PFBC–SLC20A2

The first PFBC gene, SLC20A2, was identified in 2012 by fine-mapping of the IBGC3 locus and subsequent candidate gene sequencing [13]. In the same study, six other SLC20A2 mutations were found, leaving little doubt as to whether or not SLC20A2 was the right candidate. In more than 60 publications that followed, nearly 200 SLC20A2 mutation carriers, mostly of Asian and Caucasian origin, were reported. These carry more than 50 different SLC20A2 sequence changes, mostly missense, nonsense, and frameshift mutations [16, 22]. The majority of the changes are novel, i.e. not reported in public databases such as ExAC [23], gnomAD [23], or dbSNP [24], and are predicted to be highly deleterious (reflected in an average combined annotation dependent depletion (CADD) [25] score of >30, range: 14–48) [22]. Importantly, five large deletions, involving single exons (2 and 4) [26], multiple exons (4–5 and 6–10) [15, 26], and even the entire SLC20A2 gene [27] were also discovered, stressing the importance of gene dosage analysis in SLC20A2. The coding region of SLC20A2 encompasses 1,959 base pairs, spanning 11 exons. There seem to be no mutational hotspots and the reported variants can be found in each of the exons, with only a few changes detected in more than one index patient. Interestingly, among the SLC20A2-tested individuals, there were also members of the two large families initially linked to the IBGC1 and IBGC2 loci. They were found to carry deletions (single-nucleotide; c.508delT, and multiexonic; c.(613+1_614–1)_(1794+1_1795–1)del, respectively) [14, 15] in SLC20A2, collapsing loci IBGC1-3 into a single locus, IBGC1.

PFBC–PDGFRB

PRGFRB became implicated in PFBC by means of exome sequencing, carried out in a large three-generational French family [18]. This analysis revealed a missense PDGFRB change, completely segregating with the brain calcification phenotype [18]. However, only a few of the subsequent PDGFRB screening studies reported mutation carriers, indicating that changes in this gene are a cause of PFBC but not a very frequent one. Currently, only six different PDGFRB mutations (five missense and one variant of unknown effect, affecting the first codon) in three PFBC families and three sporadic patients have been identified [16, 22]. Three of the six variants are novel and three can be found at a very low frequency in ExAC (with CADD scores around 29, ranging from 23 to 34) [22]. The PDGFRB coding region has 3,321 base pairs, comprises 22 exons, and to date, mutations were found in the coding exons 1, 13–15, 21, and 22.

PFBC–PDGFB

Shortly after variants in PDGFRB had been linked to PFBC in 2013, mutations in a functionally related gene, PDGFB, have been discovered through next-generation sequencing in two families with basal ganglia calcification [19]. Since then, more than 50 predominantly Caucasian and Asian individuals, carrying ~15 different (mostly missense and nonsense) PDGFB mutations have been identified [16, 22]. One large structural change, a deletion of exons 3–5, has been reported [28]. Interestingly, none of the changes found in these PFBC patients is listed in ExAC. The mean CADD score of known PDGFB mutations is 29, with a range of 11–41 [22]. In comparison to SLC20A2 and PDGFRB, PDGFB is relatively small and its coding region consists of 726 base pairs encompassing six exons. One-third of the mutations are situated in exon 4, and three changes were reported in two independent studies each [22].

PFBC–XPR1

XPR1 is the most recently identified gene for autosomal-dominant PFBC. The initial study, which uncovered the very first XPR1 mutations, utilized exome sequencing in a large family of Swedish ancestry, also reported two more families and two sporadic patients with different XPR1 changes [20]. To date, only one additional publication detected an XPR1 mutation in a single patient [29], thus increasing the number of PFBC families/sporadic patients to six (all of the European origin), which collectively carry five different XPR1 variants. All of those five mutations are missense variants, absent from the public databases, predicted to be pathogenic (CADD scores of 27–29) [22], and situated in the first third of the 2,091 base-pair-long coding region, in exons 4 and 6 of 15 exons.

PFBC–MYORG

In 2018, after excluding the previously known PFBC causes in their collection of 51 families with bilateral brain calcifications, Yao and colleagues decided to focus on the subset of pedigrees with a seemingly autosomal-recessive mode of inheritance [21]. As a starting point, they performed exome sequencing in a family with documented consanguinity and detected homozygous mutations in MYORG and four other genes. Exome sequencing in another autosomal-recessive family revealed compound-heterozygous variants in MYORG, thus determining this as a nominee for further mutational screening efforts. Indeed, the authors discovered one more family with homozygous and three with compound-heterozygous MYORG variants. Of the nine different MYORG mutations found in the six autosomal-recessive families, three were nonsense, four were missense, and two were in-frame insertions [21]. The mutations were evenly distributed across the 2,145 base-pair-long coding region located within the single exon, and four were novel (i.e. absent from the publicly available databases). The CADD scores ranged between 22 and 39.

PiT2

The solute carrier family 20, member 2 gene (SLC20A2) encodes a 652-amino acid-long and ubiquitously expressed type III sodium-dependent Pi transporter 2 (PiT2), responsible for uptake of Pi into the cell (Figure 18.1a). PiT2 is integrated into the plasma membrane through 12 transmembrane regions but also contains one large intracellular domain [30]. PFBC-causing changes severely decrease the ability of the mutant PiT2 to import Pi, thus, likely leading to an extracellular accumulation of this anion, and ensuing build-up of calcium phosphate [13]. Slc20a2 homozygous knockout mice are viable and, at 19 weeks of age, show calcifications in the thalamus, basal ganglia, and cortex [31], which are absent in slightly older wild-type mice. Eight-week-old knockout mice already display histologically detectable and weakly calcified nodules in the brain, which become larger, more abundant, and more calcified as these mice age [32]. Almost all of the aggregates were associated with blood vessels, and electron microscopy indicated that calcifications are located intracellularly, mainly in pericytes and astrocytes [32]. In agreement with the autosomal-dominant inheritance pattern seen the PFBC patients, two reports described the presence of calcium aggregates in heterozygous knockout mice, first appearing at the age of 6 months [32, 33]. However, while one study found weak calcification in one out of seven investigated mice [32], other studies labeled the detected bilateral calcifications as “prominent” [33]. The discrepancy was hypothetically attributed to environmental differences [32].

Figure 18.1 Schematic representation of the putative pathogenic mechanism of PFBC. (a) Pi in the brain parenchyma is taken up by PiT2 into the cells (possibly astrocytes and/or pericytes). From astrocytes/pericytes, Pi is exported through XPR1. If mutated, PiT2 is unable to import Pi, leading to an accumulation of Pi and subsequently of calcium phosphate in the extracellular matrix, and cell death. Mutated XPR1 impairs Pi transport outside of the cell, leading to intracellular accumulation of calcium phosphate and cell death. (b) Structure of the NVU and recruitment of pericytes by endothelial cells. The NVU is an interactive network of vascular cells (pericytes and endothelial cells), glia (astrocytes), and neurons. During angiogenesis, PDGF-B is synthesized and secreted by angiogenic endothelial tip cells and retained in close vicinity to the growing blood vessel by the extracellular matrix. Incoming pericytes express PDGF-Rβ and are, hence, attracted to the PDGF-B-decorated endothelial cells, and depend on PDGF-B for proliferation and correct migration along the forming blood vessel.

Interestingly, in the mouse brain, Slc20a2 is expressed in tissues that produce and/or regulate cerebrospinal fluid (CSF) [33], such as the choroid plexus and arteriolar smooth muscle cells, and Pi levels in the CSF of the knockout mice are ~2.5-fold higher than in their wildtype littermates [34]. Hence, PiT2 might be important in sustaining the low levels of Pi in the CSF by exporting Pi from the CSF to the blood [34].

As previously explained, both nonsense and missense mutations in SLC20A2 cause PFBC. While haploinsufficiency represents an obvious genetic mechanism behind the heterozygous null alleles, the same was initially suggested for the missense mutations (e.g. co-expression of the missense allele did not interfere with the activity of the wild-type protein Xenopus laevis oocytes, i.e. no dominant-negative effect was observed) [13]. Nonetheless, it is conceivable that at least some of the missense mutations may confer a dominant-negative effect. This was supported by a significantly (more than twice, as would be expected from a simple lack of one functional allele) lowered expression of PiT2 in the brain (and in particular in the astrocyte processes) of an autopsied p.Ser637Arg mutation carrier, in comparison to age-matched neurologically healthy controls [35]. In addition, three different mutants, each co-expressed with the wild-type PiT2 in fibroblasts of the homozygous knockout Slc20a2 mice exhibited a dominant-negative effect evident from a significant decrease in Pi-uptake ability in comparison to cells only transfected with wild-type Slc20a2 [36]. The possible dominant-negative effect may be explained by the fact that the functional PiT2 acts in a form of a homodimer. Thus, if a wild-type and mutated PiT2 oligomerize, they may be identified by the cell’s quality-control system, and degraded.

XPR1

The xenotropic and polytropic retrovirus receptor 1 (XPR1) gene codes for a 696-amino acid multipass transmembrane Pi exporter. The XPR1 mutations linked to PFBC have been shown to impair Pi efflux from the cell, thus likely increasing intracellular Pi levels (Figure 18.1a), leading to calcium phosphate precipitation [20]. Although the XPR1 protein is ubiquitously expressed, particularly high amounts of XPR1 messenger RNA were detected in mouse brain and were peaking in the frequently calcified regions such as the cerebellum and striatum [37]. Functional studies indicate that PFBC-causing XPR1 mutations might act through haploinsufficiency rather than via a dominant-negative effect [37]. Interestingly, analysis of the relation of XPR1 to the other PFBC-related proteins showed no interaction with PiT2 or PDGF-B. However, XPR1 co-localizes with PDGF-Rβ and may form a complex with this receptor, and such ability may be impaired by PFBC-inducing mutations [37].

PDGF-Rβ

PDGF-Rβ is a 1,074-amino acid-long protein cell-surface tyrosine kinase receptor. It consists of an extracellular domain (that binds homodimerized PDGF-B), a transmembrane region, and an intracellular kinase domain that spans amino acid residues 600 to 962. Upon binding of the ligand, PDGF-Rβ becomes autophosphorylated, thus triggering downstream signaling pathways (e.g. cellular proliferation, differentiation, and migration) [40]. The two PFBC-associated mutations situated within the kinase domain, p.Leu658Pro and p.Arg695Cys, interfere with autophosphorylation by completely or partially abolishing it and thus impairing the receptor signaling. Another investigated change p.Arg987Trp was found to decrease protein stability [41, 42]. PDGFRB is the only PFBC-relevant gene in which missense mutations can cause conditions that do not have a primarily neurological phenotype. Namely, several heterozygous missense variants in PDGFRB, none of which have been reported in PFBC patients, are responsible for (i) a common benign soft-tissue tumor in childhood and infancy (infantile myofibromatosis; OMIM 228550), (ii) Penttinen syndrome of premature aging syndrome (PENTT; OMIM 601812), and (iii) Kosaki overgrowth syndrome (KOGS; OMIM 616592), the latter two being caused by de novo changes [43]. Importantly, mutations leading to the non-PFBC phenotypes are considered to be the gain-of-function hypermorphic changes, whereas the calcification-inducing loss-of-function variants are regarded as hypomorphic [43].