LONP1 de novo dominant mutation causes mitochondrial encephalopathy with loss of LONP1 chaperone activity and excessive LONP1 proteolytic activity

Abstract

LONP1 is an ATP-dependent protease and chaperone that plays multiple vital roles in mitochondria. LONP1 is essential for mitochondrial homeostasis due to its role in maintenance of the mitochondrial genome and its central role in regulating mitochondrial processes such as oxidative phosphorylation, mitophagy, and heme biosynthesis. Bi-allelic LONP1 mutations have been reported to cause a constellation of clinical presentations. We report a patient heterozygous for a de novo mutation in LONP1: c.901C>T,p.R301W presenting as a neonate with seizures, encephalopathy, pachygyria and microcephaly. Assays of respiratory chain activity in muscle showed complex II-III function at 8% of control. Functional studies in patient fibroblasts showed a signature of dysfunction that included significant decreases in known proteolytic targets of LONP1 (TFAM, PINK1, phospho-PDH E1α) as well as loss of mitochondrial ribosome subunits MRPL44 and MRPL11 with concomitant decreased activity and level of protein subunits of oxidative phosphorylation complexes I and IV. These results indicate excessive LONP1 proteolytic activity and a loss of LONP1 chaperone activity. Further, we demonstrate that the LONP1 N-terminal domain is involved in hexamer stability of LONP1 and that the ability to make conformational changes is necessary for LONP1 to regulate proper functioning of both its proteolytic and chaperone activities.

Introduction

LONP1 is an ATP-dependent protease that plays multiple roles in mitochondria. It’s an effector of protein quality control for the mitochondria by degrading unassembled, misfolded or oxidized proteins and its expression is induced under conditions of hypoxia, ischemia and endoplasmic reticulum stress (Bota et al 2002; Fukuda et al 2007; Hori et al 2002; Kita et al 2012; Papa et al 2014; Teng et al 2013). In addition, LONP1 regulates cellular processes through degradation of proteins that play regulatory roles such as steroidogenic acute regulatory protein (STARD1)(Granot et al 2007; Ondrovicova et al 2005), mitochondrial processing peptidase alpha (MPPα) which cleaves the mitochondrial localization signal from proteins post-entering the mitochondria enabling their functionality (Ondrovicova et al 2005; Suzuki et al 1994), COX4–1 to modulate respiration at varying oxygen concentrations (Fukuda et al 2007), 5-Aminolevulinic acid synthase (ALAS-1) an enzyme that controls cellular heme biosynthesis (Kita et al 2012). Additionally, LONP1 modulates mitophagy though the degradation of PINK1 (Jin et al 2013; Thomas et al 2014).

LONP1 is integral to mitochondrial genome transcription, translation and replication. One clear aspect of its influence over mtDNA lies in the ability of LONP1 to bind single stranded DNA and RNA including mitochondrial DNA (Fu et al 1998; Liu et al 2004; Lu et al 2007). While the majority of LONP1 is present in the mitochondrial matrix it has also been observed to co-localize to the mitochondrial nucleoid (Bota et al 2002; Liu et al 2004) and to coimmunoprecipitate with POLG and Twinkle (Liu et al 2004). Further, LONP1 has been demonstrated to degrade TFAM, which is essential for mtDNA packaging, transcription and replication, and the knock down of LONP1 resulted in higher mtDNA copy number and higher mitochondrial transcript levels (Ekstrand et al 2004; Lu et al 2013; Matsushima et al 2010; Pohjoismaki et al 2006). Supporting a role for LONP1 in translation, coimmunoprecipitation studies showed five components of the mitochondrial ribosome coimmunoprecipitate (MRPL17, MRPL22, MRPL23, MRPL39, and MRPL44) with LONP1 (Zurita Rendon et al 2018). Moreover, impaired ribosome biogenesis was observed in a human cellular LONP1 loss of function model, including impaired maturation and solubility of FASTKD2 (Zurita Rendon et al 2018) which plays a role in mitochondrial ribosome biogenesis, indicating that LONP1’s role in translation may stem from its chaperone function. An extensive response of the mitoribosome to LONP1 perturbation was observed in melanoma cells (Quiros et al 2014).Taken together, LONP1 appears to play an essential role in mitochondrial nucleoid dynamics and mtDNA maintenance, transcription and translation of the mitochondrial genome.

Multiple studies support a chaperone role for yeast, human and plant LONP1 (Li et al 2017; Rep et al 1996; Suzuki et al 1994). Evidence that LONP1 functions as a chaperone and that this chaperone function depends on the N-terminal and ATPase domains includes protein structural and sequence motif similarity to other chaperone proteins that have homology to LONP1 N-terminal and ATPase domains but do not have a proteolytic domain (Neuwald et al 1999). Experimental studies showed that LONP1/lon overexpression in yeast or human cells leads to greater amounts of assembled complex I, IV, V in a non-proteolytic dependent manner (Cheng et al 2013; Rep et al 1996). Moreover, hypoxia and H₂O₂ led to a LONP1-dependent increase in complex I subunit NDUFB8 that is not dependent on the proteolytic domain of LONP1 (Cheng et al 2013).

The first report of human disease caused by mutations in LONP1 was for CODAS syndrome, an autosomal recessively inherited disorder characterized by a specific constellation of cerebral, eye, ear, dental and skeletal anomalies (Dikoglu et al 2015; Khan et al 2018; Strauss et al 2015). These patients carried bi-allelic mutations, which were present throughout the ATPase and protease domains of the protein. A single patient with clinical presentation typical of a mitochondrial encephalopathy and no evidence of skeletal or dental defects was reported to have compound heterozygous missense variants in the ATPase domain and proteolytic domains of LONP1 (Peter et al 2018). This subject’s dermal fibroblasts showed normal levels of mtDNA copy number, TFAM as well as oxidative phosphorylation (OXPHOS) proteins. Affected siblings with a homozygous missense mutation in the proteolytic domain of LONP1 were reported to have a clinical presentation that was distinct from CODAS and included progressive cerebellar atrophy, congenital cataracts, severe hypotonia and developmental delay, with mitochondrial studies in muscle showing mild dysfunction and reduced pyruvate dehydrogenase complex activity in fibroblasts (Nimmo et al 2019). Fibroblasts from the siblings reportedly showed normal levels of mtDNA, mitochondrial transcripts, normal OXPHOS activity and control levels of TFAM and PINK1.

In this study we report a new mode of inheritance for LONP1 disease as manifested in a patient with LONP1 de novo dominant mutation exhibiting mitochondrial encephalopathy. The distinct clinical presentation of the Subject reported here along with the novel observation of a dominantly acting mutation in LONP1, pointed toward a gain of function mutation as opposed to the previously reported presumed loss of function mutations. However, LONP1 functions as both a chaperone and an ATP-dependent protease, and we demonstrate through extensive functional studies in patient fibroblasts that LONP1 mutant p.R301W incurs loss of chaperone activity while retaining proteolytic activity leading to a complex cellular and clinical phenotype.

Results

A subject with mitochondrial encephalopathy with neonatal-onset seizures, microcephaly, pachygyria, progressive white matter atrophy and cortical blindness

The subject was a female born to unrelated parents who presented with seizures on the first day of life. She developed a severe lactic acidosis with a peak blood lactate of 13.0 mmol/L in the setting of status epilepticus. Hyperlactatemia improved prior to NICU discharge, but in subsequent lab draws, lactate was consistently in the 4.0–6.0 mmol/L range. G-tube was placed during her initial NICU stay which was later removed. An echocardiogram conducted on the first day of life was normal and there were no further cardiac concerns after that time and no additional studies.

CT of the brain in the first days of life showed abnormal sulcation of the brain consistent with pachygyria and small irregular calcifications in the periventricular white matter of the frontal lobes bilaterally (Figure 1A). There was also abnormal decrease in signal intensity and thickening of the white matter of the temporal lobes and posterior parietal lobes. MRI of the brain at age 2 months showed pachygyria, microcephaly and white matter volume loss (Figure 1A). Much of the supratentorial white matter was also abnormally T2 hyperintense. Ventricles, sulci and extra-axial spaces over the convexities were diffusely increased since the prior exam, in a pattern suggestive of progressive global volume loss superimposed on white matter volume loss. Cerebellar volume and cortical thickness were normal, and while appearing completely formed, the corpus callosum was diffusely thin. Hippocampi were abnormally small bilaterally. There were subtle areas of intrinsic T1 shortening in the periventricular region that likely correspond to calcification seen on the comparison CT. Optic nerves were moderately hypoplastic bilaterally.

Figure 1.

Mitochondrial encephalopathy caused by de novo LONP1 mutation. A. CT and MRI of the brain of Subject with severe encephalopathy show pachygyria and small irregular calcifications in the periventricular white matter of the frontal lobes bilaterally. CT in the first days of life showed abnormal decrease in signal intensity and thickening of the white matter of the temporal lobes and posterior parietal lobes, and MRI of the brain at age 2 months showed progressive white matter volume loss. B. Electron transport chain activity was measured in muscle and in dermal fibroblasts. Citrate synthase and lactate dehydrogenase activities were also measured in both tissues. Dashed red line is placed at 40% of normal which is the threshold for Modified Walker clinical diagnostic criteria of being designated abnormal. C. Electron transport chain protein components were measured by Western blotting in three controls (C1, C2, C3) and Subject (LONP1) fibroblasts. D. Parents were shown to be negative for the mutation found in the Subject, LONP1 c.901C>T, p.R301W.

Developmental progress was minimal. The subject did not roll over, did not regard faces, had poor head control and minimal limb movements. She had roving eye movements and ophthalmologic exam concluded she was cortically blind. Clinical seizures decreased overall and were controlled on clobazam and keppra. However, the subject continued to have myoclonic jerks several times per day. An EEG obtained at 11 months showed absence of sleep architecture, frequent bifrontal epileptiform discharges oftentimes associated with myoclonic seizures or spasms, as well as disorganization, slowing and lack of expected background features for age, consistent with a diffuse and nonspecific encephalopathy. She died at 12 months of age in the setting of a viral respiratory illness.

Subject tissues show decreased mitochondrial electron transport chain protein activity and subunits

With persistent hyperlactatemia demonstrated as lactate ranging from 3 – 13 mmol/L and consistently 4–6 mmol/L (normal 0.5 – 1 mmol/L) a primary mitochondrial disorder was suspected and patient muscle and fibroblasts were interrogated for mitochondrial dysfunction. ETC complexes I through V were assessed in subject fibroblasts and muscle (Figure 1B and Supplemental Tables 1,2). In muscle, ETC complex I-III had 36% of normal activity and complex II-III showed 8% of normal activity. Activities of complexes II and III were both higher than normal with complex II activity 128% and complex III 162%. Testing of subject fibroblasts showed complex I-III activity was diminished similarly to muscle at 20% of normal. Complex II-III activity in fibroblasts was 58% of normal, which is much less compromised than it was in muscle. Complex II activity in fibroblasts was 111% of normal with complex III activity at 51%. In both muscle and fibroblasts Complex IV was lower than normal but not below modified Walker criteria, with 80% normal activity in muscle and 77% of normal activity in fibroblasts (Bernier et al 2002).

ETC complexes I through V were also assessed in subject fibroblasts using western blotting for proteins NDUFB8, SDHA, UQCRC2, COXIV, and ATP5B (Figure 1C). Complexes I and IV were most severely affected as protein levels of NDUFB8 and COXIV were 10% and 30% of healthy controls. This could be indicative of a loss of chaperone function of LONP1. Complex II and III were 80% and 60% of normal. Complex V appeared unaffected in patient fibroblasts.

Additional biochemical studies showed Coenzyme Q in Subject’s muscle was 5.6 mcg/g (normal = 24–33 mcg/g). Subject was started on CoQ10 supplementation without apparent clinical benefit. Plasma amino acids testing included normal alanine with slightly elevated glutamine (Supplemental Table 3).

Genetic explorations identify a de novo dominant mutation in LONP1

Multiple genetic tests were conducted clinically to identify causative pathogenic variants. Whole genome sequencing of the subject’s mitochondrial genome in muscle tissue did not reveal any known pathogenic mutations, nor did it show any large deletions of mtDNA. A panel of 666 nuclear genes considered relevant for mitochondrial disease also did not result in the identification of any pathogenic variants. Whole exome sequencing of the subject and both of her parents showed a do novo variant in LONP1, NM_004793 c.901 C>T, p.R301W. Neither parent carried the LONP1 c.901 C>T mutation (Figure 1D). No second mutation in LONP1 was identified and a chromosomal microarray did not show any deletions in this gene, nor did it show any deletions considered possibly pathogenic.

LONP1 c.901C>T was not observed in the gnomAD database and was not observed in any previously reported patients with mutations in LONP1. LONP1 has three functional domains an ATPase, protease, and N-terminal substrate binding domain (Figure 2A). All previously reported patient mutations were in the ATPase or protease domains, none were reported in the N terminal domain of LONP1 (Figure 2A) (Dikoglu et al 2015; Khan et al 2018; Nimmo et al 2019; Peter et al 2018; Strauss et al 2015). Cryo-electron microscopy showed LONP1 oligomerizes to forms hexamers; the protease and ATPase domains form a hexameric chamber while the N-terminal domain is arranged as a trimer of dimers (Kereiche et al 2016). The crystal structure of the full length of the human LONP1 protein has not yet been determined so we bioinformatically generated a model of both the monomeric (Figure 2A) and the hexameric form of human LONP1 (Figure 2B) in order to determine any potential structural consequences of the p.R301W mutation. In these models, LONP1 p.R301 is a buried charged residue and the missense change p.R301W replaces a charged amino acid (R) with an uncharged residue (W). Further, this missense substitution disrupts all side-chain H-bonds formed by the R residue. The folding energy of p.W301 is significantly higher than for p.R301 (p-value = 1E-100) (Figure 2C). GERP and PhyloP indicate the LONP1 c.901 residue is constrained (Figure 2D) and a multiple species alignment shows the p.R301 is invariant across twelve species (Figure 2E).

Figure 2.

LONP1 gene and protein structure. A. The N-terminal domain of the protein is comprised of a mitochondrial localization signal (MTS) and a protein interaction domain that facilitates polypeptide binding (blue). The ATPase domain is shown in yellow and the proteolytic domain in the C terminus is shown in red. Mutations reported in patients with LONP1 related disease are noted. Black font and lines indicate mutations that were observed in patients with CODAS clinical presentation. Blue font and lines indicate mutations that were observed in patients with classical mitochondrial disease clinical presentation. Red is for the patient reported here as having a de novo dominant mutation and mitochondrial encephalopathy. PHYRE2 was used to generate the monomer three-dimensional structure of NM_004793, amino acids 124 – 959. B. Three-dimensional modeling of the hexamer of LONP1 protein sequence from NM_004793, amino acids 200 – 959. C. The difference in energy for protein folding between the reference sequence and the mutant, p.R301W. D. GERP and PhyloP both show constraint at nucleotide c.901C>T. E. Multiple species alignment of LONP1 nucleotide sequence centered on the LONP1 c.901C>T patient mutation shows conservation across species at this site.

LONP1 patient fibroblasts show decreased mtDNA transcription factor TFAM, mitochondrially encoded proteins and mitochondrial ribosome protein subunits

LONP1 plays an essential role in the mtDNA maintenance, transcription and translation through its interaction with mtDNA nucleoid proteins. Co-IP studies have shown LONP1 binding to TFAM, SSB1, Twinkle, POLG, and mitochondrial ribosome subunits (Liu et al 2004; Zurita Rendon et al 2018). Further, TFAM is degraded by LONP1 when TFAM is not bound to DNA (Lu et al 2013). Primary dermal fibroblasts from subject showed TFAM is diminished to 0.3 of controls (Figure 3A). Given TFAM’s role as an mtDNA transcription factor it would be expected that decreased TFAM would lead to a decrease in mitochondrially encoded proteins and we observed decreased protein levels of mitochondrially encoded MT-ND5 at 11% (C1) and MT-COX I at 50% (CIV) (Figure 3A). This finding is consistent with our observation of decreased complex I and IV protein and activity in fibroblasts as indicated by NDUFB8 (CI) and COX4I1 (CIV) (Figure 1B and 1C). Likewise, subject fibroblasts showed mitochondrial ribosome subunit proteins MRPL11 and MRPL44 were diminished to 22% and 10% of normal, respectively (Figure 3A). Despite these abnormalities in levels of proteins involved in transcription and translation of mtDNA, primary dermal fibroblasts from the subject showed the same amount of mtDNA as healthy controls (Figure 3A).

Figure 3.

mtDNA transcription, translation and mitophagy pathway targets of LONP1 are diminished in LONP1 Subject. A. Mitochondrial transcription, ribosome proteins, TFAM, MT-ND5, MT-COX1, MRPL44, MRPL11, were measured by Western blotting in three controls (C1, C2, C3) and Subject (LONP1) fibroblasts. mtDNA copy number in Subject fibroblasts (S) were the same as control fibroblasts (C) B. Mitophagy & Kreb’s cycle entry proteins PINK1 and phospho-PDH-E1α Ser293/Ser300 show deficiency in LONP1 Subject fibroblasts.

LONP1 patient fibroblasts show loss of mitophagy pathway factors PINK1 and PDHE1α

PINK1 is well-established to play a central role in mitophagy. The accumulation of PINK1 leads to the mitophagy of depolarized mitochondria as well as energetically healthy mitochondria harboring unfolded proteins (Jin et al 2013; Narendra et al 2010; Vives-Bauza et al 2010). LONP1 regulates PINK1-induced mitophagy by degradation of PINK1 to prevent mitophagy and by allowing PINK1 levels to rise under the conditions of damaged (depolarized) mitochondria and as part of the mitochondrial unfolded protein response (Jin et al 2013; Thomas et al 2014). Knockdown of Lon in Drosophila results in accumulation of PINK1 (Thomas et al 2014). Fibroblasts from the subject show a severe reduction in PINK1, indicating the subject’s mutation may be a gain of function rather than a loss of function mutation (Figure 3B).

Pyruvate is produced by the glycolysis pathway and becomes fuel for mitochondrial respiration when it is oxidized into acetyl CoA by the pyruvate dehydrogenase complex. Pyruvate has also been shown to be a key driver of mitophagy under conditions of mitochondrial membrane depolarization where elevated pyruvate leads to the stabilization of PINK1 (Park et al 2015). Pyruvate dehydrogenase kinase 4 (PDK4) negatively regulates PDH activity in response to changing nutrient levels by the phosphorylation of PDH subunit PDHE1α and PDK4 has been identified as a co-factor of the pyruvate-PDH mitophagy axis (Crewe et al 2017; Lee et al 2004; Park et al 2015). Moreover, LONP1 has been shown to degrade PDK4 (Crewe et al 2017). Subject fibroblasts show reduced phosphorylation of PDHE1α at Ser293 and Ser300 compared to controls (Figure 3B). This may indicate increased degradation of PDK4 by LONP1 p.R301W.

LONP1 Subject fibroblast dysfunction is not rescued by expression of wild type LONP1

Subject fibroblasts were demonstrated to have decreased OXPHOS proteins from complex I (NDUFB8) and complex IV (COX4I1), mitochondrial ribosome subunits MRPL44 and MRPL11, as well as proteolytic targets PDHE1α, TFAM, and PINK1 (Figures 1 and 3). Overexpression of wild type LONP1 in the subject’s cells did not restore any of these proteins to control levels (Figure 4). Typically, expression of the wild type version of a gene in cells from a patient with loss of function mutations rescues observed cellular dysfunction (Besse et al 2015; Bonnen et al 2013); the fact that we do not observe rescue points to a gain of function mutational effect for p.R301W.

Figure 4.

Wild type LONP1 does not rescue defects in LONP1 Subject cells and overexpression of LONP1 induces dysfunction. A. Electron transport chain proteins NDUFB8 which is part of complex I (C1) and COX4I1 which is part of complex IV (CIV). B. Proteins involved in mtDNA transcription, translation & maintenance TFAM, MRPL44 and MRPL11. C. Mitophagy proteins PINK1 and PDH-E1α Ser300 are not affected by overexpression of LONP1. Subject fibroblasts (S), control fibroblasts (C).

Over-expression of wild type LONP1 in control fibroblast leads to dysfunction

Interestingly, overexpression of wild type LONP1 in control cells appeared to affect some LONP1-related protein levels. NDUFB8, COXIV, MRPL44 and TFAM all showed decreased levels in control cells overexpressing wild type LONP1 while MRPL11, PDHE1α and PINK1 protein levels were unchanged between controls transduced with eGFP and wild type LONP1 (Figure 4).

Oligomerization of LONP1 into stable hexamers is increased by p.R301W

LONP1 protein functions as a hexamer composed of LONP1 monomers. We tested the ability of LONP1 p.R301W to form homo-oligomers (Mut-Mut) and hetero-oligomers with WT LONP1 (WT-Mut). The mutation, p.R301W, formed WT-Mut hetero-oligomers well in excess of WT-WT oligomers (Figure 5A). The p.R301W mutant formed slightly more homo-oligomers than WT-WT. Since the subject is heterozygous for this mutation the hetero-oligomeric state is representative of the patient’s genotype. It was previously reported that mutation LONP1 NM_004793.3: c.2161C>G, p.R721G which was found in the homozygous state in a patient with the classic CODAS clinical presentation reduced the protein’s ability to form stable homo-oligomers (Strauss et al 2015). We also tested the LONP1 p.R721G mutant in our assay and replicated these results with the p.R271G mutant showing decreased amounts of homo-oligomers (Figure 5A). We tested the oligomerization of an additional mutation that was found in a patient who was reported to have a clinical presentation consistent with primary mitochondrial disease and was homozygous for LONP1 c.2282C>T, p.P761L (Nimmo et al 2019). While this mutation showed a mild decrease in WT-Mut hetero-oligomers it showed a large increase in homo-oligomeric hexamers (Figure 5A). Interestingly, the p.P761L homo-oligomer showed one band migrating at the same size as other hexamers (110 kDa) and another slightly larger band, both of which had greater abundance compared to WT-WT hexamers (Figure 5A).

Figure 5.

Oligomerization is changed in LONP1 mutants and TFAM binding appears normal. A. LONP1 oligomerization was assessed through co-transfection of HEK293T cells with expression vectors allowing the concomitant expression of tagged proteins: co-immunoprecipitation of concomitant expression of tagged proteins: 1) V5-LONP1 WT + FLAG-LONP1 mutants, and 2) V5-LONP1s + FLAG-LONP1s. B. Binding of LONP1 binding with TFAM was assessed similarly to oligomerization using co-transfection followed by co-immunoprecipitation of FLAG-TFAM + V5-LONP1 WT and mutants. Colocalization of LONP1 and TFAM was tested through co-transfection of plasmids with FLAG-tagged TFAM and V5-tagged (Green) LONP1 WT or V5-tagged LONP1 mutants. Immunoflourescence shows TFAM-FLAG expression in Red and V5-LONP1 expression in Green. Merged images are shown.

TFAM binding is not disrupted by LONP1 p.R301W

LONP1 binds TFAM when TFAM is not bound to DNA either because mtDNA is depleted or because TFAM is phosphorylated. We tested the ability of WT and mutant LONP1 to bind TFAM using an in vitro co-immunoprecipitation assay. The p.R301W mutation showed comparable TFAM binding to control (Figure 5B). The p.R721G mutant showed slightly diminished TFAM binding which is in agreement with previous findings (Strauss et al 2015) and the p.P761L mutant appeared to have normal TFAM binding. Immunoflourescence studies also showed co-localization of TFAM and LONP1 in these cells for both WT and mutant LONP1 (Figure 5B).

Discussion

We show a dominant mutation in LONP1 c.901C>T, p.R301W causes severe mitochondrial encephalopathy with neonatal onset seizures, little to no neurological development, brain abnormalities including pachygyria, calcifications and progressive atrophy with death at the age of 1. Previous reports of disease caused by LONP1 mutations were all attributed to recessive mutations showing a range of clinical presentations. Several patients with bi-allelic mutations have been described as having CODAS syndrome characterized by a brain, eye, ear, dental and skeletal anomalies (Dikoglu et al 2015; Khan et al 2018; Strauss et al 2015). A single patient with clinical presentation typical of a mitochondrial encephalopathy and confirmed muscle mitochondrial dysfunction was reported to have compound heterozygous missense variants in the ATPase domain and proteolytic domains of LONP1 (Peter et al 2018). Affected siblings with bi-allelic mutations presented with progressive encephalopathy, congenital cataracts, and evidence of pyruvate dehydrogenase complex deficiency in fibroblasts (Nimmo et al 2019).

The novel observation of dominantly acting mutation in LONP1 along with the distinct clinical presentation of the subject reported here, pointed toward a gain of function mutation as opposed to the previously reported presumed loss of function mutations. However, since LONP1 functions as both a chaperone and protease as well as plays an integral role regulating several pathways in response to the cellular environment (Cheng et al 2013; Fukuda et al 2007; Hori et al 2002; Li et al 2017; Rep et al 1996; Suzuki et al 1994; Teng et al 2013), we proposed that a single application of the terminology of Mueller’s morphs may be an oversimplification. We dissected the impact LONP1 p.R301W has on the chaperone and protease functions of the protein as well as LONP1’s ability to make conformational change which is key to LONP1 regulating its activities in response to environmental cues.

LONP1 harnesses the energy generated by hydrolyzing ATP to exert conformational change or transposition of polypeptides (Bota et al 2002; Kereiche et al 2016). This conformational change results in an opening of both the protease and N terminal domains enabling access to the proteolytic chamber (Kereiche et al 2016). Moreover, the N terminal domain is necessary for the correct conformation and consequent activity of the proteolytic domain (Kereiche et al 2016). Furthermore, Kereiche et al showed that substrate binding increases LONP1 ATPase and peptidase activity. We report that the LONP1 N terminal domain mutation p.R301W has a stabilizing effect on the structure, increasing hexamer formation and increasing proteolytic activity. The LONP1 p.R301W missense change replaces a buried, charged residue (R) with an uncharged residue (W). Buried charged surfaces in proteins can have a destabilizing effect which facilitates the ability of a protein to experience dynamic conformational changes, and the replacement of buried charged groups with neutral residues can increase protein thermostability and energetics (Kajander et al 2000). In fact, modeling showed that the folding energy of p.301W is significantly higher than for p.R301 (p-value = 1E-100) (Figure 2C). Moreover, formation of LONP1 WT-mutant p.R301W hetero-oligomers was increased in our in vitro assay which is indicative of an increase in hexamer stability for LONP1 (Figure 5A).

LONP1 is documented to play a role in modulating abundance for multiple proteins that are part of independent cellular processes in response to various environmental cues (Crewe et al 2017; Fukuda et al 2007; Gibellini et al 2018; Jin et al 2013; Quiros et al 2014; Thomas et al 2014; Zurita Rendon et al 2018). These include PINK1, phospho-PDH E1α, TFAM, COX4–1, NDUFB8, MRPL44, and MRPL11 and our subject’s cells showed excessive loss of each of these proteins (Figures 1 and 3). Some of these proteins are proteolytic targets of LONP1 (PINK1, phospho-PDH E1α, TFAM) while others are shown to be affected by LONP1 even when its proteolytic domain is inactive and are subject to LONP1 chaperone function (COX4–1, NDUFB8, MRPL44, and MRPL11). The observed increase in LONP1 hexamer thermostability introduced by p.R301W, and likely decreased ability to make conformational changes, appears to affect its ability to regulate its proteolytic activity as well as its chaperone activity. This profile of dysfunction is distinct from that seen in studies of other disease-causing mutations in LONP1 where some but not all proteolytic and chaperone targets of LONP1 are affected (Nimmo et al 2019; Peter et al 2018; Strauss et al 2015). Moreover, these variant patterns of dysfunction in LONP1 lead to clinical presentations that are distinct from one another. Our dissection of the specific patho-mechanism caused by p.R301W yields insight into the function of the N terminal domain, its role in LONP1 oligomerization and how an apparently excessive oligomerization induced by p.R301W leads to loss of chaperone function and overactive proteolytic function for LONP1.

Methods

Human Subjects

Informed consent for research studies was obtained in accordance with protocols approved by local Institutional Review Boards. Genomic DNA was extracted from peripheral-blood lymphocytes according to standard protocols.

Genetic analyses

Whole genome sequencing of the Subject’s mtDNA was conducted clinically by Baylor Genetics Laboratories at Baylor College of Medicine, Houston, TX. Clinical diagnostic exome sequencing was conducted on the Subject and both of her parents at Developmental & Neurogenetics Laboratories at Medical College of Wisconsin in Milwaukee, WI, according to standard protocols and interpretation of variants was done according to ACMGG guidelines. The LONP1 variant identified through exome sequencing was orthogonally validated and the de novo status confirmed using PCR-based Sanger sequencing.

All genetic alleles studied were annotated in reference to LONP1 NM_004793. Genome Aggregation Database (gnomAD) was queried, 5–17-19, for LONP1, NM_004793 c.901 C>T, p.R301W. The LONP1, NM_004793 c.901 C>T, p.R301W variant was deposited into ClinVar.

Bioinformatic modeling

PHYRE2 was used to determine the three dimensional structure of the LONP1 monomer using sequence NP_004784.2 isoform 1, 959aa, which is the longest of the four human LONP1 isoforms (Kelley et al 2015). PHYRE2 alignments plus the pairwise alignment of LONP1 and LONP2 were used to determine domain structure. The hexameric structure was determined by using Swiss-MODEL to align the human LONP1 sequence to the PDB:4YPL hexamer of M. taiwanensis LonA and PDB:6AHF hexamer of S. cerevisiae Hsp104 (Bienert et al 2017; Waterhouse et al 2018). Missense3D server was used to determine if p.R301W induces structural changes in LONP1 (Ittisoponpisan et al 2019).

The difference in free energy (ΔΔG) between the structure of LONP1 p.R301 and LONP1 p.W301 was determined. The PHYRE2 model for the structure of the full length human WT LONP1 monomer and separately the LONP1 p.W301 were sampled 500 times each by Monte Carlo simulation in an effort to minimize free energy of the structures. Sampling was performed using the macromolecular modeling software Rosetta (Lauck et al., 2010). LONP1 residues 280 to 410 were chosen for sampling because they span p.301 and because PHYRE2 reported high confidence in this region of the model. Once Rosetta sampling was complete, the lowest energy structures found during each Monte Carlo simulation were used to determine the change in free energy (ΔΔG) that occurred upon mutation of LONP1. Free energy estimations on 500 WT LONP1 structures and 500 mutant LONP1 structures were performed using the protein design software FoldX (Schymkowitz et al., 2005). The difference in free energy between WT and mutant LONP1 was calculated by subtracting the estimated free energy of each mutant LONP1 structure by the average estimated free energy of WT LONP1. Mean, median, and standard deviation calculations were performed on both WT and mutant LONP1 data sets. The free energy change was visualized by plotting 50 data points from both data sets centered around their respective median.

Electron transport chain activity

Electron transport chain activities in muscle and fibroblasts were conducted clinically by Center for Inherited Disorders of Energy Metabolism (CIDEM), at Case Western Reserve University School of Medicine, Cleveland, Ohio. Rotenone-sensitive NADH-cytocrome c reductase activity was reported as reflecting ETC complexes I and III. Antimycin-A-sensitive succinate-cytochrome c reductase was reported as reflecting linked activity of ETC complexes II and III. Activity of succinate dehydrogenase was reported as ETC complex II activity. The activity of decylubiquinol-cytocrome c reductase was reported at ETC complex III. Cyanide-sensitive cytochrome C oxidase activity was reported as complex IV.

Primary Fibroblast Cell Culture

Fibroblasts were grown in DMEM High Glucose (Hyclone) supplemented with 15% FBS (Atlanta Biologicals). HEK293T cells were maintained in DMEM High Glucose supplemented with 10% FBS.

Immunoblotting

Fibroblasts were harvested and cell pellets resuspended in RIPA buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 0.5% sodium deoxycholate, 1% Triton X-100, 0.1% SDS, 5 mg/ml leupeptin, 2 mg/ml aprotinin and 0.5mM PMSF). Samples were incubated on ice for 10 minutes and lysates centrifuged at 15000 × g for 10 minutes at 4°C. 10 μg of lysates were resolved by SDS-polyacrylamide gel electrophoresis and transferred to a 0.45 μm PVDF membrane (Millipore). Membranes were immunoblotted using the following antibodies: NDUFB8, SDHA, UQCRC2, COXIV, ATPB, MT-ND5, MT-COX1, MRPL44, MRPL11 (Abcam), TFAM, PINK1 (Cell Signaling), phospho-PDH-E1α Ser293/Ser300 (Millipore Sigma). The membranes were next incubated in stripping buffer (62.5mM Tris, pH 6.8, 2% SDS, 100mM 2-mercaptoethanol) for 15 minutes at 50°C, and then probed with actin (Cell Signaling) as protein loading control. ImageJ software was used to perform the relative quantification of the bands with actin used for normalization.

Quantitation of mtDNA copy number

Mitochondrial genome copy number was determined by real-time quantitative PCR as previously described with minor modifications (Bonnen et al 2013). Briefly, mitochondrial genome copy number was determined relative to the nuclear genome using a region of MTND1 to represent the mitochondrial genome and B2M representing the nuclear genome. The assay utilized StepOne Plus RT-PCR system (Applied Biosystems) and PerfeCTa SYBR Green FastMix ROX (Quanta Biosciences). The assay was performed in triplicate. mtDNA content (mtDNA/B2M ratio) was calculated using the formula: mtDNA content =1/2^ΔCt, where ΔCt=Ct^mtDNA−Ct^B2M.

Lentiviral particle production and cell infections

HEK293T cells were transfected with the plasmids of interest (pLX302-EGFP, pLX302-LONP1 WT, pLX302-LONP1 c.901C>T) and the packaging vectors using Lipofectamine 2000 (Life Technologies). For each transfection,10ug of plasmid of interest was combined with 10.5ug of a mix of the packaging vectors (pMDLg/pRRE, pRSV-Rev, pMD2-VSVG), Opti-MEM (Life Technologies) and Lipofectamine 2000. Cells were then incubated for 6 hours at 37 degrees, 5% CO2. Viral supernatants were harvested at 36, 48, 60 and 72 hours after transfection, pooled together and filtered through a 0.45 μm filter.

10 ml of viral supernatant containing 8 μg/ml of polybrene (EMD Millipore) was added to 80% confluent fibroblasts and incubated overnight. Viral supernatant was replaced with normal growth media and infected fibroblasts were incubated for 48 hours before adding antibiotic (puromycin 1μg/ml) for selection. Cells were then maintained in normal growth media until being harvested for further experiments.

Co-Immunoprecipitation

HEK293T cells were co-transfected with expression vectors (pLenti6.3) allowing the concomitant expression of tagged proteins: 1) FLAG-TFAM + V5-LONP1 WT and mutants, 2) V5-LONP1 WT + FLAG-LONP1 mutants, and 3) V5-LONP1s + FLAG-LONP1s. 1 μg of each construct was transfected using Lipofectamine 3000 (Invitrogen). 48 hours after transfection, the cells were harvested and cell pellets lysed in the following ice-cold buffer: 40 mM HEPES (pH 7.5), 120 mM NaCl, 1mM EDTA, 10 mM glycerophosphate, 5 mM NaF, 1.5 mM Na3VO3, 0.3% CHAPS containing protease inhibitors for 30 minutes on ice. Lysates were then centrifuged for 20 minutes at 15000 × g at 4°C and the supernatants collected. The total protein concentration of the lysates was measured with BCA protein assay kit (Pierce). For each immunoprecipitation, 350 μg of total protein extract was pre-cleared with protein A/G sepharose beads (Pierce) for 1 hour at 4°C on a rotisserie. After pre-clearing, anti-V5 or anti-FLAG antibodies were added and the lysates were incubated for 2 hours at 4°C on a rotisserie. Then protein A/G sepharose beads were added, and lysates were incubated for 1 hour at 4°C on a rotisserie. After the last incubation, the protein A/G sepharose beads were pelleted by centrifugation for 1 minute at 2500 × g at 4°C, and the beads were washed 5 times with lysis buffer. After the last wash, Laemmli loading buffer was added to the beads, and the samples were boiled for 5 minutes prior of being resolved by SDS-polyacrylamide gel electrophoresis and transferred to a 0.45 μm PVDF membrane (Millipore). Membranes were immunoblotted using either anti-FLAG or anti-V5 antibodies.

TFAM and LONP1 intracellular localization

HEK293T cells were co-transfected with FLAG-tagged TFAM and V5-tagged LONP1 WT and mutants. 48 hours after transfection, the cells were washed twice in pre-warmed 1X-PBS then fixed in 4% paraformaldehyde for 10 minutes at room temperature. After fixation, the cells were washed again 3 times in 1X-PBS and permeabilized in 0.25% Triton X-100 (Sigma) for 10 minutes at room temperature. After permeabilization, the double staining was done as follow: blocking for 30 minutes at room temperature with 10% goat serum, incubation with rabbit anti-FLAG antibody (Cell Signaling) for 1.5 hours at room temperature, 3 washes with 1X PBS, incubation with secondary Alexa Fluor 555 (Red) goat anti-rabbit antibody (Life Technologies) for 45 minutes at room temperature, 3 washes with 1X PBS, incubation with mouse anti-V5 antibody (Life Technologies) for 1.5 hours at room temperature, 3 washes with 1X PBS, incubation with secondary Alexa Fluor 488 (Green) goat anti-mouse antibody (Life Technologies) for 45 minutes at room temperature, 3 washes with 1X PBS. After the last wash, the coverslips were mounted on plain microslides (VWR, 48300047) and the cells were visualized using a Nikon Eclipse 90i microscope equipped with TRITC and FITC filters (at 60X magnification). The images were processed with NIS-elements v3.0 software.

Highlights.

A heterozygous de novo mutation in LONP1:c.901C>T,p.R301W causes neonatal onset encephalopathy with seizures, pachygyria and microcephaly.

Patient fibroblasts showed decreases in known proteolytic targets of LONP1 as well as loss of suspected chaperone targets indicating excessive LONP1 proteolytic activity with a concomitant decrease of LONP1 chaperone activity.

The LONP1 N-terminal domain is involved in hexamer stability of LONP1, LONP1:c.901C>T,p.R301W causes increased LONP1 stability and excessive hexamer formation.

The ability to make conformational changes is necessary for LONP1 to regulate proper functioning of both its proteolytic and chaperone activities.