PARKINSON'S

FIELD OF THE INVENTION

The present invention relates to methods of diagnosing Parkinson's or assessing the risk of developing Parkinson's in a human subject. The invention also relates to methods of assessing the extent and rate of Parkinson's progression, and methods which can be used to assist in a clinical diagnosis of Parkinson's, or to identify subjects with a predisposition to Parkinson's. The methods of the invention are based on isolating mitochondrial DNA (mtDNA) from a subject and identifying specific heteroplasmic mtDNA mutations associated with Parkinson's. Specific mtDNA variants in the ND5 region and the non-coding control region have been found to be strongly associated with

Parkinson's. The invention also relates to methods of identifying novel heteroplasmic mtDNA mutations associated with Parkinson's by comparing the frequency of

heteroplasmic mutations in a Parkinson's patient with the frequency of mutations in a healthy relative of the patient who shares the same inherited mtDNA. BACKGROUND TO THE INVENTION

Parkinson ^' s is an umbrella term which includes disorders that can be classified into three main types: familial Parkinson's disease; idiopathic Parkinson's disease (also called sporadic Parkinsons disease); and Parkinsonian disorders. Parkinson's disease is a chronic, degenerative neurological disorder characterised by tremor, muscular rigidity, and slow, imprecise movement, which primarily affects late-middle-aged and elderly people. There are no laboratory or genetic tests available at present, which makes Parkinson's disease difficult to diagnose. Currentiy, diagnosis is performed by a specialist clinician who will examine a patient for any physical signs of Parkinson's disease. The Unified Parkinson's Disease Rating Scale (UPDRS) is used by clinician's to diagnose, measure and monitor Parkinson's disease. Although the scale is a quantitative measure of Parkinson's, it relies upon qualitative assessment, which is imprecise and subjective. There is currently no screening test available to detect the early onset of Parkinson's disease.

Familial Parkinson's disease is inherited and has been linked to mutations in specific genes in a patient's nuclear DNA (e.g. LRRK2, PARK7, PINK1 , PRKN, SNCA). it comprises approximately 10% of all Parkinson's cases. Idiopathic Parkinson's disease has no known specific cause, although a number of different hypotheses have been advanced. It is more common than familial Parkinson's, comprising approximately 85% of cases. Both familial and idiopathic forms of the disease are characterised by similar motor symptoms, including akinesia and rigidity of movement. Both of these forms are responsive to L-Dopa therapy, although the effectiveness of this drug wears off with time.

The remaining patients (approximately 25%) who report at least some symptoms of Parkinson's to clinicians have different underlying causes of disease and present these clinicians with a considerable diagnostic problem. These patients are not in general responsive to L-Dopa therapy. Indeed, such therapy may cause severe, undesirable side- effects. There are several sub-types of these Parkinson's-like disorders (also called

Parkinsonian disorders) that have been identified. These include so-called "Parkinson's plus" diseases (e.g. PSP - progressive supranuclear palsy, MSA - multiple system atrophy, CBD - corticobasal degeneration, DLB - dementia with Lewy bodies) together with a number of other types of Parkinsonism (e.g. Essential tremor, Wilson's disease, Whipple's disease).

Parkinson's disease is associated with a loss of neurons in the substantia nigra in the brain, resulting in a reduction of dopamine, a neurotransmitter crucial to physical movement and coordination. Studies by Braak and others have demonstrated that by the time a patient presents with the classical motor symptoms of Parkinson's, a large proportion (up to 70%) of these neurons have been lost from the substantia nigra. There is no convenient screening diagnostic currently available to detect this early loss of neurons in the brain.

There are a number of drugs being developed to treat Parkinson's disease but as yet none has gained regulatory approval. For these drugs to be effective, if is desirable that treatment of the patient starts as early as possible in the development of the disease, before too many dopamine-producing neurons have been lost. Hence, there is a need for a screening diagnostic that can detect the early onset of Parkinson's disease which can be used as a screening tool for subjects that are at risk of developing the disease but who have not yet developed the motor symptoms.

Loss of neurons in the substantia nigra is a hallmark of Parkinson's disease.

Before death, these neurons exhibit reduced energy production and are found to contain mutant copies of mitochondrial DNA. Mitochondria are energy generating organelles and the principal energy source of neurons. Each neuron contains thousands of mitochondria and each mitochondrion possesses its own genome, which encodes 13 proteins critical for energy production. It has been shown that chemically inhibiting the function of these proteins causes permanent symptoms of Parkinson's disease (Fahn S. N Engl J Med, 1996; Porras G et al., Cold Spring Harb Perspect Med, 2012; Cannon JR et al., Neurobiol Dis, 2009). mtDNA damage is observed in the mitochondria of neurons from patients with both early and late stage Parkinson's disease, which suggests there is a disease mechanism linking mtDNA damage with the Parkinson's phenotype (Bender et al., Nat Genet, 2006; Lin et al., Ann Neurol, 2012; Parker et al., Biochem Biophys Res Commun, 2005).

It has been shown that early onset Parkinson's may be caused by mutations in two proteins, Pinkl and Parkin, which target faulty mitochondria for destruction in a process coined 'mitophagy' (Youle R. J., Narendra D. P. Nat Rev Mol Cell Biol, 2011). It has been further demonstrated that introduction of mitochondrial DNA mutations into an asymptomatic Parkin knockout mouse model causes loss of dopaminergic neurons in the substantia nigra, a classical symptom of Parkinson's disease (Pickrell AM et al., Neuron, 2015).

Although dopaminergic neurons are the most affected by mitochondrial dysfunction, mitochondrial dysfunction has also been demonstrated in muscle, platelets, lymphocytes and fibroblasts of Parkinson's patients (Swerdlow et al., Antioxid Redox Signal, 2012). Researchers have discovered that white blood cells from patients with Parkinson's disease also contain mitochondria with reduced energy production. It has been theorized that over a period of time, mitochondria containing mutant copies of mtDNA arise in white blood cells, and spread to the rest of the body via the bloodstream. Furthermore, multiple studies have concluded that non-neuronal mtDNA from Parkinson's patients is genetically different than mtDNA from healthy individuals suggesting mtDNA could provide an accessible biomarker for diagnosis (Swerdlow et al., Antioxid Redox Signal, 2012).

As mentioned above, current diagnosis using the UPDRS is imprecise, and ultimately this assessment does not allow for a) an accurate diagnosis of Parkinson's disease or a Parkinsonian disorder, nor does it allow for b) an early diagnosis prior to the onset of deteriorating motor functions. The use of mitochondria-linked biomarkers (e.g. mtDNA) as Parkinson's biomarkers has come close to addressing this first problem. For example, researchers at La Trobe University have developed the first blood test to measure markers of abnormal mitochondrial metabolism in blood and the method has been used to successfully discriminate 29 Parkinson's disease patients from 9 healthy counterparts ('World-first blood test for Parkinson's', 20 April 2016 press-release, La Trobe University).

It has been suggested that a mutant copy of mtDNA may have the capacity to reproduce faster, such that over time the share of wild-type copies will be reduced.

However this process can take many years and significant accumulation of mutant copies may not appear until a person is in their sixties or seventies. In many cases a mutant copy is harmless and does not lead to a functional change in the corresponding protein. However, in a proportion of cases it is harmful, and leads to the production of a nonfunctional protein that blocks the energy producing process. The mitochondrial theory of ageing proposes that progressive accumulation of somatic mutations in mtDNA during a lifetime leads to an inevitable decline in mitochondrial function. As a consequence, energy production by the mitochondria may be severely reduced. In neurons, which require a large and continuous supply of energy, this can cause cell death. Recent studies have reported that the brain tissue of Alzheimer's patients has an enrichment of mitochondrial DNA mutations compared to age matched control's, some of which are systemic (Chen et al., PLoS One, 2016; Coskun et al., J Alzheimer's Dis, 2014). Similarly, it has been shown mitochondrial DNA deletions accumulate in the substantia nigra dopaminergic neurons of aged individuals and are increased in individuals with

Parkinson's (Bender et al., Nat Genet, 2006).

The term heteroplasmy is used to describe a mixture of two or more mitochondrial genotypes within a cell or an individual, i.e. the presence of both wild-type and mutant copies. One theory suggests that the proportion of mutant mtDNA copies determines the severity of expression of many mitochondrial disorders, with progressively more severe symptoms emerging as the ratio of mutated to wild-type genomes rises. Therefore, in the presence of heteroplasmy, there may be a threshold level of mutation that is important for the clinical expression of Parkinson's disease. High throughput sequencing has been used to identify mtDNA mutations in Parkinson's patients by comparison of the patient genome with a healthy control genome (Hudson et al., PLoS Genet, 2014). However, a key issue with identifying mtDNA mutations associated with Parkinson's disease is that there is a significant amount of variation in mtDNA sequence between individuals. As a result, specific mutations responsible for the mitochondrial dysfunction in Parkinson's disease have been difficult to identify, and to date, no solid correlations with the

Parkinson's disease phenotype have been found.

Research conducted at University of Virginia has considered the role of low abundance heteroplasmy in mitochondrial diseases. Low frequency heteroplasmic mutations have been identified in Parkinson's patients in a restricted region of mtDNA. Dr William Davis Parker describes a method of isolating mtDNA from frontal cortex and blood samples, before amplifying and sequencing a subset of genes, in particular the ND5 gene, in order to classify Parkinson's disease and healthy patient samples based on low abundance heteroplasmic levels (Parker et al., Biochem Biophys Res Commun, 2005).

Presently there is no solution for diagnosing, measuring and monitoring

Parkinson's disease that is truly quantitative, precise and objective. In particular, there is a need for a diagnostic method that can be performed early in a subject's lifetime, and prior to the emergence of classical symptoms of the disease. Furthermore, there is no reliable method for identifying new potentially pathogenic mutations associated with Parkinson's disease. The present invention fulfils this need.

SUMMARY OF THE INVENTION

The present inventors have identified heteroplasmic mtDNA variations that are common to patients with Parkinson's. These heteroplasmic mutations share a number of common features, allowing the inventors to develop a robust method of identifying further mutations that share the same features, and therefore have a high probability of being associated with Parkinson's. The inventors have further identified a non-invasive and objective method of diagnosing a subject as having or being likely to develop Parkinson's. Importantly, the methods allow for an early diagnostic or screening test prior to the onset of clinical symptoms of the disease.

Therefore in a first aspect, the present invention provides a method of identifying a subject as having or as being at risk of developing Parkinson's, comprising the steps of:

(i) sequencing the mtDNA isolated from a sample obtained from a

subject;

(ii) identifying the presence and frequency of heteroplasmic mtDNA mutations that occur in the subject sample, by comparison to a reference sample;

(iii) wherein the subject is identified as having or as being at risk of developing Parkinson's if at least one heteroplasmic mtDNA mutation identified in step (ii):

a) occurs in the subject sample with a heteroplasmic frequency of between 7% and

90%; and

b) occurs with an occurrence frequency of less than 2.5% of the mitochondrial

sequences in a human mtDNA database.

In certain embodiments, the subject is identified as having or as being at risk of developing Parkinson's when the at least one heteroplasmic mtDNA mutation is a non- synonymous mutation located between nucleotide positions 12337 and 14673 , and/or is located between nucleotide positions 1 and 576 and/or is located between 16024 and 16569.

In certain embodiments, the subject is identified as having or as being at risk of developing Parkinson's when the at least one heteroplasmic mtDNA mutation is a non- synonymous mutation located in the MT-ND5 gene, or is located in the non-coding control region.

In certain embodiments the methods may comprise an additional step wherein the heteroplasmic frequency and occurrence frequency in parameters a) and b) are optimised using an algorithm performed on a data set that has been generated by conducting the following steps on a plurality of samples:

(i) sequencing the mtDNA isolated from a sample obtained from a

patient with Parkinson's;

(ii) identifying the presence and frequency of heteroplasmic mtDNA mutations in the test sample, by comparison to a reference sample;

wherein a heteroplasmic mtDNA mutation identified in the test sample in step (ii) occurs with a heteroplasmic frequency of between 7% and 90%; and occurs with an occurrence frequency of less than 2.5% of the mitochondrial sequences in a human mtDNA database.

In certain embodiments the algorithm is a supervised learning algorithm such as a classification algorithm, or a regression algorithm.

In another aspect, the present invention provides a method of detecting a pathological heteroplasmic mtDNA mutation in a subject, said method comprising the steps of:

(i) isolating and sequencing the mtDNA from a sample obtained from a

subject;

(ii) detecting the presence and frequency of heteroplasmic mtDNA mutations in the subject sample, by comparison to a reference sample;

(iii) wherein a heteroplasmic mtDNA mutations is pathological if:

a) it occurs in the subject sample with a heteroplasmic frequency of between 7% and 90%; and

b) it occurs with an occurrence frequency of less than 2.5% of the mitochondrial

sequences in a human mtDNA database.

In certain embodiments the pathological heteroplasmic mtDNA mutations are associated with Parkinson's.

In certain embodiments, the present invention provides a method of diagnosing and treating a subject as having or as being at risk of developing Parkinson's, comprising the steps of:

(i) isolating and sequencing the mtDNA from a sample obtained from a

subject;

(ii) identifying the presence and frequency of heteroplasmic mtDNA mutations in the subject sample, by comparison to a reference sample; (iii) diagnosing the subject as having or as being at risk of developing

Parkinson's when at least one heteroplasmic mtDNA mutation identified in step (ii) occurs in the subject sample with a heteroplasmic frequency of between 7% and 90%; and occurs with an occurrence frequency of less than 2.5% of the mitochondrial sequences in a human mtDNA database; and

(iv) treating a subject with an effective amount of a therapy designed to

prevent or slow progression of Parkinson's.

It is a second aspect of the invention to provide a method of identifying a heteroplasmic mtDNA mutation associated with Parkinson's, comprising the steps of:

(i) sequencing the mtDNA isolated from a sample obtained from a patient with Parkinson's;

(ii) identifying the presence and frequency of heteroplasmic mtDNA mutations in the test sample, by comparison to a reference sample;

(iii) wherein a heteroplasmic mtDNA mutation identified in step (ii) is

associated with Parkinson's if:

a) the mutation occurs in the test sample with a heteroplasmic frequency of between 7% and 90%; and

b) the mutation occurs with an occurrence frequency of less than 2.5% of the

mitochondrial sequences in a human mtDNA database.

In certain embodiments, a heteroplasmic mtDNA mutation identified in step (ii) is associated with Parkinson's if it is a non-synonymous mutation located between nucleotide positions 12337 and 14673 , and/or is located between nucleotide positions 1 and 576 and/or is located between 16024 and 16569.

In certain embodiments a heteroplasmic mtDNA mutation identified in step (ii) is associated with Parkinson's if it is a non-synonymous mutation located in the MT-ND5 gene or it is located in the non-coding control region.

In certain embodiments a heteroplasmic mtDNA mutation identified in step (ii) is associated with Parkinson's if there is a difference in heteroplasmic frequency in the test sample and the heteroplasmic frequency in the reference sample.

In certain embodiments the reference sample is a sample from a healthy relative who shares related mtDNA with said patient with Parkinson's. If the patient with

Parkinson's is female, the healthy relative is a biological child, mother, or sibling, preferably a biological child. If the patient with Parkinson's is male, the healthy relative is a sibling, or mother, preferably a biological sibling.

In certain embodiments the methods may comprise an additional step wherein the heteroplasmic frequency and occurrence frequency in parameters a) and b) are optimised using an algorithm performed on a data set that has been generated by conducting the same method on a plurality of samples. In certain embodiments the algorithm is a supervised learning algorithm such as a classification algorithm, or a regression algorithm.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 depicts three mitochondrial DNA 'maps' or 'heteromaps' for a mother with Parkinson's (left), an unaffected son (middle) and the difference, or delta, between the two related subjects (right). Lines projecting inward from the circumference indicate the presence of a nucleotide variant in the subject (compared to the Cambridge Reference Sequence). The length of each line represents the heteroplasmic frequency of the nucleotide variant, with no line indicating no variant and a line that reaches the centre indicating 100% variant. Variants that meet the following three criteria are coloured grey, and their heteroplasmic frequency is indicated in small circles, and their position is indicated with outer numerical labels:

(1) non-synonymous variant located in the MT-ND5 gene or variant located in the non-coding control region;

(2) threshold heteroplasmic frequency of between 7% and 90%;

(3) threshold occurrence frequency of less than 2.5% of the mitochondrial sequences in a human mtDNA database.

For ease of interpretation, the difference, or delta, in heteroplasmic frequency between two subjects is represented as a positive frequency, even in cases of a negative frequency. Data associated with each heteromap is displayed in a label (top left). Data from each line of the label can be understood with the following key:

(Linel) P=Parkinsons, H=Healthy;

(Line2) age of subject;

(Line3) relationship of subject to accompanying heteromap;

(Line4) sample type;

(Line5) sample collection date.

Figure 2 depicts three heteromaps for a sister with Parkinson's (left), an unaffected sister (middle) and the difference, or delta, between the two related subjects (right).

Figure 3 depicts three heteromaps for a mother with Parkinson's (left), an unaffected son (middle) and the difference, or delta, between the two related subjects (right). Figure 4 depicts six heteromaps from six healthy individuals. Figure 5 depicts five heteromaps from five healthy individuals.

Figure 6 depicts three heteromaps from three subjects with Parkinson's.

Figure 7 depicts three heteromaps for a mother with Parkinson's (left), an unaffected daughter (middle) and the difference, or delta, between the two related subjects (right).

Figure 8 depicts three heteromaps for an unaffected mother (left), an unaffected son (middle) and the difference, or delta, between the two related subjects (right).

DETAILED DESCRIPTION

A. Definitions

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by the ordinary person skilled in the art to which the invention pertains. Without limiting any term, further clarifications of some of the terms used herein are provided below.

As used herein, the term "mitochondrial DNA" or "mtDNA" refers to an

extranuclear double-stranded DNA found exclusively in mitochondria that in most eukaryotes is a circular molecule and is maternally inherited.

As used herein, the term "heteroplasmic" or "heteroplasmy" refers to the presence of two or more mitochondrial DNA sequences of different genotypes, i.e. the presence of both wildtype and mutant sequences within a cell or individual. The proportion of mutant mitochondrial DNA molecules may determine the severity of expression of many mitochondrial disorders.

As used herein, the term "homoplasmic" or "homoplasmy" refers to the presence of a uniform collection of mitochondrial DNA; either all copies of mtDNA are wild-type or all copies of mtDNA are mutant.

As used herein the term "heteroplasmic frequency" refers to the percentage of mtDNA copies that contain mutant sequences. The term can be used to describe the percentage of heteroplasmic copies in respect of a specific nucleotide position.

Alternatively, the term may be used to describe the number of heteroplasmic mutations at different nucleotide positions within a mitochondrial genome. Most preferably, as used herein the term "heteroplasmic frequency" refers to the percentage of heteroplasmic copies in respect of a specific nucleotide position. As used herein the term "delta mutation frequency" refers to the difference in heteroplasmic mutation frequency in a sample from one individual compared to a sample from another. As used herein the heteroplasmic frequency may refer to a pre-determined threshold reciting the percentage of heteroplasmic mtDNA copies within a specified range. Such a threshold can be determined by one of skill in the art. It is understood that the particular pre-determined threshold is dependent on the specific data set which is analysed, and that the threshold can vary as the data set varies. Because the threshold levels vary according to the data set analysed they are not fixed values but can be adapted according to the present invention by one skilled in the art. It is within the level of one of skill in the art to determine the threshold of heteroplasmic frequency required to identify a mtDNA mutation associated with Parkinson's. It is within the level of one of skill in the art to determine the threshold of heteroplasmic frequency required to identify a subject as having or as being at risk of developing Parkinson's. For example, a heteroplasmic mtDNA mutation may have a frequency that falls within the threshold heteroplasmic frequency of between 7% and 90%.

As used herein the term "occurrence frequency" may refer to a pre-determined threshold reciting the frequency with which a specific heteroplasmic mtDNA mutation occurs in a human mtDNA database. Such a threshold can be determined by one of skill in the art. It is understood that the particular pre-determined threshold is dependent on the specific data set which is analysed, and that the threshold can vary as the data set varies. Because the threshold levels vary according to the data set analysed they are not fixed values but can be adapted according to the present invention by one skilled in the art. For example, a specific mtDNA mutation may occur in less than 2.5% of the sequences in a human mtDNA database. Occurrence frequency may refer to how many of the >30,000 individuals in the Mitomap database harbour a detectable variant at a specific nucleotide position. An occurrence frequency of less than 2.5% means that less than 2.5% of the >30,000 individuals in the Mitomap database harbour a variant at a specific nucleotide position.

As used herein the term "mutation" refers to any variation in nucleotide sequence, which may be a single point mutation, an insertion, a deletion, a rearrangement, or any other variant. A point mutation may result in one of three possible base substitutions. A mutation may result in an amino acid change. The following notation shall be adhered to when describing mutations: For example, t:c reflects a change from base 't' to base 'c';— -:caca reflects an insertion of bases 'caca'; c:- reflects a deletion of base 'c'. It is to be understood that the same notation style can be used with different base letters and combinations.

As used herein the term "MT-ND5" is used interchangeably with "ND5" and refers to the ND5 gene located in the human mitochondrial genome. The ND5 gene is located between nucleotide positions 12337 and 14148 (as shown in SEQ ID N0.1). The gene encodes the ND5 protein (NADH-ubiquinone oxidoreductase chain 5), which is a subunit of NADH dehydrogenase. The T-ND5 gene produces a 67 kDa protein composed of 603 amino acids.

As used herein the term "control region" or "non-coding control region" refers to the region of the human mitochondrial genome between nucleotide positions 1 and 576 (as shown in SEQ ID NO. 3), and between nucleotide positions 16024 and 16569 (as shown in SEQ ID NO. 4). The nucleotide positions correspond to the Revised Cambridge Reference Sequence of the Human Mitochondrial DNA (rCRS Genbank number

NC_012920).

As used herein the term "MT-ND6" is used interchangeably with "ND6" and refers to the ND6 gene located in the human mitochondrial genome. The ND6 gene is located between nucleotide positions 14149 and 14673 (as shown in SEQ ID NO. 6). The gene encodes the ND6 protein ((NADH-ubiquinone oxidoreductase chain 6), which is a subunit of NADH dehydrogenase. The MT-ND6 gene produces a 19 kDa protein composed of 172 amino acids.

As used herein the term "Parkinson's" is used as an umbrella term to include true Parkinson's disease, also known as idiopathic Parkinson's disease, and all Parkinsonian disorders. The term Parkinson's disease as used herein refers to a chronic, degenerative disease that involves problems of movement control, tremor, rigidity, bradykinesia in all kinds of movements such as walking, sitting, eating, talking, etc., as well as postural instability. Symptoms of the disease are clearly associated with the selective

degeneration of dopaminergic neurons in the substantia nigra. The dopaminergic deficit induces a consequent loss of striatal neurons causing a variety of cytological changes including a-synuclein aggregation in so-called Lewy bodies.

Parkinsonian disorder is used to define other disorders that present with

Parkinson's disease-like motor symptoms. Examples of Parkinsonian disorders include Progressive Supranuclear Palsy, Multiple System Atrophy, Corticobasal Degeneration, Dementia with Lewy Bodies, and Gangliosidosis. Individually these are relatively uncommon, but collectively they may account for 25-50% of initial Parkinson's diagnoses. In addition, conditions where the body has responded to physical damage or an external insult by presenting with Parkinson's disease-like motor symptoms are often included in definitions of "Parkinson's". Such conditions include normal pressure hydrocephalus, vascular and drug induced Parkinson's.

The term "patient with Parkinson's" as used herein may refer to a patient with either Parkinson's disease or a Parkinsonian disorder. Patients suitable for use in the present invention include those with idiopathic Parkinson's and Parkinsonion disorders such as Progressive Supranuclear Palsy, Multiple System Atrophy, corticobasal degeneration, dementia with Lewy bodies, gangliosidosis, normal pressure

hydrocephalus Parkinson's, vascular Parkinson's and drug induced Parkinson's.

Particularly suitable are those with idiopathic Parkinson's and Parkinsonion disorders such as Progressive Supranuclear Palsy, Multiple System Atrophy, corticobasal degeneration, dementia with Lewy bodies and gangliosidosis. The invention is even more suitable for use with those with idiopathic Parkinson's disease.

Classical idiopathic Parkinson's disease can only be "confirmed" post-mortem by detection of Lewy bodies. Therefore, a patient is selected based on the presence of one or more symptoms, and/or one or more risk factors typically associated with Parkinson's. Preferably, for the methods of the present invention, a patient is selected who is aged 65 years or older, and is responsive to Levodopa based medication, and thus is highly likely to have Parkinson's disease. The patient may also present with one or more symptoms consistent with Parkinson's, for example symptoms defined in the UPDRS. A patient with Parkinson's disease may have the disease in any of the stages according to the Braak staging: stage 1 : the affected area is the dorsal motor nucleus and/or intermediate reticular zone; stage II: the affected area extends to coreuleus locus and to the nucleus raphes; stage III: the affected area extends to the midbrain, in particular the substantia nigra pars compacta; stage IV: the affected area extends to the transentorhinal region of the anteromedial temporal mesocortex and alocortex; stage V: the affected area extends to the insular cortex, the cingulate cortex and the temporal gyrus; stage VI: the affected area extends to frontal and parietal area of the cortex.

As used herein, the term "idiopathic" refers to a disease or condition which arises spontaneously or for which the cause is unknown. The cause of Parkinson's in most individuals is currently unknown but is believed to involve both genetic and environmental factors.

As used herein the term "associated with Parkinson's" refers to a heteroplasmic mutation that is strongly correlated with either expression of the Parkinson's phenotype, or with the risk of developing Parkinson's later in life. The statistical certainty with which a specific mutation may be considered to be associated with Parkinson's may vary depending on the mutation, in particular depending on the location and/or the

heteroplasmic frequency of the mutation. As used herein the term "diagnosing" and "diagnosed" is used in a very broad sense which encompasses clinical diagnosis of Parkinson's in subjects who present with other symptoms consistent with Parkinson's, including subjects who have been clinically diagnosed according to the UPDRS, and also diagnosis of early stage, pre-clinical, or prodromal Parkinson's, and also prediction of the risk of developing Parkinson's in an asymptomatic subject or a subject exhibiting prodromal non-motor symptoms of

Parkinson's. Therefore, "diagnosis" may refer to a method of predicting the likelihood of a subject having Parkinson's or of predicting the likelihood of a subject being at risk of developing Parkinson's. As with other diagnostics, the prediction may not be 100% accurate but a subject can be diagnosed as having or being at risk of developing

Parkinson's with a high degree of confidence. This level of confidence can be improved according to the methods of the invention. For example, if a subject presents with a mutation that has already been identified as being associated with Parkinson's, then the subject can be diagnosed with the highest degree of confidence. Similarly, if a subject presents with a mutation that satisfies a number of specific criteria, as described elsewhere herein, then the level of confidence with which the subject can be diagnosed increases. The results of the diagnostic method can also be combined with other factors, for example, age, environmental risk factors, prodromal symptoms, or symptoms defined in the UPDRS, in order to improve the accuracy of diagnosis. As more mutations associated with Parkinson's are identified in a plurality of samples, the thresholds for each criteria can be refined, and the accuracy of the diagnosis increased.

As used herein the term "subject" refers to a mammal, preferably a human. The subject may be either a male or a female. The subject may be asymptomatic for

Parkinson's, or the subject may exhibit one or more symptoms consistent with

Parkinson's. In certain embodiments, the subject may be a patient, where a patient is an individual who is under medical care and/or actively seeking medical care for treatment of Parkinson's.

As used herein the term "sample" refers to either a tissue sample or a mtDNA sample. The tissue sample may be obtained from, for example, skin, saliva, urine, hair, blood, earwax, cheek swab, tongue scrape or cerebrospinal fluid. The mtDNA sample may be isolated from the nuclear and bacterial DNA also in the sample. The term "subject sample" is used to refer to a sample obtained from a subject to be diagnosed. The term "test sample" is used to refer to a sample obtained from a patient with Parkinson's. The term "reference sample" refers to a sample used for comparison with the test or subject sample. The reference sample may be the Revised Cambridge Reference Sequence of the Human Mitochondrial DNA. In alternate embodiments, the reference sample may be a sample from a healthy control. In other embodiments, the reference sample may be a sample from a healthy relative of the patient with Parkinson's who shares the same related mtDNA.

As used herein the term "healthy control" refers to a subject who does not have or is not suspected of having Parkinson's. Preferably, the healthy control also does not have another age-related disorder, such as Alzheimer's disease.

As used herein the term "healthy relative" refers to a biological relative of a patient with Parkinson's and shares the same related mtDNA. The healthy relative does not have or is not suspected of having Parkinson's. Preferably, the healthy relative also does not have another age-related disorder, such as Alzheimer's disease.

As used herein the term "identify" and "identifying" may be used interchangeably with 'detect' or 'detecting'.

As used herein the term "inherited mtDNA" may be used interchangeably with "related mtDNA" and refers to mtDNA that is inherited from one individual to another biologically related individual. mtDNA is inherited from the mother only. Therefore, a mother and her biological children share the same inherited or related mtDNA.

Furthermore, an individual and their biologically related siblings share the same inherited mtDNA as each other, since they each inherited the mtDNA from the same mother.

As used herein the term "non-synonymous mutation" refers to a mutation that results in an amino acid change. In contrast a synonymous mutation occurs when a mutation changes the nucleotide sequence but results in translation of the same amino acid sequence due to the degeneracy of the genetic code in which different codons may specify the same amino acid.

As used herein the term "parameter" and "criteria" are used interchangeably to describe the features of a heteroplasmic mtDNA mutation required to define the mutation as being associated with Parkinson's, and/or to diagnose a subject as having or being at risk of developing Parkinson's. The parameters or criteria include but are not limited to: the location of the mutation; the heteroplasmic frequency of the mutation; the occurrence of the mutation in a human mtDNA database; and the difference in heteroplasmic frequency between the test sample and a reference sample. The thresholds of these parameters may vary according to the invention.

As used herein the term "data set" and "database" are used interchangeably to refer to the data obtained from analysing heteroplasmic mtDNA mutations in a plurality of samples obtained from both patients with Parkinson's disease and samples obtained from healthy subjects.

As used herein the term "heteromap" refers to the graphical representation of the heteroplasmic mtDNA mutations identified in either a patient with Parkinson's, in a healthy control, in a healthy relative of said patient, or in a subject to be diagnosed. The heteromap displays the percentage of heteroplasmy for each identified heteroplasmic mutation, and displays the location of each mutation within the mitochondrial genome. The length of the line indicates the percentage of heteroplasmy. The data from a heteromap of a healthy relative can be subtracted from the data from a heteromap of a patient with Parkinson's to generate a delta heteromap, which displays the differences in heteroplasmic mtDNA mutations, i.e. which displays the heteroplasmic mtDNA mutations most likely to be associated with disease.

As used herein the terms "algorithm" or "classification algorithm" are used in the same way as is commonly understood by one of skill in the art, to refer to a process or set of rules to be followed in calculations or other problem-solving operations, especially by a computer. In the terminology of machine learning, classification is considered an instance of supervised learning, i.e. learning where a training set of correctly identified

observations is available. In the present invention, the algorithm is used to identify to which of a set of categories a new observation belongs, on the basis of a training set of data containing observations whose category membership is known. For example, a subject presenting with a heteroplasmic mtDNA mutation can be diagnosed as having or being at risk of developing Parkinson's based on the presence of an observed set of characteristics common to the mtDNA mutations identified in patients with Parkinson's. It is to be understood that the present invention is not limited to a particular algorithm and that there are numerous types of algorithm suitable for implementing the methods of the present invention.

As used herein the term "optimised" is used to refer to a process of improving, or refining, or adapting, the methods of the present invention. The optimisation process is used to adapt the thresholds of the parameters depending on information in a current data set. The optimisation process may be performed by an algorithm. The optimisation process may be an iterative process.

B. Diagnostic method In a first aspect, the present invention provides a method of identifying a subject as having or as being at risk of developing Parkinson's. The methods enable early and accurate diagnosis of a subject prior to the appearance of clinical symptoms. As with other diagnostic methods, the methods of the invention may not be 100% accurate in terms of diagnosing and predicting the risk of a subject having or developing Parkinson's. The methods of the invention aim to provide a quantitative and objective genetic test that may be used to predict whether a subject has or is at risk of developing Parkinson's disease with at least 50%, at least 60%, at least 70%, at least 75%, at least 80%, at least 85%, at least 90%, at least 95%, at least 99% accuracy.

The methods of the invention are based on identifying the presence and frequency of heteroplasmic mtDNA mutations wherein the mutations fulfil a specific set of criteria relating to location and heteroplasmic frequency. For example, the present inventors have found that heteroplasmic mtDNA mutations in the ND5 region and the non-coding control region are strongly associated with Parkinson's disease.

The present diagnostic method is carried out on a tissue sample obtained from the subject. The tissue sample may be from a non-neuronal source including, but not limited to, saliva, skin, hair, urine, blood, earwax, cheek swab, tongue scrape or cerebrospinal fluid. In preferred embodiments the tissue sample is obtained from saliva or blood.

Without wishing to be bound by theory, evidence suggests that Parkinson's disease pathology can originate in or in proximity to the enteric nervous system, before spreading to the central nervous system through the vagus nerve (Sampson et al., Cell. 2016, Liu et al., Neurology. 2017). Specifically, mutations in mitochondrial DNA may arise in or in proximity to the enteric nervous system, proliferate due to the presence of electron transport chain inhibitors originating in the microbiome of the intestinal lumen and later spread to the central nervous system through the vagus nerve due to intracellular and intercellular transfer of mitochondria and mitochondrial DNA. Proliferation of MELAS mutant mitochondrial DNA has been demonstrated in response to the electron transport chain inhibitor rotenone (Ives et al., Euromit Poster Presentation. 2014) whilst mitochondrial DNA transfer has been demonstrated between mesenchymal stem cells (MSCs) and epithelial cells in vitro (Ahmad et al., EMBO J, 2014); between cultured cell lines in vitro (Jayaprakash et al., Nucleic Acids Res, 2015); and most recently between astrocytes and neurons after stroke in vivo (Hayakawa et al., Nature, 2016), suggesting that transfer of mitochondrial DNA between different cell types could be widespread. Furthermore, the intestine possesses the largest proportion of the body's immune system by mass (Salminen et al., BrJ Nutr. 1998). Therefore, white blood cells proximal to the intestine may receive mitochondria and mutant mitochondrial DNA by intercellular transport, and then act as a vector for delivery of mutant mitochondrial DNA to the rest of the body since they are able to reach distal parts of the body via the bloodstream and infiltrate most tissues during the inflammatory response.

Therefore, the present invention provides a non-invasive sampling technique to assist in diagnosis of Parkinson's. A saliva sample, for example, contains plentiful white blood cells, and is also inexpensive to obtain, store, transport and process. In one embodiment, the sample may be collected using an Oragene OG-500 kit (DNA Genotek) according to the manufacturer's instructions, and DNA may be stabilised with proprietary stabilisation solution to give 5ml of stabilised DNA mixture. Stabilised DNA is stored at room temperature in a dark dry place until shipping, and shipped at room temperature for further processing. The non-invasive sampling technique allows for the present diagnostic method to be repeated without negatively impacting the patient. Therefore, the method can be repeated and used to monitor the progress of disease over time, to determine the rate of disease progression, or to determine the likely rate of neurological decline due to Parkinson's.

The methods of the present invention may involve isolating and sequencing mtDNA from a tissue sample. The methods of the present invention may also be conducted using a mtDNA sample that has already been isolated, or that has already been sequenced. Isolation and sequencing of mtDNA may be performed using any of the techniques known to those skilled in the art. Conventional methods for isolating mitochondrial DNA involve first isolating the mitochondria by differential centrifugation before then purifying mitochondrial DNA using standard protocols. To isolate a partial mitochondrial DNA sequence, a region of interest can be PCR amplified from total DNA (Smigrodzki et al., 2004, Neurobiology of Aging, 25; 1273; Parker and Parks, 2005, Biochem. Biophys. Res. Comm. 326;667). The PCR amplicon is then sequenced by shotgun sequencing. To isolate a whole mitochondrial DNA sequence, mitochondria can be isolated from other cell organelles by differential centrifugation, with mitochondrial DNA subsequently purified from isolated mitochondria (Lang.B.F. and Burger.G., 2007, Nat. Protoc, 2, 652-660. 30). The disadvantage of this method is that yields are low and contaminated with fragmented nuclear DNA (Ameur et al., 2011 , PLoS Genet, 7, e1002028). Mseek is an alternative method to isolate a whole mitochondrial DNA sequence, which takes total DNA and subjects it to enzymatic digestion of linear DNA, destroying linear nuclear genomic DNA, but sparing and enriching circular mitochondrial DNA. The Mseek method generates higher yields of mitochondrial DNA and minimises contamination with fragmented nuclear DNA compared to the differential centrifugation method. In one embodiment, mtDNA is isolated using the Mseek method, as described in US20150275200.

In preferred embodiments, mtDNA is isolated using an adapted Mseek method.

The adapted Mseek method for use in the present invention involves a further step of removing contamination from circular bacterial DNA in the sample, which in saliva samples can reach especially high levels. Non-methylated circular mitochondrial DNA is isolated from methylated circular bacterial DNA by digestion with Dpnl restriction endonuclease, an enzyme that specifically digests methylated DNA. The adapted Mseek method allows for isolation of the mtDNA with reduced contamination from circular bacterial DNA. Thus, the present invention provides a purer sample of mtDNA than conventional methods of isolation.

Limited samples of mitochondrial DNA may be sequenced by first PCR amplifying a region of interest before high throughput sequencing of the PCR-amplicons. Larger samples of mitochondrial DNA may be sequenced directly by high-throughput

sequencing. A region or a single gene of the mitochondrial genome may be sequenced. In preferred embodiments, each of the 16569-nucleotide positions of mitochondrial DNA may be sequenced. The mtDNA copies sequenced may be derived from the same cell or may be from different cells. In one embodiment, each of the 16569-nucleotide positions of mitochondrial DNA may be sequenced at least 2000 times, on average, to obtain accurate heteroplasmy measurements. In another embodiment, each of the 16569- nucleotide positions of mitochondrial DNA may be sequenced at least 1000 mtDNA times, on average, to obtain accurate heteroplasmy measurements.

The present methods involve identifying the presence and frequency of heteroplasmic mtDNA mutations in a mtDNA sample. The mtDNA sequences in the sample may be analysed at specific nucleotide positions or specific genetic regions. In preferred embodiments, the mtDNA sequences in the sample may be analysed at all nucleotide positions. The presence of a heteroplasmic mutation may be determined by comparison of the sample with a reference sample. The reference sample may be the Revised Cambridge Reference Sequence of the Human Mitochondrial DNA (rCRS

Genbank number NC_012920). The reference sample may be a mtDNA sample from a healthy control or healthy relative.

The percentage of mtDNA copies in the sample that contain a mutation at a specific nucleotide location determines the heteroplasmic frequency at that nucleotide position. In certain embodiments, heteroplasmy is characterised by the presence of between 7% and 90% of mutant mtDNA copies in the sample. Frequencies below 7% are deemed to be clinically insignificant, while frequencies above 90% may be considered benign mutations. In certain embodiments, the methods involve identifying the presence and frequency of at least one heteroplasmic mtDNA mutation that occurs in the subject sample with a heteroplasmic frequency between 7% and 90%, or with a heteroplasmic frequency of between 10% and 90%. In certain embodiments, the methods involve identifying the presence and frequency of at least one heteroplasmic mtDNA mutation that occurs in the subject sample with a heteroplasmic frequency between 7% and 90%, or with a heteroplasmic frequency of between 10% and 90%, wherein heteroplasmic frequency refers to the percentage of mtDNA copies in the sample having a mutation at a specific nucleotide position. In certain embodiments, the methods involve identifying the presence and frequency of at least one heteroplasmic mtDNA mutation that occur in the sample with a heteroplasmic frequency of between 7% and 74%. In another embodiment, the methods involve identifying the presence and frequency of at least one heteroplasmic mtDNA mutation that occurs in the sample with a heteroplasmic frequency between 7% and 52%. In another embodiment, the methods involve identifying the presence and frequency of at least one heteroplasmic mtDNA mutation in the ND5 gene that occurs in the sample with a heteroplasmic frequency between 7% and 74%. In another

embodiment, the methods involve identifying the presence and frequency of at least one heteroplasmic mtDNA mutation in the non-coding control region that occurs in the sample with a heteroplasmic frequency between 7% and 52%.

A number of factors may be used to determine the extent of progression of disease or disorder in a subject that has or is likely to develop Parkinson's. It has been recognised that symptoms and signs of Parkinson's disease can be present up to decades in advance of classical clinical signs of Parkinson's disease characterised by fully-evolved motor deficits. This earlier stage has been coined 'prodromal' Parkinson's. Symptoms and signs of prodromal Parkinson's have been formalised by the Movement Disorder Society to create an established set of diagnostic criteria (Berg et al., Mov Disord, 2015). These diagnostic criteria can be combined with other risk factors that influence the probability of developing Parkinson's. For example, age is the largest risk factor for development of the disease. Typically, symptoms of Parkinson's do not manifest until late-middle-age or old-age, for example, at 65 years of age or older. The present inventors hypothesise that heteroplasmic mutations may be present in a subject as young as 45 years of age, prior to the onset of clinically recognised motor symptoms of the disease. Therefore, heteroplasmic mtDNA mutations may be detectable in subjects presenting with prodromal Parkinson's, or in subjects who are asymptomatic. In certain embodiments, prodromal non-motor symptoms of Parkinson's and other risk factors for Parkinson's may be used in combination with the present diagnostic method to determine the extent of Parkinson's progression in a subject. In certain embodiments, the age of the subject may be used in combination with the present diagnostic method to determine the extent of Parkinson's progression in a subject. In certain embodiments, the heteroplasmic frequency may be used to determine the extent of Parkinson's progression. It is predicted that the frequency of a heteroplasmic mutation will increase to an upper limit of about 90% as the disease progresses. For example, a heteroplasmic mutation frequency of about 20% for a mutation in the non-coding control region may indicate a 50%

progression of the disease, whereas a heteroplasmic mutation frequency of about 40% for a mutation in the ND5 gene may indicate a 50% progression of the disease. The inventors hypothesise that a lower frequency of mutations in the control region is sufficient to kill a dopaminergic neuron compared to the frequency of mutations required in the ND5 gene to kill a dopaminergic neuron. The inventors further hypothesise that a higher frequency of mutations is required to cause cell death in white blood cells than in dopaminergic neurons due to a lower ATP demand in the former. This means that mutations in the ND5 gene or the control region can reach higher levels in white blood cells and therefore may be detected earlier and more easily. In certain embodiments, the total number of heteroplasmic mutations identified at different nucleotide positions in the mtDNA sample may be used to determine the extent of Parkinson's progression.

In certain embodiments, the heteroplasmic frequency can be used to determine the rate of Parkinson's progression. The heteroplasmic frequency may be determined in a second sample obtained from the same subject at a later time point, wherein the increase in heteroplasmic frequency between the two time points indicates the rate of disease progression. The samples may be obtained from the same tissue source or from different tissue sources, preferably from the same tissue source. In certain embodiments the second sample may be obtained at least 6 months after the first sample. In certain embodiments the second sample may be obtained at least 2 months, at least 4 months, at least 6 months, at least 8 months, at least 10, months, at least 12 months, at least 18 months, or at least 24 months after the first sample. The heteroplasmic frequency may be determined in further samples from the same subject obtained at later time points, such as a third sample or a fourth sample.

The methods of the present invention are in part dependent upon the age of the subject. The mitochondrial theory of ageing proposes that progressive accumulation of damage to mitochondria and somatic mutations in mtDNA during a lifetime leads to an inevitable decline in mitochondrial function and age-related disorders. Thus, symptoms associated with Parkinson's typically arise later in a person's life, around late-middie-age and old-age. Symptoms may arise as a result of age or environmental factors, or a combination of both. The motor symptoms of Parkinson's disease (rigidity, slowness of movement and difficulty with walking) only manifest once a severe dopamine deficiency develops in the brain (Schapira et a/., 2009, Eur J Neurol. Sep;16(9):982-9). For late onset idiopathic Parkinson's disease, this deficiency follows ceil death of between 60% and 80% of the dopamine-producing neurons in the substantia nigra (Cheng et a/., 20 0, Ann Neurol. Jun; 87(8): 715-725). If the disease could be detected earlier, before presentation of motor symptoms, therapies could be developed to target cell death of the dopamine producing neurons. This could delay or entirely prevent progression to motor symptoms of the disease. Possible therapies include neuroprotective drugs, drugs that reduce the proportion of mitochondria with mutant mtDNA, diets, behavioural-regimes or any combination of the described. Early detection of heteropiasmy could further allow for the identification of subjects eligible for treatment with drugs which reduce the proportion of mitochondria with mutant mtDNA. The availability of an accurate diagnostic test for Parkinson's significantly improves the ability to determine the therapeutic efficacy of a therapy for the treatment of Parkinson's in a human subject recipient of such therapy.

Importantly, the present inventors have demonstrated that changes in

heteroplasmic mtDNA frequencies arise at an earlier stage prior to the clinical symptoms of the disease. The present invention thus provides an early diagnostic applicable to predicting and detecting Parkinson's in a subject prior to the onset of clinical symptoms. The present invention allows for an early diagnostic of a subject who is at risk of developing Parkinson's, especially detection of individuals who are asymptomatic, or who are in pre-clinical stages of the disease. The invention provides an early screening method to detect and/or monitor the progression of Parkinson's. Thus, the method can be used to identify a subject at risk of developing the disease and enable treatment to prevent the clinical and biological symptoms associated with Parkinson's before they appear. The result of the present diagnostic method may be assessed in combination with other known factors generally associated with Parkinson's in order to improve the accuracy of diagnosis. For example, the results may be assessed in combination with one or more of the following: the age of the subject; the presence of environmental risk factors for Parkinson's; the presence of prodromal non-motor symptoms of Parkinson's.

In young subjects, the frequency of heteroplasmic mtDNA mutations may be too low to detect. The ability to detect heteroplasmy may be dependent in part on the tissue sample obtained. As explained above, mutations may be detected earlier in white blood cells, for example in a saliva sample. In certain embodiments, the subject is at least 20, at least 25, at least 35, at least 40, at least 45, at least 50, at least 55, at least 60, at least 65 years of age. The frequency of a specific heteroplasmic mtDNA mutation may increase in a subject during their lifetime. Similarly, the total number of heteroplasmic mtDNA mutations at different nucleotide positions may increase in a subject during their lifetime. In some instances, the frequency of a specific heteroplasmic mtDNA mutation may decrease in a subject during their lifetime. The change in the frequency of heteroplasmic mtDNA mutations in a subject during their lifetime may indicate the rate of progression of the disease.

The one or more heteroplasmic mtDNA mutations may be a point mutation, a deletion, an insertion, a rearrangement, and/or any other variant. The point mutation may be one of three base substitutions. The mutation may result in an amino acid change, i.e. the mutation may be non-synonymous. As reported elsewhere herein, the present inventors have identified specific heteroplasmic mtDNA mutations associated with Parkinson's. In certain embodiments, the one or more specific heteroplasmic mtDNA mutations is located between nucleotide positions 12337 and 14673 (SEQ ID NOs. 1 and 6), and/or between nucleotide positions 1 and 576 and/or between 16024 and 16569 (SEQ ID NO. 3; SEQ ID NO. 4). Preferably, the one or more specific heteroplasmic mtDNA mutations is located between nucleotide positions 12337 and 14148 (SEQ ID NO. 1), and/or between nucleotide positions 1 and 576 and/or between 16024 and 16569 (SEQ ID NO.3; SEQ ID NO.4). In certain embodiments, the one or more specific heteroplasmic mtDNA mutations is located in the ND5 gene between nucleotide positions 12337 and 14148 (SEQ ID N0.1), and/or is located in the non-coding control region between nucleotide positions 1 and 576 and/or between 16024 and 16569 (SEQ ID NO.3; SEQ ID NO.4). In certain preferred embodiments, the one or more specific heteroplasmic mtDNA mutations located in the ND5 region is located between nucleotide positions 13153 and 14059 (SEQ ID NO.2). In certain preferred embodiments, the one or more specific heteroplasmic mtDNA mutations located in the non-coding control region is located between nucleotide positions 302 and 525 (SEQ ID NO.5).

The present inventors have identified heteroplasmic mtDNA mutations associated with Parkinson's; four point mutations located in the ND5 gene, one point mutation located in the ND6 gene, and six variants located in the control region. These are listed in Table 1 , along with their nucleotide positions, base changes, percentage heteroplasmy and percentage occurrence in the Mitomap database. Codon positions and amino acid changes are also shown, where appropriate.

Table 1. Heteroplasmic mtDNA mutations associated with Parkinson's disease

The diagnostic method of the present invention can be used to identify a subject having one or more of the heteroplasmic mtDNA mutations in Table 1 as having or as being at risk of developing Parkinson's. The subject may also have additional

heteroplasmic mtDNA mutations not listed in Table 1. The subject may have mutations at the same nucleotide position as those listed in Table 1 but with alternative base substitutions, deletions, insertions or rearrangements. A subject having only synonymous mutations that do not result in an amino acid change is unlikely to have or be at risk of developing Parkinson's. Therefore, in preferred embodiments, a heteroplasmic mtDNA mutation identified in a coding region is a non-synonymous mutation. The methods may be used to identify a subject having only one 'pathogenic' heteroplasmic mtDNA mutation, or the methods may be used to identify a subject having multiple 'pathogenic' heteroplasmic mtDNA mutations. The more pathogenic heteroplasmic mtDNA mutations identified in a subject, then the greater the probability that the subject has or is at risk of developing Parkinson's.

The subject may have a heteroplasmic frequency of between 7% and 90% of one or more of the mutations listed in Table 1. The subject may have a heteroplasmic frequency of between 10% and 90% of one or more of the mutations listed in Table 1. In certain embodiments, the subject may have a heteroplasmic frequency of between 7% and 74% of one or more mtDNA mutations in the ND5 gene. In certain embodiments, the subject may have a heteroplasmic frequency of between 7% and 52% of one or more mtDNA mutations in the non-coding control region.

Each of the heteroplasmic mtDNA mutations in Table 1 occur in less than 1 % of the mitochondrial DNA sequences in the Mitomap database (http://www.mitomap.org, 2017). The Mitomap database is a reference database of more than 30,000 mtDNA sequences from healthy subjects. As of 8 May 2017, the Mitomap database consists of 32059 full-length individual sequences, and 66676 control region only sequences. Any mutation that occurs in more than 2.5% of the sequences in the Mitomap database, or other similar reference mtDNA database, may be considered a 'common' mutation and thus unlikely to be associated with Parkinson's. Therefore, the present method allows for the identification of a subject having or being at risk of developing Parkinson's wherein an identified heteroplasmic mtDNA mutation also occurs with an occurrence frequency of less than 2.5% of the sequences in a human mtDNA database. In another embodiment, the present method allows for the identification of a subject having or being at risk of developing Parkinson's wherein an identified heteroplasmic mtDNA mutation also occurs with an occurrence frequency of less than 2% of the sequences in a human mtDNA database. In another embodiment, the present method allows for the identification of a subject having or being at risk of developing Parkinson's wherein an identified heteroplasmic mtDNA mutation also occurs with an occurrence frequency of less than 1 % of the sequences in a human mtDNA database.

The database may be the Mitomap database, and/or the database may be the dbSNP database (Sherry ST et al., 2001 , Nucleic Acids Res. Jan 1 ;29(1):308-11). In certain embodiments a heteroplasmic mtDNA mutation in the ND5 gene occurs with an occurrence frequency of less than 1 % of the full-length sequences in the Mitomap database, and/or a heteroplasmic mtDNA mutation in the non-coding control region occurs in less than 0.15% of the sequences in the Mitomap database. Of the 32059 full- length sequences, thirty-two sequences are from individuals with Parkinson's. Therefore, the Mitomap database must be pre-processed in order to remove these thirty-two sequences, before the percentage of occurrence frequency can be calculated.

A subject can be diagnosed as having or being at risk of developing Parkinson's if a number of distinct criteria are met. In certain embodiments a subject can be diagnosed as having or being at risk of developing Parkinson's if three distinct criteria are met. If all three criteria are met, the probability of correctly identifying a subject as having or being at risk of developing Parkinson's is significantly increased.

Therefore, in a preferred embodiment the subject may be diagnosed as having or as being at risk of developing Parkinson's if the subject has at least one heteroplasmic mtDNA mutation:

a) that occurs in the subject with a heteroplasmic frequency of between 7% and 90%;

b) that occurs with an occurrence frequency of less than 2.5% of the

mitochondrial sequences in a human mtDNA database; and

c) is a non-synonymous mutation located between nucleotide positions

12337 and 14673 (SEQ ID NOs. 1 and 6), and/or is located between nucleotide positions 1 and 576 and/or is located between 16024 and 16569 (SEQ ID NOs.3 and 4).

In another embodiment the subject may be diagnosed as having or as being at risk of developing Parkinson's if the subject has at least one heteroplasmic mtDNA mutation: a) that occurs in the subject with a heteroplasmic frequency of between 7% and 90%;

b) that occurs with an occurrence frequency of less than 2.5% of the

mitochondrial sequences in a human mtDNA database; and

c) is a non-synonymous mutation located in the ND5 gene (SEQ ID N0.1) or is located in the non-coding control region (SEQ ID NOs.3 and 4). In another embodiment the subject may be diagnosed as having or as being at risk of developing Parkinson's if the subject has at least one heteroplasmic mtDNA mutation: a) that occurs in the subject with a heteroplasmic frequency of between 7% and 90%;

b) that occurs with an occurrence frequency of less than 1 % of the

sequences in a human mtDNA database, or it occurs with an

occurrence frequency of less than 0.15% of the sequences in a human mtDNA database; and

c) is a non-synonymous mutation located in the ND5 gene (SEQ ID N0.1) or is located in the non-coding control region (SEQ ID NOs.3 and 4). Mutations in the ND5 gene and mutations in the control region are proposed to result in the clinical expression of Parkinson's by slightly different mechanisms. Mutations in ND5 cause a change in protein structure, which reduces the proton pumping efficiency of the multi-subunit mitochondrial Complex-1 (ND5 is a subunit of Complex-1). Complex-1 pumps protons to generate an electrochemical proton-motive force, which is used by multi-subunit mitochondrial Complex-5 to generate ATP. Therefore, a reduction in the proton motive force causes a reduction in ATP production. A reduction in ATP production causes cell death in dopaminergic neurons of the Substantia nigra, which typically require high levels of ATP for sustained function and survival. Loss of these neurons creates a deficit in the neurotransmitter dopamine, which is crucial for correct movement and coordination. Lack of dopamine therefore causes the clinical expression of Parkinson's disease. In addition, classical Parkinson's disease is characterised by an increase in the formation of Lewy bodies caused by the aggregation of the cytosolic protein Alpha- synuclein. It was recently demonstrated that mitochondria are utilized by the cell to dispose of cytosolic proteins prone to aggregation, and that this process requires an inner mitochondrial membrane potential (Ruan et al., Nature, 2017). Disruption of this membrane potential in dopaminergic neurons due to the presence of mutant

mitochondrial DNA could therefore cause the build-up of the aggregation prone protein Alpha-synuclein, encouraging Lewy body formation.

In contrast, mutations in the nucleotide sequence of the non-coding control region are proposed to negatively affect binding of proteins required for the transcription of mRNA from mitochondrial DNA by mitochondrial RNA polymerase. This mRNA is used both for translation of mitochondrial DNA encoded proteins, including those proteins that form part of Complex-1 or 5, but also serves as a primer for mitochondrial DNA replication. An insufficient quantity of such proteins or mtDNA copies reduces ATP production overall, causing clinical expression of Parkinson's by the mechanism outlined above. Clinical utility of the methods

The method of the invention is used to identify a subject as either having Parkinson's, or as being at risk of developing Parkinson's, or to assess the extent or rate of Parkinson's progression. However, the basic methodology of:

(i) sequencing the mtDNA isolated from a sample obtained from a subject;

(ii) identifying the presence and frequency of heteroplasmic mtDNA mutations that occur in the subject sample, by comparison to a reference sample;

(iii) wherein at least one heteroplasmic mtDNA mutation identified in step (ii): a) occurs in the subject sample with a heteroplasmic frequency of between 7% and

90%; and

b) occurs with an occurrence frequency of less than 2.5% of the mitochondrial

sequences in a human mtDNA database

can be used in a variety of different clinical applications.

As described elsewhere herein, in certain embodiments the subject may be asymptomatic. Therefore, the methodology of the invention may be used to assess risk of developing Parkinson's in a subject prior to the emergence of classical Parkinson's symptoms. In other words, the assay methodology can identify subjects who are predisposed to develop Parkinson's due to the presence of abnormal heteroplasmic frequencies of mutations in their mtDNA, irrespective of the presenting symptoms, or lack of, in that subject. In certain embodiments, the methodology of the invention may be used to screen asymptomatic subjects to assess risk/predisposition for developing Parkinson's.

In alternative embodiments, the same basic methodology may be used to screen subjects who are "symptomatic" to varying degrees. For example, the subject may already exhibit one or more symptoms consistent with Parkinson's. The subject may have one or more of the non-motor symptoms associated with prodromal Parkinson's. The subject may have been diagnosed already as having Parkinson's after clinical

assessment using the Unified Parkinson's Disease Rating Scale (UPDRS). The additional diagnostic criterion provided by the present methodology of the invention may therefore provide a useful tool to assist with clinical diagnosis in the live patient. Since a

"confirmed" diagnosis of Parkinson's disease can only be made post-mortem, a subject may present with symptoms generally associated (though not necessarily exclusively) with Parkinson's disease. Therefore, the present methodology may be used to assist diagnosis in a live patient and confirm whether the symptoms are attributable to

Parkinson's disease, and not some other disease. In other embodiments, the methodology of the present invention may be applied to a subject who has not been assessed by a clinician using the UPDRS. The present methods thus may provide an alternative approach to Parkinson's diagnosis, based on detection of abnormal heteroplasmic frequencies of mtDNA mutations, rather than evaluation of outward symptoms of the disease. The methods could be viewed as providing an alternative basis for clinical diagnosis of Parkinson's in the live patient which is independent of the UPDRS assessment.

The methods as described herein may be combined with alternative methods for diagnosing Parkinson's, for example, to further improve accuracy in prediction and diagnosis. Therefore, identifying the presence and frequency of heteroplasmic mtDNA mutations may be used as a marker to identify a subject as having or as being at risk of developing Parkinson's, optionally in combination with other existing techniques, described herein or known in the art, for the diagnosis of Parkinson's. In certain embodiments, the invention may be combined with an assessment using the Unified Parkinson's Disease Rating Scale. The diagnostic method of the present invention can be used in conjunction with other diagnostic methods, for example, to provide further information such as the location of the heteroplasmic mtDNA mutation, and therefore predict the potential effect of the mutation on protein function. The combined method may also provide further information relating to the extent and/or rate of disease progression, as reported elsewhere herein.

The results of the present method may be combined with other information concerning the subject in order to improve the accuracy of correctly identifying a subject as having or being at risk of developing Parkinson's. Therefore, the present method may be used to assist in the clinical diagnosis of Parkinson's in a subject. The confidence in the result of the present diagnostic method will increase if the subject also presents with one or more factors known to be associated with Parkinson's. For example, the age of the subject is an important factor for Parkinson's. Therefore, in certain embodiments of the invention, the age of the subject is used as an additional criterion to identify a subject as having or being at risk of developing Parkinson's. In certain embodiments of the invention, the presence of one or more prodromal non-motor symptoms of Parkinson's in the subject is used as an additional criterion to identify a subject as having or being at risk of developing Parkinson's.

Accordingly, the scope of the invention also extends to the following methods. It will be clear to those of skill in the art that the particular embodiments described above with respect to the method of identifying a subject as having or as being at risk of developing Parkinson's apply equally to the following: A method of obtaining a diagnostic criterion associated with Parkinson's in a subject, which method comprises:

(i) sequencing the mtDNA isolated from a sample obtained from a subject; (ii) identifying the presence and frequency of heteroplasmic mtDNA mutations that occur in the subject sample, by comparison to a reference sample; (iii) wherein at least one heteroplasmic mtDNA mutation identified in step (ii): a) occurs in the subject sample with a heteroplasmic frequency of between 7% and 90%; and

b) occurs with an occurrence frequency of less than 2.5% of the mitochondrial

sequences in a human mtDNA database,

and wherein the presence of one or more such heteroplasmic mtDNA mutations provides a diagnostic criterion associated with Parkinson's. A method of assessing the risk of developing Parkinson's in a subject, which method comprises:

(i) sequencing the mtDNA isolated from a sample obtained from a subject;

(ii) identifying the presence and frequency of heteroplasmic mtDNA mutations that occur in the subject sample, by comparison to a reference sample; (iii) wherein at least one heteroplasmic mtDNA mutation identified in step (ii): a) occurs in the subject sample with a heteroplasmic frequency of between 7% and 90%; and

b) occurs with an occurrence frequency of less than 2.5% of the mitochondrial

sequences in a human mtDNA database,

and wherein the presence of one or more such heteroplasmic mtDNA mutations indicates risk of developing Parkinson's.

A method to assist with clinical diagnosis of Parkinson's in a live human subject, which method comprises:

(i) sequencing the mtDNA isolated from a sample obtained from a subject;

(ii) identifying the presence and frequency of heteroplasmic mtDNA mutations that occur in the subject sample, by comparison to a reference sample;

(iii) wherein at least one heteroplasmic mtDNA mutation identified in step (ii): a) occurs in the subject sample with a heteroplasmic frequency of between 7% and 90%; and

b) occurs with an occurrence frequency of less than 2.5% of the mitochondrial

sequences in a human mtDNA database, and wherein the presence of one or more such heteroplasmic mtDNA mutations indicates that said subject has Parkinson's.

A method to assist with diagnosis of pre-clinical Parkinson's in a live human subject, which method comprises:

(i) sequencing the mtDNA isolated from a sample obtained from a subject;

(ii) identifying the presence and frequency of heteroplasmic mtDNA mutations that occur in the subject sample, by comparison to a reference sample;

(iii) wherein at least one heteroplasmic mtDNA mutation identified in step (ii): a) occurs in the subject sample with a heteroplasmic frequency of between 7% and

90%; and

b) occurs with an occurrence frequency of less than 2.5% of the mitochondrial

sequences in a human mtDNA database,

and wherein the presence of one or more such heteroplasmic mtDNA mutations indicates that said subject has pre-clinical Parkinson's.

A method of obtaining a prognostic criterion indicative of the likely rate of neurological decline due to Parkinson's in a human subject, which method comprises:

(i) sequencing the mtDNA isolated from a sample obtained from a subject; (ii) identifying the presence and frequency of heteroplasmic mtDNA mutations that occur in the subject sample, by comparison to a reference sample;

(iii) wherein at least one heteroplasmic mtDNA mutation identified in step (ii): a) occurs in the subject sample with a heteroplasmic frequency of between 7% and 90%; and

b) occurs with an occurrence frequency of less than 2.5% of the mitochondrial

sequences in a human mtDNA database,

and wherein the presence of one or more of such heteroplasmic mtDNA mutations provides a diagnostic criterion indicative of the likely rate of neurological decline due to Parkinson's.

In a specific embodiment, the present invention provides a method of diagnosing and treating a subject as having or being at risk of developing Parkinson's, comprising the steps of:

(i) isolating and sequencing the mtDNA from a sample obtained from a

subject;

(ii) identifying the presence and frequency of heteroplasmic mtDNA mutations in the subject sample, by comparison to a reference sample; (iii) diagnosing the subject as having or being at risk of developing Parkinson's when at least one heteroplasmic mtDNA mutation identified in step (ii) occurs in the subject sample with a heteroplasmic frequency of between 7% and 90%; and occurs with an occurrence frequency of less than 2.5% of the mitochondrial sequences in a human mtDNA database; and

(iv) treating a subject with an effective amount of a therapy designed to

prevent or slow progression of Parkinson's.

The therapy may be any treatment designed to prevent the development of Parkinson's; slow the progression of the disease; or reverse the symptoms of the disease. The skilled person is aware of therapies for use in the present invention.

Possible therapies could include, but are not limited to, neuroprotective drugs, drugs that reduce the proportion of mitochondria with mutant mtDNA, diets, behavioural-regimes or any combination of the described. Early detection of heteroplasmy could allow for the identification of subject's eligible for treatment with drugs which reduce the proportion of mitochondria with mutant mtDNA. The availability of an accurate diagnostic test for Parkinson's significantly improves the ability to determine the therapeutic efficacy of a therapy for the treatment of Parkinson's in a human subject recipient of such therapy.

In another specific embodiment the invention provides a method of detecting a pathological heteroplasmic mtDNA mutation in a subject, said method comprising the steps of:

(i) isolating and sequencing the mtDNA from a sample obtained from a

subject;

(ii) detecting the presence and frequency of heteroplasmic mtDNA mutations in the subject sample, by comparison to a reference sample; (iii) wherein a heteroplasmic mtDNA mutation is pathological if:

a) it occurs in the subject sample with a heteroplasmic frequency of between 7% and 90%; and

b) it occurs with an occurrence frequency of less than 2.5% of the mitochondrial

sequences in a human mtDNA database.

The pathological heteroplasmic mtDNA mutation may be associated with Parkinson's.

C. Method of identifying new mutations associated with Parkinson's

In a second aspect, the present invention provides a method of identifying a heteroplasmic mtDNA mutation associated with Parkinson's. The methods enable the identification of novel mutations associated with Parkinson's. The methods of the invention are based in part on the understanding that mtDNA is maternally inherited, and thus an individual's mtDNA is a genetic copy passed down from the mother. Therefore, the mtDNA from two individuals that have related mtDNA can be compared to identify differences that may be associated with disease. A related mtDNA sample provides the most reliable control sample since it reduces the issue of inter-individual variation in mtDNA sequence. Two related mtDNA samples will have a fewer number of variable mutations, and therefore a mutation at a nucleotide position that differs between the related mtDNA sequences has a greater likelihood of being disease-related. The present inventors have found that mtDNA from a Parkinson's patient can be compared with mtDNA from a relative who has the same inherited mtDNA. The difference in the presence and frequency of one or more heteroplasmic mtDNA mutations in a patient with Parkinson's compared to a healthy relative with related mtDNA can be used to identify mutations associated with Parkinson's with a high degree of resolution.

Identifying heteroplasmic mtDNA mutations associated with Parkinson's according to the second aspect of the methods is carried out on a tissue sample obtained from a patient with Parkinson's. In alternative embodiments the methods are carried out on a tissue sample obtained from both the patient with Parkinson's and a tissue sample obtained from a healthy relative of said patient who shares the same inherited mtDNA. The tissue sample may be from a non-neuronal source including, but not limited to, saliva, skin, hair, urine, blood, earwax, cheek swab, tongue scrape or cerebrospinal fluid. In preferred embodiments the tissue sample is obtained from saliva or blood. For comparison, the sample from the patient with Parkinson's and the sample from the relative with related DNA may be obtained from the same tissue source. For example, in certain embodiments, both samples are obtained from saliva. The invention therefore provides a non-invasive sampling technique to assist in the identification of heteroplasmic mtDNA mutations associated with Parkinson's.

The methods relating to this second aspect of the invention may involve isolating and sequencing mtDNA from a tissue sample. Isolating and sequencing methods may be performed in accordance with the methods described above relating to the first aspect of the invention. The methods of the present invention may also be conducted using a mtDNA sample that has already been isolated, or that has already been sequenced. In preferred embodiments, mtDNA is isolated using an adapted Mseek method. The adapted Mseek method for use in the present invention involves a further step of removing contamination from circular bacterial DNA in the sample. Non-methylated circular mitochondrial DNA is isolated from methylated circular bacterial DNA by digestion with Dpnl restriction endonuclease, an enzyme that specifically digests methylated DNA. The adapted Mseek method allows for isolation of the mtDNA with reduced contamination from circular bacterial DNA. Thus, the present invention provides a purer sample of mtDNA than conventional methods of isolation.

The present methods involve identifying the presence and frequency of heteroplasmic mtDNA mutations in a mtDNA sample from a patient with Parkinson's by comparison with a reference mtDNA sample. The mtDNA sequences in the paired samples may be analysed at specific nucleotide positions or locations. The mtDNA sequences in the paired samples may be analysed at all nucleotide positions. The presence of a heteroplasmic mutation may be determined by comparison of the test sample with a reference sample. The reference sample may be the Revised Cambridge Reference Sequence of the Human Mitochondrial DNA (rCRS Genbank number

NC_012920). The reference sample may be a mtDNA sample from a healthy control. In preferred embodiments, the presence and frequency of one or more heteroplasmic mtDNA mutations in a mtDNA sample from a patient with Parkinson's is compared to the presence and frequency of heteroplasmic mtDNA mutations in a mtDNA sample from a healthy relative of said patient who shares the same related mtDNA. As defined elsewhere herein, related mtDNA refers to mtDNA that is inherited from one individual to another biologically related individual. mtDNA is inherited from the mother only; therefore, a mother and her biological children share the same inherited or related mtDNA.

Furthermore, an individual and their biologically related siblings share the same inherited mtDNA as each other, since they each inherited the mtDNA from the same mother. In certain embodiments of the present invention, where the patient with Parkinson's is female, the heathy relative may be a biological child, mother, or sibling of said patient. In preferred embodiments where the patient is female, the healthy relative is a biological child. In certain embodiments, where the patient with Parkinson's is male, the healthy relative may be a biological sibling, or mother, preferably a biological sibling. In some embodiments, the sibling may be a maternal half sibling. The healthy relative is a relative who does not themselves have Parkinson's, or is not known to be at risk of developing Parkinson's disease. In preferred embodiments, the healthy relative also does not have another age-related disorder, such as Alzheimer's disease.

The method of identifying new mutations in accordance with this second aspect is in part dependent upon the age of the patient with Parkinson's. Symptoms associated with Parkinson's typically arise later in a person's life, around late-middle-age and old- age. Symptoms may arise as a result of age or environmental factors, or a combination of both. The motor symptoms of Parkinson's disease (rigidity, slowness of movement and difficulty with walking) only manifest once a severe dopamine deficiency develops in the brain (Schapira, AH et a/., 2009, Eur J Neurol. Sep; 16(9):982-9). The classical diagnosis of Parkinson's, for example using the UPDRS, depends upon the presence of clinical symptoms. Therefore, the patient with Parkinson's as referred to in the present invention typically has some or all of the classical symptoms associated with Parkinson's. For this reason, in preferred embodiments, the patient with Parkinson's disease is at least 65 years of age. However, the present inventors have demonstrated that changes in heteroplasmic mtDNA frequencies arise at an earlier stage prior to the clinical symptoms of the disease. Thus, the present invention allows for the detection of Parkinson's in a patient prior to the onset of clinical symptoms. In light of this, the invention foresees that the patient with Parkinson's disease may be at least 35, at least 40, at least 45, at least 50, at least 55, at least 80, at least 65 years of age. Similarly, the health relative may be at least 20, at least 25, at least 35, at least 40, at least 45, at least 50, at least 55, at least 60, at least 65 years of age. Preferably, the healthy relative is younger than the patient with Parkinson's.

As mentioned previously, the number and frequency of heteroplasmic mtDNA mutations in an individual increases over time as a consequence of both the aging process and a number of environmental factors. The present invention provides a method of identifying which of these heteroplasmic mtDNA mutations are specifically associated with Parkinson's. The human mitochondrial genome consists of 16569 nucleotide positions. A heteroplasmic mutation may occur at any one of these positions, and may be either a deletion, an insertion, a rearrangement, or one of three different substitutions. However not all of these positions and their potential mutations are of interest. It is possible to reduce the total number of mutations of interest to include only those mutations that satisfy specific parameters. For example, the present inventors have identified the following parameters as important for identifying a heteroplasmic mtDNA mutation that is likely to be associated with Parkinson's: the location of the heteroplasmic mutation; whether the mutation results in a change in the amino acid sequence i.e. a non- synonymous mutation; the frequency of heteroplasmy; the occurrence of the

heteroplasmic mutation in a human mtDNA database; and the difference in heteroplasmic frequency between the patient and their healthy relative. Each of these parameters are discussed below. i. Frequency of heteroplasmy

The present invention involves identifying mtDNA mutations that occur in the patient with Parkinson's at a given heteroplasmic frequency. The percentage of mtDNA copies in the sample that contain a mutation at a specific nucleotide location determines the heteroplasmic frequency at that nucleotide position. In certain embodiments, heteroplasmy is characterised by the presence of between 7% and 90% of mutant mtDNA copies in the sample. In certain embodiments, the methods involve identifying the presence and frequency of a heteroplasmic mtDNA mutation that occurs in the sample from the patient with Parkinson's with a heteroplasmic frequency of between 7% and 90%, or with a heteroplasmic frequency of between 10% and 90%, wherein

heteroplasmic frequency refers to the percentage of mtDNA copies in the sample having a mutation at a specific nucleotide position.. In certain embodiments, the methods involve identifying the presence and frequency of a heteroplasmic mtDNA mutation that occurs in the patient sample with a heteroplasmic frequency of between 7% and 74%. In another embodiment, the methods involve identifying the presence and frequency of a heteroplasmic mtDNA mutation that occurs in the patient sample with a heteroplasmic frequency of between 7% and 52%. In another embodiment, the methods involve identifying the presence and frequency of a heteroplasmic mtDNA mutation in the ND5 gene that occurs in the patient sample with a heteroplasmic frequency of between 7% and 74%. In another embodiment, the methods involve identifying the presence and frequency of a heteroplasmic mtDNA mutation in the non-coding control region that occurs in the patient sample with a heteroplasmic frequency of between 7% and 52%. ii. Difference in heteroplasmic frequency between paired samples In addition to identifying a mtDNA mutation that occurs with a specific

heteroplasmic frequency in a patient with Parkinson's, it is a useful tool of the present invention to compare the heteroplasmic frequency of the mutation in the patient with Parkinson's with the heteroplasmic frequency, if any, at the same nucleotide position in a healthy relative of said patient. In certain embodiments the difference in frequency observed in the patient with Parkinson's and the frequency observed in the healthy relative is between 5% and 83%. The frequency observed in the patient with Parkinson's may be higher than the frequency observed in the healthy relative. The frequency observed in the patient with Parkinson's may be lower than the frequency observed in the healthy relative. The difference in heteroplasmic frequency observed in the patient with Parkinson's and the heteroplasmic frequency observed in the healthy relative may be at least 5%. The difference in heteroplasmic frequency observed in the patient with

Parkinson's and the heteroplasmic frequency observed in the healthy relative may be at least 7%. The difference in heteroplasmic frequency observed in the patient with

Parkinson's and the heteroplasmic frequency observed in the healthy relative may be at least 10%. iii. Occurrence in reference database

The present inventors have found that a heteroplasmic mtDNA mutation identified in a patient with Parkinson's is more likely to be associated with Parkinson's where that same mutation occurs infrequently in a reference database of healthy human mtDNA sequences. In contrast, a mutation that occurs frequently in a reference database is likely to be relatively common, and thus not associated specifically with disease. In certain embodiments the heteroplasmic mtDNA mutations identified in the patient with

Parkinson's occurs with an occurrence frequency of less than 2.5% of the mitochondrial sequences in a human mtDNA database. In certain embodiments the heteroplasmic mtDNA mutations identified in the patient with Parkinson's occurs with an occurrence frequency of less than 2% of the mitochondrial sequences in a human mtDNA database. In certain embodiments the heteroplasmic mtDNA mutations identified in the patient with Parkinson's occurs with an occurrence frequency of less than 1 % of the mitochondrial sequences in a human mtDNA database.

The database may be the Mitomap database, and/or the database may be the dbSNP database (Sherry ST et al., 2001 , Nucleic Acids Res. Jan 1 ;29(1):308-11). In preferred embodiments, the human mtDNA database is the Mitomap database. The Mitomap database is a reference database of more than 30,000 mtDNA sequences from healthy subjects. Any mutation that occurs frequently in the Mitomap database, or other similar reference mtDNA database, may be considered a 'common' mutation and thus unlikely to be associated with Parkinson's. As of 8 May 2017, the Mitomap database consists of 32059 full-length individual sequences, and 66676 control region only sequences. Any mutation that occurs in more than 2.5% of the sequences in the Mitomap database, or other similar reference mtDNA database, may be considered a 'common' mutation and thus unlikely to be associated with Parkinson's. In certain embodiments a heteroplasmic mtDNA mutation in the ND5 gene occurs with an occurrence frequency of less than 1 % of the full-length sequences in the Mitomap database, and/or a

heteroplasmic mtDNA mutation in the non-coding control region occurs in less than 0.15% of the sequences in the Mitomap database. Of the 32059 full-length sequences, thirty-two sequences are from individuals with Parkinson's. Therefore, the Mitomap database must be pre-processed in order to remove these thirty-two sequences, prior to calculating the percentage occurrence of a mutation. iv. Location of mutation

As reported elsewhere herein, the present inventors have found specific heteroplasmic mtDNA mutations associated with Parkinson's. These heteroplasmic mutations appear in specific regions within the mitochondrial genome. In certain embodiments, the present method can be used to identify the presence and frequency of one or more heteroplasmic mtDNA mutations located between nucleotide positions 12337 and 14673 (SEQ ID NOs. 1 and 6), and/or is located between nucleotide positions 1 and 576 and/or is located between 16024 and 16569 (SEQ ID NO. 3 and 4).

The inventors have shown a heteroplasmic mtDNA mutation that occurs in one of two regions, the MT-ND5 gene and the non-coding control region has a greater probability of being associated with Parkinson's. Therefore, in certain embodiments, the present method can be used to identify the presence and frequency of one or more heteroplasmic mtDNA mutations located in the MT-ND5 gene and/or the non-coding control region. In preferred embodiments, the present method can be used to identify the presence and frequency of one or more heteroplasmic mtDNA mutations located between nucleotide positions 12337 and 14148 (SEQ ID NO. 1), and/or is located between nucleotide positions 1 and 576 and/or is located between 16024 and 16569 (SEQ ID NO. 3 and 4).

Mutations in the ND5 gene and mutations in the control region are proposed to result in the clinical expression of Parkinson's by slightly different mechanisms, as explained elsewhere herein. For example, in the protein coding sequence, different types of mutation can have a very different impact on protein sequence and structure. A substitution may not change the amino acid sequence of a protein (synonymous) or may alter a single amino acid in the amino acid sequence (non-synonymous). In contrast, a rearrangement, or an insertion or deletion with a nucleotide length non-divisible by three, can alter a much larger stretch of amino acid sequence in what is known as a 'frame- shift'. Therefore, the type of mutation can have a dramatic effect on the amount of disruption to the amino acid sequence, which consequently may affect protein function and mitochondrial function. In contrast to protein coding sequences, different types of mutation are not expected to have such a variable impact on the function of non-coding sequences (e.g. mutations in the control region). In certain embodiments the one or more heteroplasmic mtDNA mutations identified in the patient with Parkinson's may be a point mutation, a deletion, an insertion, a rearrangement, and/or any other variant. The point mutation may be one of three base substitutions. The mutation may result in an amino acid change, i.e. the mutation may be non-synonymous. A synonymous heteroplasmic mtDNA mutation that does not result in an amino acid change is unlikely to be associated with Parkinson's.

In certain embodiments, the heteroplasmic mtDNA mutations identified in the patient with Parkinson's may be located in the ND5 gene between nucleotide positions 12337 and 14148 (SEQ ID N0.1), and/or may be located in the non-coding control region between nucleotide positions 1 and 576 and/or between 16024 and 16569 (SEQ ID NO.3; SEQ ID NO.4). In certain embodiments, the heteroplasmic mtDNA mutations located in the ND5 region may be located between nucleotide positions 13153 and 14059 (SEQ ID NO.2). In certain embodiments, the heteroplasmic mtDNA mutations located in the non-coding control region may be located between nucleotide positions 302 and 525 (SEQ ID NO.5).

In certain embodiments a heteroplasmic mtDNA mutation identified in a patient with Parkinson's may be confirmed as being associated with Parkinson's if one or more distinct criteria are met. If all criteria are met, the probability of correctly identifying a heteroplasmic mtDNA mutation associated with Parkinson's is significantly increased.

Therefore, in a preferred embodiment a heteroplasmic mtDNA mutation identified in the patient with Parkinson's may be classified as being associated with Parkinson's with a high probability if the heteroplasmic mtDNA mutation identified:

a) occurs in the patient with a heteroplasmic frequency of between 7% and

90%;

b) occurs with an occurrence frequency of less than 2.5% of the

mitochondrial sequences in a human mtDNA database; c) is a non-synonymous mutation located between nucleotide positions

12337 and 14673 (SEQ ID NOs. 1 and 6), and/or is located between nucleotide positions 1 and 576 and/or is located between 16024 and 16569 (SEQ ID NO. 3 and 4); and

d) there is a difference in frequency observed in the patient with Parkinson's and the frequency observed in a reference sample.

In another embodiment a heteroplasmic mtDNA mutation identified in the patient with Parkinson's may be classified as being associated with Parkinson's with a high probability if the heteroplasmic mtDNA mutation identified:

a) occurs in the patient with a heteroplasmic frequency of between 7% and 90%; b) occurs with an occurrence frequency of less than 2.5% of the mitochondrial sequences in a human mtDNA database; c) is a non-synonymous mutation located in the ND5 gene (SEQ ID NO.1) or is located in the non-coding control region (SEQ ID NOs.3 and 4); and d) there is a difference in frequency observed in the patient with Parkinson's and the frequency observed in a reference sample.

In one embodiment a heteroplasmic mtDNA mutation identified in the patient with Parkinson's may be classified as being associated with Parkinson's with a high probability if the heteroplasmic mtDNA mutation identified:

a) occurs in the patient with a heteroplasmic frequency of between 7% and

90%;

b) occurs with an occurrence frequency of less than 1 % of the sequences in a human mtDNA database or occurs with an occurrence frequency of less than 0.15% of the sequences in a human mtDNA database ;

c) is a non-synonymous mutation located in the ND5 gene (SEQ ID NO.1) or is located in the non-coding control region (SEQ ID NOs.3 and 4); and d) there is a difference in frequency observed in the patient with Parkinson's and the frequency observed in a reference sample.

A patient with Parkinson's may have more than one heteroplasmic mtDNA mutation that satisfies each of the above criteria. Therefore, the method of identifying a

heteroplasmic mtDNA mutation associated with Parkinson's may be repeated on the same mtDNA sample in order to identify further heteroplasmic mutations of interest. v. Other parameters

While the above listed parameters, (i-iv) have been identified as being the most important criteria for assessing heteroplasmic mtDNA mutations associated with

Parkinson's, a number of other parameters can be additionally used to confirm the identification of a pathogenic mutation associated with Parkinson's with greater probability. For instance, the present inventors have identified a positive correlation between scores generated by the computational pathogenicity prediction tool 'Mitlmpact' and mutations identified in Parkinson's patients. Computational prediction methods can be used to reveal the impact that a specific mutation has on the overall nucleotide and/or protein structure. It is to be understood that a mutation that disrupts the nucleotide and/or protein structure may disrupt cellular functions and so has a greater probability of being associated with disease. Mitlmpact provides an exhaustive collection of pre-computed pathogenicity predictions of human mitochondrial non-synonymous variants (Castellana. S. et al., 2015, 36(2) :E2413-22). The present methods may include a further step of confirming a heteroplasmic mtDNA mutation associated with Parkinson's by determining whether the mutation disrupts the nucleotide and/or protein structure using computational prediction methods. Different computational models may predict whether a given mutation is likely to be damaging, or have a neutral impact. The results from a number of different computational models can be combined and the mutation assigned a score, termed a 'Mitlmpact score' (see Table 6).

A further parameter that can be used to confirm a potentially pathogenic mutation associated with Parkinson's is to observe whether the wild-type nucleotide sequence is conserved across a range of species. It is to be understood that a mutation occurring in a patient with Parkinson's at a nucleotide sequence location where the wild-type sequence is a highly conserved sequence has a greater probability of being associated with disease. The present methods may include a further step of confirming a heteroplasmic mtDNA mutation associated with Parkinson's wherein the corresponding wild-type nucleotide sequence is conserved across a defined set of species above a set threshold. In certain embodiments, the wild-type nucleotide sequence may be conserved in more than 50% of a defined set of species.

A further parameter that can be used to confirm a potentially pathogenic mutation associated with Parkinson's is to observe whether the heteroplasmic frequency of the identified mtDNA mutation can be altered in response to small molecule drugs that have the capacity to alter the heteroplasmic frequency of alternative mutations. This chemical method can be used to predict the functional effects of mutations. The present methods may include a further step of confirming a heteroplasmic mtDNA mutation associated with Parkinson's wherein the frequency of the heteroplasmic mtDNA mutation is reduced in response to small molecule drugs that have demonstrated capability to reduce the frequency of alternative pathogenic mutations. In one embodiment, the heteroplasmic frequency may be reduced by more than 10%.

It is to be understood that a potentially pathogenic heteroplasmic mtDNA mutation identified by the methods of the present invention can be confirmed as being associated with Parkinson's with a greater probability by performing one or more of the above techniques, or any other technique known in the art that can be used to confirm pathogenic status. D. Optimisation of methods

As with other diagnostic methods, the methods of the invention may not be 100% accurate in terms of identifying mutations associated with Parkinson's and/or diagnosing a subject as having or being at risk of developing Parkinson's. It is another aspect of the present invention that the methods can be optimised in order to diagnose a subject as having or being at risk of developing Parkinson's with improved confidence and accuracy. It is another aspect of the present invention that the methods can be optimised in order to identify new heteroplasmic mtDNA mutations associated with Parkinson's with improved confidence and accuracy. Each mutation identified in a patient with Parkinson's, presents with a known set of parameters. As more mutations are identified in more patient samples, it is possible to further define the thresholds of these parameters, i.e. to define threshold values that are common to mutations associated with Parkinson's. Thus, as more mutations are identified the confidence with which any new mutation can be classified as "associated with Parkinson's" increases depending on whether the new mutation falls within the same or a similar set of parameters. Consequently, the confidence with which a subject presenting with a new mutation can be diagnosed as having or being at risk of developing Parkinson's also increases.

The method of identifying mutations associated with Parkinson's, as described elsewhere herein, can be repeated on a plurality of samples so as to generate a data set of known heteroplasmic mtDNA mutations associated with Parkinson's. The data set may include data from both patients with Parkinson's, and data from heathy controls or healthy relatives. This data set may be generated using an initial set of parameters. For example, the heteroplasmic mtDNA mutations associated with Parkinson's identified in the initial data set may each occur with a heteroplasmic frequency of between 7% and 90%, and occur in less than 2.5% of the sequences in a human mtDNA database. The actual percentage values of each sample are recorded in the database. The parameter thresholds can then be optimised by applying an appropriate algorithm to the data set. The algorithm will recalculate the preferred thresholds that allow for the most accurate classification of all the samples in the data set with the lowest number of false positives and false negatives, i.e. with the fewest number of samples from Parkinson's patients being incorrectly classified as 'healthy', and the fewest number of samples from healthy subjects being incorrectly classified as associated with Parkinson's. Thus, as further mutations are identified, the size of the data set increases, and the parameter thresholds can be adjusted to improve the accuracy of the methods described herein.

The present invention therefore provides a process of optimising the method of identifying heteroplasmic mtDNA mutations associated with Parkinson's, wherein the thresholds can be optimised using an algorithm performed on a data set that has been generated by repeating the same method on a plurality of samples.

For example, all of the mutations listed in Table 1 were identified using an initial set of 24 samples; 7 samples from patients with Parkinson's, and 17 samples from healthy controls. Applying the following threshold parameters wherein a mutation associated with Parkinson's: (i) is located between between nucleotide positions 12337 and 14673, and/or is located between nucleotide positions 1 and 576 and/or is located between 16024 and 16569; (ii) has a heteroplasmic frequency of between 7% and 90%; (iii) and occurs in less than 2.5% of the Mitomap database, will result in accurate classification of 23 out of 24 samples as either a Parkinson's patient or a healthy control.

For example, the mutations located in the ND5 gene or the non-coding control region listed in Table 1 were identified using an initial set of 20 samples; 6 samples from patients with Parkinson's, and 14 samples from healthy controls. Applying the following threshold parameters wherein a mutation associated with Parkinson's: (i) is present in the ND5 gene or the non-coding control region; (ii) has a heteroplasmic frequency of between 7% and 90%; (iii) and occurs in less than 2.5% of the Mitomap database, will result in accurate classification of 19 out of 20 samples as either a Parkinson's patient or a healthy control. Alternatively, applying the narrower threshold parameters wherein a mutation associated with Parkinson's: (i) has a heteroplasmic frequency of between 7% and 90%; (ii) if it is present in the ND5 gene then it occurs in less than 1 % of the full-length sequences in the Mitomap database; (iii) and if it is present in the non-coding control region then it occurs in less than 0.15% of the sequences in the Mitomap database, will result in accurate classification of all 20 samples as either a Parkinson's patient or a healthy control. The threshold values can therefore be refined in order to identify mutations associated with Parkinson's with the greatest probability and with the fewest number of false positives and negatives.

However, as new mutations are identified in more samples, the threshold parameters can be refined further. For example, the present inventors hypothesise that there may be a maximum observed heteroplasmic frequency for mutations occurring in the non-coding control region. The inventors consider that it is highly likely that the level of heteroplasmy can be used to indicate the degree of progression of the disease.

Therefore the upper limit of 90% heteroplasmic frequency may be reduced as more mutations are identified in patients with late-stage Parkinson's disease that reveal a maximum observed heteroplasmic frequency that is lower than 90%. These thresholds are therefore not fixed, but may be altered as new mutations are identified in more samples. The thresholds may need to be adapted, as more samples are added to the existing database, in order to increase the probability of correctly identifying a new mutation as being associated with Parkinson's.

This optimisation step can be performed using a standard algorithm to calculate the optimum thresholds. In this way, the accuracy of the methods can be continually improved as more patient data is added to the existing database. The thresholds may be re-estimated any number of times until a stopping criterion is met - the stopping criterion being the correct classification of all the heteroplasmic mutations in the database with the minimum sum of false positive and false negatives. For example, a series of iterations may be performed for each of the different threshold parameters until the minimum possible sum of false positive and false negative results is reached. It will typically require thousands of iterations to reach the minimum possible sum, so this method of calculation is best performed by a computer program. Each successive iteration step may be set at an incremental value of, for example, 0.5%, but this incremental value may be reduced to 0.1 % or smaller.

The algorithm may be any appropriate algorithm known to those skilled in the art.

The algorithm may be a supervised learning algorithm. For example, the algorithm may be a classification algorithm. Examples of classification algorithms for use in the present invention include but are not limited to: artificial neural network algorithms, decision trees, genetic algorithms and Bayesian networks. The algorithm may be a regression algorithm. Examples of regression algorithms for use in the present invention include but are not limited to: linear regression, logistic regression, stepwise regression, ordinary least squares regression, multivariate adaptive regression splines, and locally estimated scatterplot smoothing. It is to be understood that the algorithm can be any algorithm that allows for the re-calculation of threshold values based on known mutations in a current data set so as to improve the accuracy of classifying a mutation as being associated with Parkinson's.

An optimisation step can also be incorporated into the methods of diagnosing a subject as having or being at risk of developing Parkinson's. In this way, the accuracy of the diagnostic method increases. The present invention therefore provides a process of optimising the method of diagnosing a subject as having or being at risk of developing Parkinson's, wherein the thresholds can be optimised using an algorithm performed on a data set of known heteroplasmic mtDNA mutations identified in patients with Parkinson's and heteroplasmic mtDNA mutations identified in healthy controls or healthy relatives.

The algorithm may be any appropriate algorithm known to those skilled in the art. The algorithm may be a supervised learning algorithm. For example, the algorithm may be a classification algorithm. Examples of classification algorithms for use in the present invention include but are not limited to: artificial neural network algorithms, decision trees, genetic algorithms and Bayesian networks. The algorithm may be a regression algorithm. Examples of regression algorithms for use in the present invention include but are not limited to: linear regression, logistic regression, stepwise regression, ordinary least squares regression, multivariate adaptive regression splines, and locally estimated scatterplot smoothing. It is to be understood that the algorithm can be any algorithm that allows for the re-calculation of threshold values based on known mutations in a current data set so as to improve the accuracy of diagnosing a subject as having or being at risk of developing Parkinson's.

The present methods allow for the collection of data relating to heteroplasmic mtDNA mutations. This data may be collected from both Parkinson's patients and healthy subjects. Certain parameter values may be recorded to create the data set, for example the parameters may include; the location of the mutation; the heteroplasmic frequency of the mutation; the percentage occurrence of the mutation in a human mtDNA database; and/or the difference in heteroplasmy between paired samples. A statistical learning algorithm may be applied to the data set such that the samples are classified correctly as being from either a Parkinson's patient or a healthy subject, with the lowest sum of false positives and false negatives. The output thresholds generated by the algorithm may then be applied to the method of identifying mutations associated with Parkinson's, and/or to the method of diagnosing a subject. This process may be repeated on any data set, and may be repeated as the size of the data set increases. In this way, the statistical algorithm can be used to improve the accuracy of the methods of the present invention effectively.

The invention will now be further understood with reference to the following non- limiting examples.

Example 1 : Method of diagnosing a subject as having or being at risk of developing Parkinson's

Method steps

1. For fibroblast samples obtain at least 1.5 million cells from the subject. For blood samples obtain at least 2ml of blood by venepuncture. For saliva samples use an OG-500 or similar kit (DNA Genotek) according to the manufacturer's instructions to collect the specified volume of saliva from the subject. DNA is stabilised with a proprietary stabilisation solution to give a stabilised DNA mixture. Stabilised DNA is stored at room temperature in a dark dry place until shipping, and shipped at room temperature for further processing.

For saliva samples, isolate non-methylated circular mitochondrial DNA from methylated circular bacterial DNA by digestion with Dpnl restriction endonuclease, an enzyme that specifically digests methylated DNA, prior to isolation of mitochondrial DNA by the Mseek method (step 3).

Use Mseek (US20150275200) or an alternative method to isolate circular mitochondrial DNA from the sample. 600-900μΙ of stabilised DNA mixture is used for mitochondrial DNA isolation by the Mseek method.

Sequence mitochondrial DNA an average of 2000 times at each nucleotide sequence position to accurately detect heteroplasmy.

Compare the sequence of the subject with that of the Revised Cambridge

Reference Sequence (rCRS Genbank number NC_012920) or other healthy control sample in order to detect nucleotide sequence variants.

Determine if any variants at a given nucleotide position fulfil all three of the following criteria;

a. Fall within ND5 or the Control region genetic loci

i. If variant falls within ND5, it must additionally be an amino acid- changing mutation

b. Heteroplasmy of variant is between 7% and 90%

c. ND5 variant occurs in less than 1 % of sequences, and Control Region variant occurs in less than 0.15% of sequences from the Mitomap database

If a variant fulfils all three of these criteria this indicates a high probability that the subject has, or is at risk of developing Parkinson's.

If that variant has been previously linked to Parkinson's in the Mitomap database, then this indicates an even greater level of probability that they have, or are at risk of developing Parkinson's.

If that variant is also found in a database of confirmed pathogenic Parkinson's variants then this indicates the highest level of probability that the subject has, or is at risk of developing Parkinson's. Results

Table 2: Heteroplasmic variants identified in a healthy male subject

Position Loots Variant Variant % ΑΛ & Database %

73 Corstroi a:g 90.5 - €4.8

150 Control c:t 95,5 1LS

152 Control t:c 95.0 23.0

185 Control g:c 93.5 O.Q

1721 165 r! A c:t 95.5 0.

2160 1SS rf¾ A 93.G 0.0

2706 .O 77.4

3197 leS rRNA i c 95.5 4. ϋ

4732 D2 a:g 96.5 N>S 0.7

5918 COl :c 95.5 0, 1

7028€01 c:t 95.5 - 79.0

7768 CG2 a¾ 91.5 - 1.8

947?€03 g:a 94.5 V>f 4.2

11457 i\D4 a:p 95.5 - 13.2

11719 MD g:a 94,0 - 75.4

12308 tR A leu 2 a:* 95.0 - 1111111111

12372 ^D5 g:a 95.5 - 14.1

13617 ¾D5 t:c 95.0 - 4,1

13637 ND5 3::g 93,5 Q R 0.9

14182 ^DS t:c 96.5 ■ 1111111:1111^

14323 ND6 $Ά 9S.5 - 0, The mtDNA sequence of the subject contains variants that cause amino acid changes (A. A Δ) in ND2 (NADH:ubiquinone oxidoreductase core subunit 2), C03 (Cytochrome c oxidase subunit 3), ND5 ((NADH:ubiquinone oxidoreductase core subunit 5), and CYB (Cytochrome b), compared to the Revised Cambridge Sequence for Human Mitochondrial DNA. However, these mutations do not occur with a heteroplasmic frequency of between 7% and 90%. The only variant that occurs with a frequency within this range is in the non- coding control region at position 16189, which occurs with a heteroplasmic frequency of 60%. However, this variant also occurs frequently in the Mitomap database, occurring in 26.6% of the sequences in the database, and thus according to the present methods, is unlikely to be associated with disease. The subject is diagnosed as not being at imminent risk of having or developing Parkinson's.

Example 2: Method of identifying one or more heteroplasmic mtDNA mutations associated with Parkinson's Method steps

1. For fibroblast samples obtain at least 1.5 million cells from the patient with

Parkinson's disease. For saliva samples use an OG-500 or similar kit (DNA Genotek) according to the manufacturer's instructions to collect 2ml saliva from the patient. DNA is stabilised with proprietary stabilisation solution to give 5ml of stabilised DNA mixture. Stabilised DNA is stored at room temperature in a dark dry place until shipping, and shipped at room temperature for further processing.

2. Obtain a sample from a healthy sibling or biological child of the patient with

Parkinson's in the same way as step 1.

3. For saliva samples, isolate non-methylated circular mitochondrial DNA from

methylated circular bacterial DNA by digestion with Dpnl restriction endonuclease, an enzyme that specifically digests methylated DNA, prior to isolation of mitochondrial DNA by the Mseek method (step 4).

4. Use Mseek (US20150275200) or an alternative method to isolate circular

mitochondrial DNA from the sample. 600-900μΙ of stabilised DNA mixture is used for mitochondrial DNA isolation by the Mseek method.

5. Sequence mitochondrial DNA an average of 2000 times at each nucleotide

sequence position to accurately detect heteroplasmy.

6. If subject is female, compare the sequence of the subject with Parkinson's with that of their child or sibling to identify any differences.

7. If the subject is male, compare the sequence of the subject with Parkinson's with that of their sibling to identify any differences.

8. In the subject with Parkinson's, determine if any of these differences in sequence at a given nucleotide position fulfil all the following criteria;

a. For genetic loci that code for proteins, it must be an amino acid-changing mutation

b. Heteroplasmy of a variant is between 7% and 90%

c. ND5 variant occurs in less than 1 % of sequences and control region

variant occurs in less than 0.15% of sequences from the Mitomap database

9. If a difference in sequence fulfils all these criteria, add that genetic locus to a

database of confirmed pathogenic variants observed in subjects with Parkinson's. Results

Table 3: Mother (P1) Son (H1) pair:

Vsitsrei 1% Ml* k % A.M Database

95,5 SS.5 ^■ 3.0

27-2 m 4 5

3107 165?ί*ΝΑ S7.S ios,a -23 o.o

NC 2 : : 13.?

4769 95,0 100,0 -4.Q §7.4

II: 100.0 ^■ 5.0 13.8

fc323 COI ^ : 100.0 •6.0 0,2

« f 2.5

m 3S.0 39.S Hill ^F 5J

14768 96.5 130.0 ■>3„s n>§ 1.3

CYB 3S..S I ;\C

15693 CYB 97.5 130.0 -2..S M T 1.2

It 265 95,5 i .: ^■ .: ^■ 4 5 ■ 11:11::

1635S 37.0 92,0 5,0 ^■ 2,1

. : : ii-;-.:- -7.5 · 20.0

18519 Capital S5.S 100,0 •O.S ·> 56,5 Analysis performed on a mother and son pair, aged 74 and 54, respectively. The mother is an individual with Parkinson's (P1) and her son (H1) is a healthy relative with related mtDNA. Table 3 lists data including the position, type and percentage heteroplasmy of each variant in both P1 and H 1. The percent-delta is calculated by subtracting the heteroplasmic frequency observed in the son from the heteroplasmic frequency observed in the mother. The percent-delta can be positive or negative. Highlighted in the table are heteroplasmic mutations of interest. The highlighted variants identified in P1 fulfil the following criteria:

(i) the mutation occurs in the non-coding control region (ii) the mutation occurs with a heteroplasmic frequency of between 7% and 90%

(iii) the Control Region variant occurs in less than 0.15% of the sequences in the Mitomap database.

(iv) there is a difference between the heteroplasmic frequency of the mutation in P1 compared to the frequency in H 1 , i.e. there is a delta heteroplasmic frequency of between 5% and 83%

Therefore, variants 514,— -:caca and 525,— -:acac are identified as being associated with Parkinson's. The data in Table 3 was used to create 3 'heteromaps' labelled 'Ρ1 ', Ή Τ and 'Delta'. This produces the diagram depicted in Figure 1. The heteroplasmic mutation frequency observed in H 1 is subtracted from the heteroplasmic frequency observed in P1 to determine the delta mutation frequencies. To generate the delta heteromap, the negative sign is ignored (if present) and all delta mutation frequency data is treated as positive.

Table 4: Sister (P2) Sister (H2) pair

Position Lams Va iant ?2 % H2 % Delta % A.A ώ Database %

263 Control a:g 97.5 97.Q O.S - 77JS

Analysis performed on a woman aged 81 with Parkinson's (P2) and her healthy sister (H2) aged 79 who shares related mtDNA. Table 4 lists data including the position, type and percentage heteroplasmy of each variant in both P2 and H2. The highlighted variant identified in P2 fulfils the following criteria:

(i) the mutation occurs in the non-coding control region

(ii) the mutation occurs with a heteroplasmic frequency of between 7% and 90%

(iii) the Control Region variant occurs in less than 0.15% of the sequences in the Mitomap database. (iv) there is a difference between the heteroplasmic frequency of the mutation in P1 compared to the frequency in H1 , i.e. there is a delta heteroplasmic frequency of between 5% and 83%

Therefore, variant 302, a:c is identified as being associated with Parkinson's.

The data in the Table 4 was used to create 3 'heteromaps' labelled 'Ρ2', Ή2' and 'Delta'. This produces the diagram depicted in Figure 2.

Table 5: Data set of known heteroplasmic mtPNA mutations associated with Parkinson's 0 as identified by the methods of the present invention

Confirmation step using computational tools

The inventors have identified a positive correlation between scores generated by 5 computation pathogenicity prediction tool 'Mitlmpact' and mutations identified by the methods of the present invention in patients with Parkinson's. Computational prediction methods can be used to reveal the impact that a specific mutation has on the overall nucleotide and/or protein structure. It is to be understood that a mutation that disrupts the nucleotide and/or protein structure may disrupt cellular functions and so has a greater 0 probability of being associated with disease. Mitlmpact provides an exhaustive collection of pre-computed pathogenicity predictions of human mitochondrial non-synonymous variants. Different computational models may predict whether a given mutation is likely to be damaging, or have a neutral impact. The results from a number of different computational models can be combined and the mutation assigned a score, termed a 5 'Mitlmpact score'. Table 6 shows the results of 21 different computational models. In general, Parkinson's individuals have higher Mitlmpact scores for their ND5 variants than healthy individuals.

Table 6:

The present invention is not to be limited in scope by the specific embodiments described herein. Indeed, various modifications of the invention in addition to those described herein will become apparent to those skilled in the art from the foregoing 10 description and accompanying figures. Such modifications are intended to fall within the scope of the appended claims. Moreover, all aspects and embodiments of the invention described herein are considered to be broadly applicable and combinable with any and all other consistent embodiments, including those taken from other aspects of the invention (including in isolation) as appropriate.

Various publications and patent applications are cited herein, the disclosures of which are incorporated by reference in their entireties.