The GRACILE mutation introduced into Bcs1l causes postnatal complex III deficiency: A viable mouse model for mitochondrial hepatopathy§


  • The project was funded by grants to V.F. from Swedish Research Council (2005-6599, 2008-2898), Lund University funding, and several foundations (A Tielman, HG Dunkers, and Segerfalk) and to H.K. from the Royal Physiographic Society in Lund.

  • §

    Author contributions: P.L. was responsible for the mouse model. H.K. and P.L developed and executed the key methods, R.K. was responsible for histopathology, M.M. for TEM, E.E. for the high-resolution respirometry. V.F. had overall responsibility for the project. P.L. drafted the article, H.K. and V.F. revised it, and all coauthors have been involved in finalizing it.

  • Potential conflict of interest: Nothing to report.


Mitochondrial dysfunction is an important cause for neonatal liver disease. Disruption of genes encoding oxidative phosphorylation (OXPHOS) components usually causes embryonic lethality, and thus few disease models are available. We developed a mouse model for GRACILE syndrome, a neonatal mitochondrial disease with liver and kidney involvement, caused by a homozygous BCS1L mutation (232A>G). This gene encodes a chaperone required for incorporation of Rieske iron-sulfur protein (RISP) into complex III of respiratory chain. Homozygous mutant mice after 3 weeks of age developed striking similarities to the human disease: growth failure, hepatic glycogen depletion, steatosis, fibrosis, and cirrhosis, as well as tubulopathy, complex III deficiency, lactacidosis, and short lifespan. BCS1L was decreased in whole liver cells and isolated mitochondria of mutants at all ages. RISP incorporation into complex III was diminished in symptomatic animals; however, in young animals complex III was correctly assembled. Complex III activity in liver, heart, and kidney of symptomatic mutants was decreased to 20%, 40%, and 40% of controls, respectively, as demonstrated with electron flux kinetics through complex III. In high-resolution respirometry, CIII dysfunction resulted in decreased electron transport capacity through the respiratory chain under maximum substrate input. Complex I function, suggested to be dependent on a functional complex III, was, however, unaffected. Conclusion: We present the first viable model of complex III deficiency mimicking a human mitochondrial disorder. Incorporation of RISP into complex III in young homozygotes suggests another complex III assembly factor during early ontogenesis. The development of symptoms from about 3 weeks of age provides a convenient time window for studying the pathophysiology and treatment of mitochondrial hepatopathy and OXPHOS dysfunction in general. (HEPATOLOGY 2011:53:437-447.)

Respiratory chain disorders are found in at least 1:5,000 live births.1 Mitochondrial hepatopathies present at an early age and are found in 10%-20% of patients with respiratory chain defects.2 Oxidative phosphorylation (OXPHOS) is dependent on the five respiratory chain complexes as well as assembly factors, cofactors, and electron carriers like cytochrome c and ubiquinone.3 Nuclear DNA genes influencing OXPHOS encode either subunits of the complexes, assembly (ancillary) factors, proteins affecting maintenance and expression of mitochondrial DNA, or proteins related to mitochondrial dynamics.3, 4

One of the ancillary factors is BCS1L, the only known assembly factor for complex III.5 It was originally identified in yeast as an adenosine triphosphate (ATP)-dependent chaperone, essential for the incorporation of Rieske iron-sulfur protein (RISP) in the last steps of complex III assembly.6–8 More than 20 pathogenic human mutations have been reported in BCS1L.5, 9–16 The most severe pathology is caused by a missense mutation (232A>G) substituting serine for glycine (S78G) in the N-terminus of the protein, causing the autosomal recessive GRACILE syndrome (Fellman syndrome, MIM 603358). The acronym refers to the main presenting findings; fetal Growth Restriction, Aminoaciduria, Cholestasis, Iron accumulation in the liver, Lactacidosis, and Early death within days or weeks.10, 17 The pathogenesis of the severe phenotype remains obscure. Other mutations in BCS1L, both homozygous and compound heterozygous, cause a variety of phenotypes, such as neonatal complex III deficiency and encephalopathy with or without visceral involvement,5 as well as the least severe one, Björnstad syndrome,13 with sensorineural hearing loss and brittle hair.

To date, experimental studies on BCS1L and its mutations have been limited to yeast or in vitro cell culture models. We have developed a model of GRACILE syndrome by introducing the Bcs1l 232A>G mutation into mice, using gene targeting. Mice homozygous for the mutation resemble the human syndrome in many ways, including hepatopathy with progressive complex III deficiency and short lifespan. This is the first viable mouse model of complex III deficiency and also the first nuclear transgenic model mimicking a human mitochondrial disorder.

Abbreviations: BN-PAGE, blue native polyacryl amide electrophoresis; FCCP, the protonophore and uncoupler carbonylcyanide-p-trifluoromethoxyphenylhydrazone; H&E, hematoxylin-eosin; ORO, Oil-Red-O; OXPHOS, oxidative phosphorylation; PAS, periodic acid-Schiff; PCR-RFLP, polymerase chain reaction - restriction fragment length polymorphism; RCR, respiratory control ratio; RISP, Rieske iron sulfur protein; ROS, reactive oxygen species; SUIT, substrate-uncoupler-inhibitor titration.

Materials and Methods

Bcs1l 232A>G Mutant Mice and Screening Procedures.

The gene-targeting vector was constructed from a diphtheria toxin expression vector (pPGKdt-AbpA: a kind gift from Professor Phillippe Soriano, Fred Hutchinson Cancer Research Center, Seattle, WA), into which two genomic Bcs1l fragments and a neomycin G418 resistance gene (loxHSVtk-neo-tk-lox), driven by the thymidine kinase promoter (tk) were introduced. The two genomic fragments comprised upstream and downstream genomic Bcs1l sequences, respectively. Sequences were amplified from the murine genomic BAC-clone RPCI-22 (Taconic, Hudson, NY) derived from the 129/SvEvTac mouse strain, using the proofreading Phusion DNA-polymerase (Finnzymes Oy, Espoo, Finland). The 3′-fragment was mutated in exon 2 to achieve the 232A>G base substitution, using the mutagenesis kit QuickChangeII (Stratagene, La Jolla, CA). The entire gene construct was sequenced before introduction into 129/Sv embryonic stem cells using standard gene targeting techniques. The loxP-flanked tk-neo cassette was removed by transient transfection of a cre-expression plasmid. Mice heterozygous for the 232A>G mutation on a mixed 129/SvEvTac × C57BL/6 background were interbred to create homozygous 232A>G mutant mice.

G418-resistant embryonic stem cell clones were screened for homologous recombination using the polymerase chain reaction (PCR) primers P1 (5′-TGG ACCAGCTGAGTTGCAGTAAAC-3′) and P2 (5′-CGA AGTTATTAGGTCTGAAGAGGAGT-3′). Following removal of tk-neo, the 232A>G mutation was verified by restriction fragment length polymorphism (RFLP)-PCR using the primers P3 (5′-ATCCCTCGGTTTTGACA GTGCC-3′) and P4 (5′-TTTCCTCCATCGATCCAC CCTC-3′) and Taq polymerase from Fermentas (St. Leon-Rot, Germany). The PCR product was digested with the restriction enzyme BsaJ1 (Fast Digest, Fermentas) before gel electrophoresis.

Animal Experiments.

All animal experiments were performed with the approval of the Lund regional animal research ethics committee, Sweden (permits M170-06, M158-08, 31-8265/08). Mice were maintained on rodent diet (Labfor R34, Lactamin, Stockholm, Sweden) and water was available ad libitum in a vivarium with a 12-hour light/dark cycle at 22°C. Littermate wildtype mice (Bcs1lA/A) or heterozygotes (Bcs1lA/G) were used as controls. Animals were sacrificed by cervical dislocation or decapitation.

Lactate Measurements.

Blood lactate was measured using Accusport/Accutrend Lactate Portable Lactate Analyzer (Accusport, Sports Resource Group, USA) immediately following heart puncture of sacrificed animals.


Tissues obtained after sacrifice of the mice were immediately fixed in 4% buffered formalin or snap-frozen for storage at −75°C. The formalin-fixed tissues were processed according to a normal pathology laboratory routine to be paraffin-embedded. Sections were stained using hematoxylin-eosin (H&E) for basic morphology.

Liver sections were additionally stained with periodic acid-Schiff (PAS) without and with diastase for glycogen, and silver staining (for reticulin fibers). From snap-frozen liver tissues standard Oil-Red-O (ORO) fat staining was performed.

Accumulation of iron was assessed using standard Perl's iron staining (Berlin blue). The intensity of positive iron staining was graded from 0 to 4 (intense positivity in hepatocytes and Kupffer cells) in liver sections.18

Histological findings were assessed by three investigators, including an experienced pediatric pathologist (R.K., H.K., V.F.), who were unaware of the mouse age and phenotype.

Electron Microscopy.

Liver tissue samples (about 1 mm3) were prepared for electron microscopy (JEOL JEM 1230) and images were recorded with a Gatan Multiscan 791 CCD camera as described.19

Mitochondrial Preparations.

Tissue for isolation of mitochondria was kept in ice-cold isolation buffer (0.32 M sucrose, 10 mM Trizma Base, 2 mM EGTA), and homogenized in isolation buffer supplemented with bovine serum albumin (BSA). Mitochondria were prepared from homogenates by sequential centrifugation including density purification on 19% Percoll.20 The resulting mitochondria were either used directly in respirometry or aliquoted and frozen at −75°C for subsequent complex III measurement or blue native polyacryl amide electrophoresis (BN-PAGE). All chemicals were from Sigma Aldrich (Stockholm, Sweden).

BN-PAGE and Immunoblotting.

Respiratory chain complexes were prepared from frozen mitochondria as described.21 For liver, kidney, and heart, the ratio digitonin per mg protein was 0.05%, 0.1%, and 0.1%, respectively. Depending on the tissue and protein investigated, 2 to 12 μg protein was run on a Native PAGE 4-16% Bis-Tris gel and blotted to PVDF membrane using Iblot equipment (Invitrogen, Carlsbad, CA). After blocking in 5% dry milk the blots were incubated with antibodies detecting subunits of complex III (RISP; MS 305, Core I; MS 303), complex IV (subunit I; MS 404), and complex II (30 kDa IP; MS 203) using antibodies from MitoSciences (Eugene, OR).

Complex III Enzymatic Assay.

Complex III activity was measured as electron flux kinetics from ubiquinol through complex III to cytochrome c22 using chemicals from Sigma Aldrich. Decylubiquinol was prepared from decylubiquinone by reduction with NaBH4.22 Frozen mitochondrial pellets were thawed and resuspended in buffer (20 mM KH2PO4, pH 7.4, 2% sodium cholate). Mitochondria (1-5 μg protein depending on organ) were added to 200 μL reaction buffer (100 μM EDTA, 0.2% defatted BSA, 3 mM NaN3, 60 μM cytochrome c, 50 mM KH2PO4, pH 7.4) and parallel triplicate samples with equal amounts of Bcs1lG/G and littermate control mitochondria were measured in 96-well plates. The reaction was initiated by the addition of 2 μL 10 mM decylubiquinol dissolved in ethanol, and the absorbance of cytochrome c was determined spectrophotometrically at 550 nm and 30°C every 15 seconds. Complex III activities were calculated as the increase of absorbance per second from the linear part of the curve and corrected for nonenzymatic reduction of cytochrome c by subtracting the absorbance values of antimycin-treated mitochondria for each sample. Relative complex III activity for each Bcs1lG/G sample was expressed as percent of control activity.

Mitochondrial Oxygen Consumption.

Mitochondrial oxygen consumption in freshly isolated mitochondria was measured with Oroboros Oxygraph-2k (Oroboros Instruments, Innsbruck, Austria) using DatLab 4 software allowing on-line respiration rate output with high sensitivity, low noise, and concentration-dependent background correction.23 Experiments were run at 37°C in MIR0523 using identical mitochondrial concentrations for mutants and controls (150-800 mg/mL).

For titration of tolerance to the complex III specific inhibitor myxothiazol,24 experiments were run with MIRO5 containing malate and glutamate. ADP was added (5 μL of 0.25 M) prior to each titration with myxothiazol (all from Sigma Aldrich) to obtain state 3 respiration. Oxygen consumption rate was expressed as pmol/s × mL × mg. The concentration of myxothiazol inhibiting it by 50% (IC50) was determined and defined as the tolerance of mitochondria to the inhibitor. Relative values of tolerance were expressed as percent of controls.

Mitochondrial oxygen consumption was further analyzed using the substrate-uncoupler-inhibitor titration (SUIT) protocol.23 Complex I-driven basal respiration was induced by adding malate (5 mM) and pyruvate (5 mM) and OXPHOS was determined following addition of ADP (1 mM) and glutamate (5 mM). Maximal substrate dependent respiration by way of convergent input through both complex I and complex II was obtained by adding succinate (10 mM). State 4 basal respiration was evaluated by adding oligomycin, an ATP-synthase inhibitor. Maximal capacity of the electron transport system was obtained by titrating the optimal uncoupling concentration of the protonophore and uncoupler carbonylcyanide-p-trifluoromethoxyphenylhydrazone (FCCP) in each experiment. Inhibition of complex I by rotenone revealed the contribution of complex II, and finally addition of antimycin-A determined the residual oxygen consumption.


Statistical analyses were performed with SPSS 17.0.1 for Microsoft Windows (SPSS, Chicago, IL). Group differences of continuous variables were analyzed with independent samples t test or Mann-Whitney U test, littermate comparisons with paired t test. Group differences of relative complex III activity over time were analyzed with linear regression and myxothiazol inhibition capacity as a function of age in the two groups with multiple linear regression. SUIT results were analyzed with unpaired t test.


Verification of Gene Targeting.

We generated mice to harbor a base substitution in exon 2 of the Bcs1l gene (232A>G) (Fig. 1A). Embryonic stem (ES) cells were screened by PCR for homologous recombination, using one 5′-external primer and one primer complementary to the loxHSVtk-neo-tk lox in the construct (Fig. 1B). The 232A>G mutation was verified in targeted ES cells using PCR-RFLP. Amplification of the exon 2 region was followed by digestion of the PCR product (433 basepairs [bp]) with the restriction endonuclease BsaJ1, which cleaves a novel recognition sequence introduced by the 232A>G mutation (Fig. 1C). The same method was used for genotyping of animals. The integrity of the targeted locus was verified by sequencing outer and flanking regions of the construct, the loxP sites, and the 232A>G mutation in the ES-clone used to generate mutant animals.

Figure 1.

Gene targeting of the Bcs1l gene. (A) A gene targeting construct harboring the Bcs1l 232A>G mutation was introduced into the genome of wildtype (wt) mouse ES cells, resulting in a recombinant allele (rec.). The tk-neo antibiotic resistance gene was subsequently removed by cre-lox recombination (rec.Dneo). (B) PCR screening for targeted ES clones using the external primers P1 and P2 indicated in (A). Targeted clones: T1 and T2. Negative clones: N1 and N2. (C) Verification of Bcs1l mutagenesis in ES cells by PCR-RFLP using the primers P3 and P4 indicated in (A) and the restriction endonuclease Bsa J1. The 355-bp and 283-bp fragments identify the wildtype (A) and mutant (G) alleles, respectively.

232A>G Homozygotes Are Growth Restricted and Have a Short Life Span.

Inbreeding heterozygotes yielded homozygous mutants (Bcs1lG/G) at expected Mendelian ratios (Bcs1lA/A, Bcs1lA/G, and Bcs1lG/G distribution: 23%, 53%, and 24%, respectively, n = 614). Heterozygous mice (Bcs1lA/G) developed normally, were fertile, and indistinguishable from wildtype mice (Bcs1lA/A) in all aspects of phenotype and mitochondrial function examined. Thus, both Bcs1lA/A and Bcs1lA/G littermates to Bcs1lG/G mice were used as controls. The initial abnormality in Bcs1lG/G mice was slower weight gain, starting from postnatal days 21-25 (P21-25, Fig. 2A). After reaching a plateau, the weight decreased concomitantly with appearance of disease symptoms. At the terminal stage, the mice became hunchbacked and inactive, and for ethical reasons the majority were sacrificed an estimated 2-3 days before spontaneous death (Fig. 2B,C). The earliest deaths occurred at day 29. However, about 5% of the animals had no other symptoms than growth restriction until P70-165. The relative body weight at P31-38 of Bcs1lG/G mice was 59% and 68% of littermate control males and females, respectively, and was accompanied by a proportional reduction in size of liver (Fig. 2D), kidney, and heart. In general, homozygous mutant mice appeared healthy and active, with normal movements and muscular strength until the last few days before death.

Figure 2.

Phenotype of Bcs1lG/G mice. (A) Weight development of females and males. Genotypes are indicated as A/A = wildtype, A/G = heterozygote, and G/G = homozygote. Number of mice in each group aged less than P16 was 2-15, thereafter at least 5. Significant weight decrease was observed at P21-25 in Bcs1lG/G females (n = 9, *P = 0.012 vs. A/A, n = 5, and P = 0.001 vs. A/G, n = 21) and males (n = 15, **P = 0.002 vs. A/G, n = 20, and P = 0.10 vs. A/A, n = 15). Data represent mean, error bars indicate SD. (B) Survival of Bcs1lG/G mice (n = 90) and controls (n = 29) are shown by solid and broken lines, respectively. (C) Appearance of mice aged 35 days. (D) Livers taken from the same mice. The Bcs1lG/G liver is pale, indicative of liver steatosis; however, the relative size is similar to littermates.

Symptomatic Bcs1lG/G mice had elevated blood lactate concentrations (mean ± standard deviation [SD] in Bcs1lG/G 6.5 ± 1.4 mmol/L, n = 13 and in controls 4.5 ± 0.5 mmol/L, n = 11, P = 0.0001).

Bcs1lG/G Mice Have Structural Abnormalities in Liver, Kidney, and Testis.

Bcs1lG/G mice had a pale yellowish liver (Fig. 2D). The other organs examined, including kidney, lung, heart, brain, spleen, stomach, and intestines appeared macroscopically normal. Microscopic studies revealed a normal liver morphology in Bcs1lG/G animals less than 3 weeks of age (Fig. 3). From about P24, periportal degenerating hepatocytes with compensatory regeneration appeared. The hepatocytes showed glycogen depletion and accumulation of microvesicular lipid droplets indicative of steatosis (Fig. 3). In P30 or older symptomatic Bcs1lG/G animals, fibrotic fibers and nodular hepatic structure started to appear (Fig. 3). They had no marked hepatic iron positivity (Fig. 3); grade 1 positivity18 was found in 50% of symptomatic Bcs1lG/G animals, but only in 12% of healthy littermates. In electron microscopy, the overall cellular structure and the number of mitochondria in Bcs1lG/G hepatocytes were comparable to controls, but mitochondria from P30 mutant mice were considerably enlarged and less electron-dense (Fig. 3).

Figure 3.

Liver histopathology of Bcs1lG/G mouse (G/G) and littermate control (A/A) aged 16 (P16) and 30 (P30) days. At P16 the H&E findings were similar in both genotypes, only a slight decrease in glycogen could be found in Bcs1lG/G mouse. In symptomatic Bcs1lG/G mice at P30, H&E staining showed periportal shrunken hepatocytes (arrow) with eosinophilic condensed cytoplasm, where small amounts of iron could be detected. A few inflammatory cells were present, suggesting ongoing hepatocellular destruction. PAS staining showed depletion of glycogen, silver staining fibrotic fibers, and ORO intensive staining compared to isolated orange microvesicular lipid droplets in control mice. All scale bars are 5 μm. Transmission electron microscopy (TEM) of liver showed similar mitochondria in Bcs1lG/G and control mice at P16, whereas a considerably enlarged mitochondrion with preserved cristae in Bcs1lG/G mice at P30. The arrow indicates a lipid droplet.

In kidneys of symptomatic Bcs1lG/G mice, proximal tubuli were smaller, reduced in number, and contained tubular exudate indicating tubulopathy (Supporting Fig. S). No gender differences were found in liver or kidney histology. In testes of symptomatic Bcs1lG/G animals, Leydig cells were decreased in number and size, in some areas completely absent (Supporting Fig. S). No histopathological abnormalities were detected in the heart (Supporting Fig. S), brain, lung, skeletal muscle, stomach, or intestine of symptomatic animals.

Decreased Incorporation of RISP into Complex III in Symptomatic Bcs1lG/G Mice.

Western blotting of hepatocellular lysates of symptomatic (>P30) Bcs1lG/G mice demonstrated a substantial decrease of BCS1L protein compared to littermates, whereas RISP levels were similar (Fig. 4A). In BN-PAGE of liver mitochondria, BCS1L was present in lower amounts than in controls in P15 Bcs1lG/G and became, in agreement with hepatocellular lysate, undetectable at later ages (Fig. 4B).

Figure 4.

Bcs1lG/G mitochondria show decreased levels of BCS1L at all ages and impaired assembly of complex III from 24 days of age (P24). (A) In western blot of hepatocyte whole cell lysate, a decreased amount of BCS1L and an unaffected RISP amount were found in Bcs1lG/G. Blue Native PAGE on liver (B), kidney and heart (C) mitochondria from Bcs1lG/G and controls followed by western blot investigating the presence of BCS1L and the assembly of complex III. Complex II was used as a loading control. Age and genotype of the mice are indicated. CII, CIII, CIV indicate complexes II, III, IV, CIII/CIV, a supercomplex containing both components. Anti-Core I is an antibody detecting complex III subunit Core I present in both precomplex and fully assembled complex III. (B) BCS1L is decreased at all ages in Bcs1lG/G mice liver, whereas RISP is incorporated in complex III in young animals and depleted in P32 animals. (C) Compared to liver, RISP was incorporated in complex III to a larger extent in kidney and heart of sick (P30 and older) mutants. No supercomplexes were found in these tissues.

Normal RISP incorporation into complex III was found in the youngest animals, whereas at P24 incorporation decreased. The levels of complex III without RISP (precomplex III) were comparable to controls (Fig. 4B).

In the kidney and heart mitochondria of Bcs1lG/G mice, RISP incorporation was in general less affected at all ages, but clearly decreased in all symptomatic mice (Fig. 4C).

Impaired Function of Complex III in Bcs1lG/G Mice.

Bcs1lG/G mice had a progressive decrease of electron flux kinetics through complex III with age (Fig. 5). In affected animals, complex III activity relative to littermate controls decreased to about 20%, 40%, and 40% in the liver, heart, and kidney, respectively (Fig. 5B-D).

Figure 5.

Complex III activity in mitochondria from Bcs1lG/G and control mice. (A) Spectrophotometric assessment of cytochrome c reduction following addition of decylubiquinol. The rate of absorbance increase (Δ abs per minute) was determined during the first minute of reaction. Antimycin treated samples were included to adjust for complex III-independent reduction of cytochrome c. Filled and open circles represent Bcs1lG/G and control mitochondria, respectively. (B-D) Relative complex III activities (% Bcs1lG/G activity of control) of mitochondria from liver, heart, and kidney showing decreasing values with increasing age in all tissues (P < 0.001).

Bcs1lG/G mice showed a progressive decline with age in tolerance to the complex III specific inhibitor myxothiazol24 in liver, heart, and kidney (Fig. 6A-D). In contrast, control mice showed increasing tolerance with age.

Figure 6.

Tolerance of mitochondrial respiration to complex III inhibition by myxothiazol decreases with age in Bcs1lG/G mouse tissues. (A) Titration curves for liver mitochondria isolated from one P35 Bcs1lG/G (filled circles) and littermate control (open circles) showing decline of mitochondrial O2-consumption rate with increasing concentration of myxothiazol. Tolerance of each sample to myxothiazol was defined as the 50% inhibition concentration (IC50 value). The steeper slope of the Bcs1lG/G curve is a representative sample of strongly reduced tolerance to myxothiazol of affected mice aged P30 or more. (B-D) Bcs1lG/G and littermate control samples from liver, heart, and kidney mitochondria run in parallel showing myxothiazol IC50 values as a function of age. In liver, there was a group difference in young animals (P = 0.03) and the slopes differed significantly (P< 0.001). Tolerance decreased in Bcs1lG/G and increased in controls with age (P = 0.001). In heart and kidney, no significant group difference was found in young animals, but the tolerance increased in controls, whereas it decreased in Bcs1lG/G over age (P = 0.01 and P < 0.001, respectively, multiple linear regression).

Impaired Mitochondrial Respiration in Bcs1lG/G Liver.

For direct assessment of integrated respiratory chain function, we examined oxygen consumption in freshly isolated liver mitochondria (Table 1).23 Basal nonphosphorylating respiration was lower in BCS1lG/G mice. However, mitochondria from wildtype and mutant mice were both efficiently coupled, as illustrated by comparable respiratory control ratio (RCR complexI+complexII), 3.3 and 3.8 for BCS1lA/A and BCS1lG/G mice, respectively. Under state 2 conditions, using malate and pyruvate for complex I driven respiration, slightly higher oxygen consumption was found in Bcs1lG/G mitochondria compared to controls. This difference, however, was not present when mitochondria were using endogenous substrates (basal) or during complex I-driven, ADP-stimulated respiration (Table 1). However, convergent substrate input by the subsequent addition of the complex II substrate succinate (in the presence of complex I substrates) resulted in significantly lower oxygen consumption in Bcs1lG/G liver mitochondria compared to control (Table 1). This implies a lower electron transport capacity through Bcs1lG/G liver mitochondria, most likely caused by complex III deficiency. These differences remained following titration of the uncoupler FCCP (representing the maximal electron transport capacity in mitochondria). These results suggest impaired mitochondrial respiration in the Bcs1lG/G mouse liver during high oxygen consumption.

Table 1. High-Resolution Respirometry: SUIT Protocol of Liver Mitochondria
Substrate/ InhibitorBcs1lA/A O2 flux/ pmol/(s*mg) Mean +/- SEMBcs1lG/G O2 flux/ pmol/(s*mg) Mean +/- SEMP-value
  1. Respiratory activities are deduced by way of sequential addition of substrates, uncoupler, and inhibitors. Complex I-driven basal respiration was induced by adding malate (5 mM) and pyruvate (5 mM). Maximal complex I-driven phosphorylating respiration was determined following addition of ADP (1 mM) and glutamate (5 mM), an additional complex I substrate. Maximal substrate-dependent respiration by way of convergent input through both complex I and complex II was obtained by adding succinate (10 mM). Basal nonphosphorylating respiration was evaluated by adding oligomycin, an ATP-synthase inhibitor. Maximal capacity of the electron transport system was obtained by titrating the protonophore FCCP. Inhibition of complex I by rotenone revealed the contribution of complex II.

Basal20.9 ± 10.49.6 ± 2.10.345
Malate/pyruvate30.5 ± 4.554.1 ± 6.10.036
ADP106.2 ± 28.4132.4 ± 15.60.464
Glutamate152.0 ± 48.5165.4 ± 29.30.825
Succinate394.9 ± 52.1201.3 ± 39.90.042
Oligomycin121.0 ± 10.552.1 ± 10.50.016
FCCP max887.7 ± 55.3387.4 ± 78.10.006
Rotenone698.9 ± 52.40349.3 ± 57.10.011


Mutations in the complex III chaperone BCS1L cause a broad spectrum of clinical symptoms and often display unexplained tissue specificity.25 The homozygous point mutation 232A>G, however, causes a consistent phenotype, the lethal GRACILE syndrome,17, 26 which motivated us to generate a transgenic mouse with this specific mutation for the investigation of disease mechanisms and functions of BCS1L. The main findings in Bcs1lG/G mice were growth restriction and progressive liver disorder from the fourth week of life leading to death before 6 weeks of age in the majority of the animals, corresponding to characteristics of GRACILE syndrome.

Several transgenic mouse models have been developed for the study of nuclear encoded OXPHOS genes. Conventional knockouts have resulted in embryonic lethality,27, 28 which can be circumvented by the generation of conditional knockouts, exemplified by models of complex I29, 30 and complex IV deficiency.31–33 However, mouse models of specific mitochondrial diseases caused by nuclear mutations are missing.

The Bcs1lG/G mouse presented with the typical tissue specificity of the human mutation, i.e., liver and kidney involvement,34 but without any obvious changes during early postnatal life. In livers from Bcs1lG/G mice, only small amounts of iron were detectable, whereas in GRACILE syndrome hepatic iron accumulation was found at 13 weeks of gestation35 and considerable hemosiderosis is present at birth.36 The discrepancy between mice and human might be related to differences in placental transport of iron. The postnatal decrease of hemosiderosis in GRACILE patients is also consistent with disturbed placental transport.17, 36

The earliest abnormalities observed in Bcs1lG/G mice were depletion of liver glycogen stores and steatosis. A similar phenotype was observed in a knockout model of liver-specific complex IV subunit COX10 causing reduced OXPHOS activity.31 For both models it is conceivable that lack of stored glycogen is caused by, on the one hand, OXPHOS dysfunction leading to decreased glycogen synthesis and, on the other, by increased glycolysis as an alternative source of ATP. OXPHOS deficiency might also explain the hepatic steatosis observed. Mitochondrial β-oxidation of fatty acids requires NAD+, and decreased electron flux through complex III or other components of OXPHOS inhibit the formation of NAD+ from NADH by reducing electron entry through complex I. This would lead to impaired lipid breakdown and the accumulation of microvesicular fat droplets.37, 38

In Bcs1lG/G tissues, less BCS1L protein was detected compared to controls. This could potentially be explained by decreased stability of the protein, as we observed in COS cells transfected with a 232A>G mutant BCS1L.10 BCS1L depletion or functional impairment result in incomplete assembly of complex III, as shown in several yeast8, 39 and human5, 21, 40 studies. Also in our mouse model, BCS1L depletion was accompanied with deficient assembly of complex III. All tissues showed, to various extents and in an age-dependent fashion, decreased content of RISP in complex III. Despite the decreased levels of BCS1L in animals aged less than 3 weeks, RISP incorporation was normal or only slightly affected. Part of the complex III assembly in Bcs1lG/G tissues might be ascribed to residual activity of the mutant protein. However, the existence of another, yet unknown, assembly factor(s) involved in RISP incorporation cannot be ruled out. Such a factor could be expressed during embryonic development but later become down-regulated, making complex III assembly increasingly dependent on BCS1L. In a recent study of yeast strains harboring mutations in BCS1L, lack of incorporation of RISP into CIII was compensated by extragenic mutations, suggesting an additional factor contributing to the assembly of complex III.39

The depletion of BCS1L and the impaired CIII assembly in Bcs1lG/G mice were associated with reduced functionality of the complex. These findings correlated with enlargement and altered morphology of mitochondria. A recent study suggests interaction of BCS1L with another mitochondrial protein, LETM1, which maintains mitochondrial shapes. Down-regulation of either of them resulted in enlarged mitochondria.41

Interestingly, although there was a pronounced and progressive decrease in relative complex III activity in mitochondria from liver, kidney, and heart from Bcs1lG/G animals, only liver and kidney showed histological abnormalities. In humans, proximal tubulopathy has been a common finding in BCS1L-mutations.9–16, 26, 40 In Bcs1lG/G heart, complex III activity less than 40% of controls (Fig. 5) did not cause any microscopic abnormalities. Neither have any myocardial abnormalities been found in GRACILE syndrome.34 Thus, it seems that complex III in the myocardium has a high reserve electron transport capacity, which is supported by the fact that cardiomyopathy is rare in complex III deficiency.28

Leydig cells are sensitive to decreased complex III activity according to our results. This is in line with the dramatic decline in spermatogenesis in the Ant4−/− mouse (testis-specific ADP/ATP carrier gene),42 and with the inhibition of testosterone production by myxothiazol in a Leydig cell study.43

ADP-stimulated complex I-driven respiration was comparable to controls, whereas decreased respiratory capacity was observed in liver mitochondria from Bcs1lG/G mice when mitochondria were stressed to maximum electron transport. Thus, respiratory deficiency becomes apparent under conditions requiring high capacity of the respiratory chain including electron transport by way of complex III. An association between complex III deficiency and secondary decreased complex I activity has been proposed, but normal activity has been shown in several patients.40, 44 In our mutant mice we did not find a functional deficit in complex I.

Proposed mechanisms for phenotypic variability in human OXPHOS disorders include altered organization, composition, or content of complexes,45, 46 different demands of energy,47 and whether glycolysis or OXPHOS is preferred for ATP generation.31 Complex III deficiency may increase reactive oxygen species (ROS) formation,13 but wide variations in ROS production and antioxidant capacity have been found in human mutations.13, 21, 40

One of the most puzzling questions regarding GRACILE is why the point mutation, 232A>G, causes a more consistent and far more devastating clinical disease than a number of other mutations in the same region of the gene, with at least as significant effects on complex III assembly and function as the GRACILE mutation. The death mechanisms in the Bcs1lG/G mice is also unclear, but may be related to ATP depletion, as proposed for the COX10 knockout mouse.31 Future studies should be directed to investigation of other potential function(s) of BCS1L than complex III assembly.

Due to the favorable time range for disease development, occurring at 3-5 weeks of age, and the reproducible pathology, we consider this model valuable for the investigation of the molecular pathology, clinical manifestations, and therapeutic modalities of the GRACILE syndrome, as well as pathophysiology and treatment of cirrhosis and OXPHOS dysfunction in general.

Acknowledgment: We thank David Nicholls for expert advice regarding complex III activity assessments, Mats Ehinger for initial histopathology screening, Valentina Fedele for brain histology, Magnus Hansson and Saori Morota for advice relating to high-resolution respirometry, and Eva Hanson for skilled technical execution of several methods.