Dissecting the Phenotype and Genotype of PLA2G6‐Related Parkinsonism

Complex parkinsonism is the commonest phenotype in late‐onset PLA2G6‐associated neurodegeneration.

A BS TRACT: Background: Complex parkinsonism is the commonest phenotype in late-onset PLA2G6associated neurodegeneration. Objectives: The aim of this study was to deeply characterize phenogenotypically PLA2G6-related parkinsonism in the largest cohort ever reported. Methods: We report 14 new cases of PLA2G6-related parkinsonism and perform a systematic literature review. Results: PLA2G6-related parkinsonism shows a fairly distinct phenotype based on 86 cases from 68 pedigrees. Young onset (median age, 23.0 years) with parkinsonism/ dystonia, gait/balance, and/or psychiatric/cognitive symptoms were common presenting features. Dystonia occurred in 69.4%, pyramidal signs in 77.2%, myoclonus in 65.2%, and cerebellar signs in 44.6% of cases. Early bladder overactivity was present in 71.9% of cases. Cognitive impairment affected 76.1% of cases and psychiatric features 87.1%, the latter being an isolated presenting feature in 20.1%. Parkinsonism was levodopa responsive but complicated by early, often severe dyskinesias. Five patients benefited from deep brain stimulation. Brain magnetic resonance imaging findings included cerebral (49.3%) and/or cerebellar (43.2%) atrophy, but mineralization was evident in only 28.1%. Presynaptic dopaminergic terminal imaging was abnormal in all where performed. Fifty-four PLA2G6 mutations have hitherto been associated with parkinsonism, including four new variants reported in this article. These are mainly nontruncating, Little is known about late-onset PLAN progression because ongoing natural history studies mainly focus on INAD. 17,18 Finally, treatments with disease-modifying potential have not hitherto been explored in late-onset PLAN. 19 Some PLAN cases show brain iron deposition, thus raising the option of chelation therapy, as in pantothenate kinase-associated neurodegeneration. 20 More promisingly, small molecule therapies are under investigation in cell and murine models, and a viral vectorbased gene therapy has been tested in the PLA2G6-INAD mouse with encouraging results and is approaching completion of preclinical studies, 15,21 as reviewed elsewhere. 19 Promptly recognizing PLA2G6related phenotypes, in particular PLA2G6 parkinsonism among early-onset parkinsonism from different etiologies, may therefore have considerable therapeutic implications in the not-too-distant future.
We report 14 new cases of PLA2G6-associated parkinsonism carrying 13 different mutations, 4 of which are novel. By merging data from this series and a systematic literature review, we deeply characterize phenotypically and genotypically PLA2G6-related parkinsonism, highlight clinicoradiological hints for diagnosis, outline its natural history, and discuss poorly understood issues in late-onset PLAN.

Subjects and Methods
We identified new cases of PLA2G6-associated parkinsonism from six centers and systematically searched PubMed (14/03/2021) for parkinsonism in genetically confirmed PLAN published since 2006 (discovery paper). The search strategy was "PLA2G6" AND "parkins*," with no language restriction. Additional references from relevant articles were identified and reviewed. According to the search strategy, the systematic review was driven by the presence of parkinsonism in genetically confirmed PLAN cases; thus, subsequent results were not restricted to any age at symptom onset. Only cases carrying biallelic PLA2G6 mutations with individual information were included. Predefined categories for data extraction were sex; ethnicity; age and symptom(s) at onset; age at last assessment; different neurological and psychiatric symptoms/signs; findings from laboratory, neuroimaging, and other investigations; family history; genotype; and treatment response. Neuropathology specimens available at Queen Square Brain Bank and videos of published cases were reviewed. Phenotypic features were recorded as nonmissing if explicitly stated to be present/absent. PLA2G6 variants were (re-)annotated in reference to transcript NM_003560.4. Combined Annotation Dependent Depletion, PolyPhen-2, Sorting Intolerant From Tolerant, MutationTaster, and Protein Variation Effect Analyzer were used to predict the impact of missense mutations on the protein structure and function. We assessed sequence conservation across species using Genomic Evolutionary Rate Profiling and/or visual multiple sequence alignment (Clustal Omega). 22 gnomAD and ClinVar were retrieved on August 11, 2021. Mutations were finally classified according to American College of Medical Genetics and Genomics guidelines. 23 Descriptive statistics were performed using IBM SPSS Statistics. Results are provided as valid percentages (ie, counts divided by the total number of nonmissing observations) for dichotomous variables and median with interquartile range (IQR; weighted average) for continuous variables.
Eye movement abnormalities were reported in 24/41 (58.5%; Fig. 1E) cases, with fragmented pursuit and/or reduced gaze range in the vertical plan being the most frequent. Eyelid-opening apraxia was reported in six cases. 8,24,25,55 Dysarthria was described in 29/30 (96.7%; Fig. 1E) cases. Swallowing difficulties were reported in 17/20 (85.0%) cases (Fig. 1E) and, in 6 cases with details available, 7,8,12,16,24,30,51 overt dysphagia and/or the need of percutaneous endoscopic gastrostomy occurred a median interval of 10 years (IQR, 7.0) after symptom onset. Sensory signs were reported in one case. 16 Generalized seizures occurred in five patients some years into the disease course. 8,16,24,31,39 Oculogyric crises were reported in four cases. 10,30,58 Parkinsonism responded to levodopa in 71/73 (97.3%) cases. Levodopa-induced dyskinesias were reported in 46/57 (80.7%) cases and appeared within the first year of treatment in most cases, even with low levodopa doses (≤300 mg/day). In some cases, levodopa caused behavioral changes with psychotic manifestations, or dystonic reactions mainly affecting the oromandibular or cervical region. In 17/18 cases, dopamine agonists had a beneficial response; however, some cases experienced negative side effects, including psychosis and hypersexuality. Response to other pharmacological treatments is detailed in Supporting Information File 3. Four patients underwent bilateral deep brain stimulation (DBS) of the subthalamic nucleus (STN) with excellent outcome, 57 and one patient underwent bilateral DBS of the GP internus (GPi; case 1) with improvement of trunk dystonia and control of levodopa-induced dyskinesias. One patient underwent unilateral pallidotomy with transient relief of motor fluctuations. 52,53 Five patients died at a median age of 36.0 years (IQR, 13.0), showing a median disease duration of 13.0 years (IQR, 14.5).
Brain MRI was available in 82/86 (95.3%) patients. Cerebral atrophy was reported in 34/71 cases (47.9%), being described as mild to moderate in severity and generalized or mainly involving the frontotemporal lobes ( Fig. 1E; Supporting Information File 2B,F,G,L, O). Mild to marked cerebellar atrophy affecting the vermis and/or the hemispheres was observed in 32/81 (39.5%; Fig. 1E; Supporting Information File 2B,G,L, N) cases. In 21/82 (25.6%) cases, there was evidence of iron deposition on MRI-T2/T2*/SWI sequences ( Fig. 1E; Supporting Information File 2D-I). Interestingly, iron deposition was not detected in 15 cases in which T2*/SWI sequences were performed ( Fig. 1E; Supporting Information File 2A-P). White matter T2 hyperintensities were reported in three cases, 8,24,40,56 mainly in the frontal lobes. Additional findings on brain MRI (Fig. 1E) were claval hypertrophy in four patients, 7,12,14,31,56 vertically oriented corpus callosum in two, 14,31 and the swallow tail sign in two. 61 Followup brain MRI scans were available in seven patients, showing progression of cerebral and/or cerebellar atrophy in three cases, 16 as well as appearance of iron deposition in two. 8,24,51 In three patients, CT scan excluded MRI-T2/SWI hypointensity corresponding to basal ganglia calcifications. Spine MRI, available in eight cases, showed no signal abnormalities from the spinal cord.
Dopamine imaging with presynaptic tracers was available in 37/86 cases (43.0%) and invariably abnormal ( Fig. 1E; Supporting Information File 2C,E,H,M, Q). In one patient, 11 C-raclopride (RAC)-positron emission tomography for postsynaptic receptor function revealed increased RAC uptake in the putamen more than in the caudate. 25 18 F-fluorodeoxyglucose-positron emission tomography, performed in seven cases, showed global hypometabolism of cerebral cortex and cerebellum in one case, 47 hypometabolism in the frontoparietal regions in three, 29 and hypometabolism in the temporoparietal regions or parieto-occipital lobes in one case each. 10,39 EEG was abnormal in four of eight cases where reported, with diffuse slowing and multifocal epileptiform abnormalities a few years into the disease course. 16,24,31,59 Nerve conduction studies were performed in seven patients, showing signs of distal sensory neuropathy in two. 7,16 Four patients with pyramidal signs underwent motor-evoked potentials (MEPs), which showed delayed central motor conduction time in two cases, including case 1. 50 In case 12, motor-evoked potential was abnormal despite the absence of clinically detectable pyramidal signs.
The new cases carried 13 distinct PLA2G6 mutations, including four novel missense variants (c.956C>T, c.1924A>G, c.2311G>A, and c.1937>T), four missense variants previously reported in only childhood-onset phenotypes (c.1061T>C, c.673C>T, c.1021G>A, and c.1898C>T), and one nonsense (c.2370T>G) and three missense variants (c.238G>A, c.2222G>A, and c.2239C>T) described in both childhood-and late-onset phenotypes ( Table 2). The novel variants were either absent (n = 2) or exceedingly rare (minor allele frequency, MAF < 0.1%; n = 2) in gnomAD, highly conserved across species, and predicted pathogenic by at least four predetermined prediction tools (Table 2). Among variants previously reported in only childhood-onset phenotypes, the variant c.1061T>C was in compound heterozygosity with another missense variant in a case of neurodegeneration with brain iron accumulation (NBIA) 3 and with nonsense and frameshift variants in two INAD cases. 3,16 The variant c.238G>A was found in compound heterozygosity with a nonsense variant in an NBIA case 3 and in homozygosity in two cases (onset 8 and 14 years, respectively). 25,26 The variant c.1021G>A was previously found in compound heterozygosity with a missense variant in two siblings with INAD 3 and in homozygosity in another INAD case, 36  variant c.673C>T was detected in the homozygous state in two kindreds with childhood-onset cases. 27 Among the variants reported in both childhood-and late-onset phenotypes, the nonsense variant c.2370T>G was in compound heterozygosity with a missense variant in case 7 (age at onset: 25 years) and found in the homozygous state in an INAD case 3 and in compound heterozygosity with another nonsense variant (c.1674del) in two sisters with INAD. 33 Overall, these observations suggest a gradient in the age of onset and phenotype severity reflecting variants' impact on the transcript.
Enzymatic activity of the mutant iPLA2β was documented in only one included report. 45 Transfection of cDNA encoding PLA2G6 carrying the homozygous variant c.991G>T into HEK293T cells revealed about 30% of residual protein activity compared with the wild-type protein. 45 Pathology Brain pathology has been reported in three cases. 16,48,52,53 In one case, brain autopsy revealed cerebellar cortical atrophy with severe loss of granule cells, gliosis in the molecular layer, and to a lesser extent, Purkinje cell depletion. Cytoarchitecture of the dentate and inferior olivary nuclei was comparably well preserved. There were occasional neuroaxonal spheroids in the GP, STN, thalamus, and ambiguus nucleus, and frequent ones in the gracile and cuneate nuclei and posterior horns of the spinal cord. There was severe atrophy of the ventrolateral parts of the SN with better pigmented neuron preservation medially. The locus ceruleus showed mild depletion, and the vagus nerve nucleus in the medulla showed no evidence of severe neuronal loss. Nevertheless, Lewy bodies were present in the brainstem nuclei in the tegmentum and frequently in the less atrophic medial part of the SN, in the Meynert nucleus, medial temporal lobe, and occasionally in the putamen. Lewy pathology was particularly widespread across the deep layers of the neocortex affecting all, including occipital, lobes (Braak stage 6).
There was also very mild tau pathology in the medial temporal lobe with rare neurofibrillary tangles and occasional neuropil threads (Fig. 3). 16 One patient underwent brain biopsy of the frontal cortex, which showed severe Lewy pathology and moderate tau pathology (neuropil threads). 16 Postmortem evaluation in another case showed widespread cortical and limbic structure atrophy, Lewy bodies in the SN and locus ceruleus, Alzheimer's disease-like pathology, mainly in the temporal lobe structures, abundant gliosis detected by glial fibrillary acid protein, and excessive iron accumulation in the reticularis portion of the SN, but also the GP and ventral forebrain. 52,53 No cases hitherto reported had nerve or rectal biopsy available. Muscle biopsy was unremarkable in three cases and showed neurogenic changes and reduced cytochrome oxidase activity (complex IV) in one case. 7,12,16 Skin biopsy, which, however, did not search for α-synuclein, was unremarkable in three cases. 8,24 Bone marrow aspiration was normal in two cases. 8,24 Discussion We provided in-depth phenotypic and genotypic characterization of the largest cohort of PLA2G6related parkinsonism hitherto reported. Distinct aspects from this phenogenotypic overview are discussed later.
Although the age of symptom onset ranged between the first and seventh decades of life, most cases manifested between late second and third decades of life. Patients with earlier onset 7,12,56 showed overlapping clinicoradiological features with ANAD cases but lacked the typical rapid deterioration in late childhood, 4 and instead had slower symptom progression with parkinsonism from the second decade of life, which supports the notion of PLAN phenotypes as a phenotypic continuum. 12,15 Psychiatric features were frequently observed early in the disease course, often preceding extrapyramidal manifestations, and were the only presenting symptoms in one-fifth of the cohort. 10,16,25,29,30,44,59,60 This evidence should warn clinicians against misdiagnosis of psychiatric disorders and initiation of potentially detrimental treatments (ie, antipsychotics) in the context of (at least preclinical) extrapyramidal involvement. 29 Interestingly, onset or rapid deterioration of symptoms in PLA2G6-related parkinsonism occurred during pregnancy/postpartum (cases 2, 3, and 11) 30 or in vitro fertilization (case 1) in five cases. This is in keeping with likely hormonerelated symptom deterioration during pregnancy and, most often, the perimenopausal and postmenopausal period previously reported in idiopathic Parkinsonʼs disease (PD). 65 These observations, along with the age of onset and high prevalence of psychiatric manifestations, suggest that female individuals with PLA2G6 parkinsonism can be at risk to be misdiagnosed as having pregnancy-or postpartum-related psychiatric morbidity, autoimmune encephalitis, or functional motor disorders in the early disease stages.
Parkinsonism frequently showed dramatic, albeit unsustained, response to levodopa. Levodopa efficacy was most often limited by the occurrence of severe levodopa-induced dyskinesias, as well as exacerbation of psychiatric symptoms a few weeks to a few years after treatment initiation. We suggest that early-onset levodopa-induced dyskinesias are a tell-tale sign and, along with other clinicoradiological red flags discussed in this article, essentially limit the differential diagnosis to Kufor-Rakeb syndrome. Five patients underwent bilateral DBS (STN or GPi) with good to excellent outcome, 54,55,57 thus making DBS a potential option in early disease stages with intractable treatment fluctuations. Dystonia was the second most frequent motor feature in PLA2G6-related parkinsonism. Isolated foot dragging was the presenting feature in $10% of cases, 8,24,40,45 thus initially suggesting PRKN-PD. 66 We noticed a high prevalence of extensor truncal dystonia in our cases and previous literature, thus confirming dystonic opisthotonus as a hint to suspect NBIA syndromes. Oculogyric crises were also observed, 10,30,58 thus expanding the spectrum of disorders possibly manifesting with these paroxysmal dystonic manifestations. 67,68 Finally, oromandibular dystonia seemed less prevalent than in pantothenate kinase-associated neurodegeneration. Although pyramidal features are not particularly diriment in early-onset parkinsonian, pallidopyramidal, or NBIA syndromes, (even subtle) cerebellar signs could point toward PLA2G6. Finally, myoclonus was frequently reported a few years into the disease course, and we first proved its cortical origin on neurophysiology in case 2.
Cerebellar atrophy was detected on MRI in more than 40% of cases. Genetic PLA2G6 ablation in mice causes cerebellar atrophy, with loss of Purkinje cells, reactive astrogliosis, microglia activation, and upregulation of proinflammatory cytokines. Less frequent and (in most cases) severe degree of cerebellar atrophy in late-onset PLAN might reflect higher residual iPLA 2 β activity. It could also correlate with disease duration because we documented cerebellar atrophy on followup MRI in three cases where it was not initially detected. Iron deposition was found in $25% of cases. Interestingly, it was often not documented on dedicated MRI-T2*/SWI sequences. When available, dopamine imaging invariably documented signs of nigrostriatal degeneration. Notably, in one patient, RAC for postsynaptic receptor function revealed increased RAC uptake in the putamen more than in the caudate, thus indicating a largely presynaptic dopaminergic abnormality as seen in idiopathic PD. 25 Pathogenic PLA2G6 mutations causing different PLAN phenotypes are scattered throughout protein domains and may therefore impair iPLA 2 β function through a variety of loss-of-function mechanisms, affecting either its enzymatic activity, regulation, or interactions at the macromolecular level. No mutations hitherto linked to PLA2G6 parkinsonism affect the primary structure of iPLA 2 β domains responsible for its enzymatic activities. Most of them are nontruncating and might determine less detrimental effects than truncating variants, which are found more frequently in INAD/ANAD. As previously suggested, different mutation sites in iPLA 2 β domains can lead, directly or indirectly, to different changes in its enzymatic activity, which might be a critical factor in the phenotypic heterogeneity of PLAN. 69 This is supported by the observation that all individuals with two null alleles manifest with INAD, while most patients with ANAD or late-onset phenotypes carry two missense PLA2G6 mutations. 15 Furthermore, it is consistent with the few biochemical and enzymatic studies available. Engel et al. 69 demonstrated in vitro that mutations associated with different PLA2G6 phenotypes have different impact on its catalytic activity. Mutations associated with INAD/NBIA cause loss of enzyme activity, with mutant proteins exhibiting less than 20% of wild-type enzymatic activity in both lysophospholipase and phospholipase assays. In contrast, three mutations associated with dystonia-parkinsonism (c.1894C>T, c.2222G>A, and c.2239C>T) do not impair phospholipase or lysophospholipase catalytic activity. Shi et al. 45 documented that the PLA2G6 variant c.991G>T is associated with approximately 30% residual protein activity compared with the wild-type iPLA 2 β. Finally, Zhou et al. 70 found defective activation of endogenous store-operated Ca 2+ and iPLA 2 β activation in cells from a patient with familial PD carrying the c.2239C>T mutation in PLA2G6. Functional studies of the tertiary/quaternary structure of mutant iPLA 2 β and its residual enzymatic activity/regulation are needed to further explore phenotype-genotype correlations.
Limitations of this study are mainly attributable to its retrospective nature. Frequencies of clinicoradiological features were calculated as valid percentages, and this, or any possible alternative imputation, is not exempt from risk for underestimation or overestimation. Furthermore, considerations on the natural disease history were limited by the lack of precise information on symptom/sign onset and follow-up in most published cases. Nevertheless, we are confident this is the most comprehensive analysis of the largest population with PLA2G6-related parkinsonism hitherto reported. Finally, we acknowledge that the search strategy did not allow to compare PLA2G6 cases with and without parkinsonism in perspective of gene-specific therapies. However, we believe that clinicoradiological clues highlighted by this study might increase the clinical suspicion and prompt genetic testing for PLA2G6 parkinsonism among different etiologies of early-onset parkinsonism, thus favoring precision medicine in the not-too-distant future.
In conclusion, biallelic PLA2G6 mutations cause early-onset parkinsonism with dystonia, pyramidal and cerebellar signs, myoclonus, and cognitive impairment. Early psychiatric manifestations and bladder overactivity are common. Early, severe dyskinesias are a tell-tale sign. Against this background, irrespective of whether iron deposition is present on brain MRI, the detection of cerebellar atrophy points toward PLA2G6 among other genetic causes of early-onset parkinsonian, pallidopyramidal, and NBIA syndromes. These clinicoradiological features should prompt PLA2G6 mutation analysis, which might have considerable therapeutic implications in the near future, given promising preclinical disease-modifying strategy development. 19