impaired.
Discussion
Greater than 400 genes (mtDNA + nDNA) are currently
linked to mitochondrial disease, and many additional disorders
and nongenetic factors are associated with SMD (74). In
addition, greater than 300 pathogenic mutations localize to the
nuclear POLG gene alone and are associated with different
types of mitochondrial disease, including mtDNA depletion
and deletion disorders (75,76). To better understand how a
POLG mutation affects mitochondrial function at the cellular
and molecular levels, we constructed a human cell line model
harboring a knock-in of the most common autosomal domi-
nant POLG mutation, Y955C. Another goal was to understand
if the mutation confers increased sensitivity to the well-known
mitochondrial stressor, ddC.
In agreement with previous reports of reduced mtDNA
content in POLG Y955C adPEO patient skeletal muscle
samples (10), patient-derived broblast culture (11), and
model systems harboring the orthologous mutation including
yeast (20), Drosophila (10), and mouse (22)weobserved
decreased mtDNA content in SJCRH30 POLG Y955C. How-
ever, in agreement with other studies discussed above, we only
observed an 14% to 24% decrease in Y955C mtDNA copy
number relative to WT over 7 days, Figure 7. These ndings
suggest that sufcient expression of the mutantsWTPOLG
locus provides a necessary stoichiometry of p140 catalytic
subunits in vivo or another DNA polymerase localizing to
the mitochondrion assists the WT enzyme in replicating
Figure 11. SJCRH30 POLG Y955C cells have increased sensitivity to ddC. In A, wildtype (WT) passage 16 and B, Y955C passage 12 experiments, cells
were exposed to 128, 8, 4, 2, 1, 0.5, 0.25, and 0 μM ddC for 9 days and mean percent survival values (n = 4) and SDs are reported for each representative IC
50
curve. C,IC
50
values were calculated in quadruplicate from three independent experiments (n = 12) utilizing WT at passages P10, 12, and 16, and Y955C at 6,
8, and 12. IC
50
values were calculated using the least-squares t of inhibitor concentration versus normalized response in Graph Pad Prism, and mean values
are reported in micromolar (μM) with error as SD. D, the relative rate of mtDNA depletion is similar in WT and Y955C cells exposed to 1 μM ddC. The rate of
mtDNA depletion in WT and Y955C was determined by preparing WCE DNA samples from cells separately exposed to ddC followed by BamHI digestion,
Southern blotting, and dual DIG-labeled probe detection as outlined in Figure 7. On each blot and for each cell type, the day 0 to 6 samples were loaded in
duplicate, the average normalized band intensity values of mtDNA relative to nDNA on day 0 samples were set to 100%, and all other samples were
compared with it. The experiments were repeated in triplicate (n = 12, quadruplicates from three experiments using different passages, a total of six blots
were analyzed for each cell type, two blots per experiment). The average half-life values for mtDNA depletion were WT 29.6 ± 10.5 h and Y955C