Seckel syndrome is an autosomal recessive disorder of intra-uterine growth retardation, severe proportionate short stature and marked microcephaly1,2
. The ataxia-telancgiectasia and Rad3-related (ATR)
gene is mutated in Seckel syndrome (ATR-Seckel)4
, and encodes a phosphotidylinositol-3-kinase-like kinase which has distinct, but overlapping functions with ATM, in co-ordinating the response to DNA damage10
. ATR is activated by single stranded DNA whilst ATM responds to DNA double strand breaks. MCPH1, encoded by another gene whose mutations cause severe (primary) microcephaly, also acts in the ATR damage response pathway7
. Furthermore, other, non-ATR Seckel patient cell lines also exhibit ATR-pathway dysfunction3
, suggesting that mutation of further genes encoding proteins involved in this DNA damage response cascade may cause this disorder.
To identify such components, we performed a SNP-microarray genome-wide homozygosity scan on two consanguineous families from the Middle-East with a clinical diagnosis of Seckel syndrome (Supplementary Fig. 1 and 2
) and showing cellular evidence of defective ATR signalling. Multipoint linkage analysis identified a novel locus, Sckl4
, with a maximum LOD score of 4.03 at 70.9cM on chromosome 21q22.3 between rs1598206 and rs2330591 ( and Supplementary Fig. 2
). 55 Refseq annotated genes lay within this interval, including pericentrin (PCNT
), which encodes a centrosomal protein. As several centrosomal genes, CenpJ
cause the related condition, primary microcephaly8,9
, we postulated that PCNT
might be the causative gene and therefore sequenced its 47 coding exons in affected individuals from both families.
Schematic of the Sckl4 critical region and the PCNT gene depicting location of identified mutations
We identified a homozygous nonsense mutation in exon 4 of family 1 (E220X). A homozygous single base-pair deletion in exon 12 in family 2 (S629fs, ) produces a frameshift, predicted to result in premature protein truncation after an additional 65 amino acids. Both mutations segregated with the disease in each family, with all parents being heterozygous for the respective mutations. From screening of additional cases we identified a further patient with a homozygous single base-pair insertion in exon 18, resulting in a frameshift at codon 1190 (C1190fs). Neither this, nor the other two mutations were present in over 200 control alleles screened. Collectively, these mutations disrupt all described mammalian protein isoforms of PCNT11,12
. We therefore concluded that homozygous truncating mutations in PCNT
cause Seckel syndrome.
Pericentrin (also known as kendrin) is a huge 360kDa coiled-coil protein with a C-terminal PACT domain that targets it to the centrosome (). Two protein isoforms have been described in humans11
, the full length protein, pericentrin B, and a C-terminally truncated isoform, pericentrin A. Pericentrin localises to the pericentriolar material (PCM), where it interacts with several structural centrosomal proteins, including γ-tubulin and PCM113,14
and is important for microtubular nucleation and spindle organisation15,16
Anti-pericentrin antibodies disrupt mitosis, suggesting it is essential for mitotic progression15
. However, its Drosophila
homolog, D-PLP, though required for timely recruitment of PCM components, is apparently dispensible for cell division17
. Furthermore, as well as its structural roles, pericentrin acts as a scaffold to recruit signalling proteins, such as Protein Kinase A (PKA) and Protein Kinase CβII, to the centrosome5,18
As all mutations identified were homozygous and significantly truncated the protein, this would suggest that they are functionally ‘null’ alleles. To test this prediction, we examined pericentrin localisation in patient lymphoblastoid cell lines (LCLs). Centrosomal pericentrin staining was lost in three independent LCLs, in contrast to control cell lines ( and Supplementary Fig. 3
), both in interphase and mitosis.
Pericentrin localisation and function is disrupted in PCNT-Seckel cell lines
We also examined protein expression by immunoblotting (). Two bands were detected in controls (Wild-type (WT), and heterozygous relatives, PCNTE220X/+, PCNTS629fs/+), representing two isoforms of pericentrin, which were not present in mutant LCLs. A smaller ~170kD band was present in PCNTE220X lymphoblastoid cells, (potentially representing an aberrant truncated PCNT protein product), while no significant PCNT protein was apparent in the other two mutant cell lines, therefore supporting the molecular genetic findings that protein function is significantly, if not completely, disrupted.
We next examined other components of the centrosome, to establish if they were affected by the absence of pericentrin. ASPM, PCM-1, ninein and centrin were all localised normally ( and Supplementary Fig. 4
). However, during mitosis, γ-tubulin was significantly reduced or absent at prometaphase/metaphase (), consistent with previous RNAi experiments6
Other patients with Seckel syndrome are defective in the ATR DNA damage response pathway3
. We therefore assessed PCNT-Seckel patient cell lines to see if loss of this key centrosomal protein is associated with the defects in ATR signalling previously described for this disorder3,4
. Indeed, PCNT-Seckel (PCNTC1190fs
) LCLs exhibited defective UV-induced G2-M checkpoint arrest, similar to that seen in ATR-Seckel patients (), in contrast to cell lines established from heterozygote relatives (PCNTE220X/+
) and an unrelated wild type control. This suggested that mutations in PCNT impact on ATR-dependent cell cycle checkpoint activation. Similarly, PCNT-Seckel LCLs also exhibited elevated nuclear fragmentation following replication fork stalling () and supernumerary mitotic ‘centrosomes’ after prolonged mitotic arrest with nocodazole (), two additional cellular phenotypes caused by impaired ATR-signalling3
PCNT is required for ATR dependent DNA damage signalling
Since, these findings suggested a direct role for PCNT in ATR-dependent DNA damage response signalling, we examined further steps in this signalling cascade in PCNT-Seckel cells. We found that ATR-dependent γH2AX formation was normal in Seckel-PCNT LCLs, similar to controls (). This was distinct from ATR-Seckel LCLs and suggested that ATR is able to phosphorylate a key substrate in PCNT-Seckel syndrome. Thus, Pericentrin functions downstream of ATR kinase activation in this pathway.
We next examined hydroxyurea-induced 53BP1 foci formation in PCNT-Seckel cell lines. Strikingly, both PCNT-Seckel and ATR-Seckel LCLs failed to form such foci in contrast to control cell lines (). This phenotype is dependent on both ATR and Chk1 kinases19
, indicating that PCNT has a downstream role in ATR-pathway function.
To determine whether PCNT mutations affect other DNA damage responses, we assessed the G2-M checkpoint response to DNA double strand breaks induced by ionising radiation. This response is specifically dependent on ATM and not ATR signalling. PCNT-Seckel LCLs arrested as efficiently as control LCLs under these conditions. ATR-Seckel, unlike ATM mutant LCLs, also had a normal response, confirming the ATR-independence of this assay (). Therefore, we concluded that mutations in PCNT have specific effects on the downstream stage of the ATR-pathway including G2-M checkpoint activation.
We also performed siRNA experiments in HeLa cells to provide confirmatory evidence that PCNT is required for ATR-dependent signalling. siRNA depletion of PCNT also results in defective ATR-dependent (UV) checkpoint activation (Supplementary Fig. 5
), as we had observed in PCNT Seckel LCLs. Similarly, no defect was seen in ATM-dependant checkpoint response after ionizing radiation. This substantiates our findings that PCNT mutations cause imparied ATR signalling.
PCNT is a key structural centrosomal protein nucleating microtubular assembly during mitosis15,16
. It was thus surprising to identify homozygous truncating mutations that appear to cause absence of known forms of the PCNT protein from centrosomes. So remarkably, PCNT would seem to be dispensable for most aspects of human development. Pericentrin mutations do however dramatically reduce both body and brain size by ~ 8 standard deviations below the norm. Other centrosomal genes (ASPM
) have been implicated in determining human brain size, being mutated in primary microcephaly. They are presumed to act through effects on neurogenic mitosis8,9,20
, perhaps modulating symmetric/asymmetric cell division. Given PCNT's role in mitotic spindle assembly15,16
, perturbation of cell division may similarly result in globally reduced cell number, so additionally causing reduced body size. However for ATR
and now PCNT
genes, there is a specific correlation between defects in ATR signalling and reduced brain and body size4,7
. This suggests an alternative explanation, that impaired ATR signalling might itself, directly cause these phenotypes through impaired mitotic progression or reduced cell survival.
Primary microcephaly genes have undergone significant adaptive evolution in primates, suggesting that functional alterations in the proteins they encode may have contributed to the evolution of human brain size21
. Morphologically, Seckel syndrome also has evolutionary parallels. Homo Floresiensis,
the recently described Indonesian hominid, has similarly marked reduced brain and body size (cranial volume -6.1 s.d, height -9.5 s.d, relative to modern humans)22
. Consequently, changes in this or similar genes could underlie the anatomical differences seen in this hominid23
Through study of a human disease phenotype, we have for the first time implicated a structural protein of the centrosome in the ATR-dependent DNA damage response. Centrosomes have a central role in regulating cell cycle progression24
and pericentrin has an established role in recruiting regulatory proteins to the centrosome5,18
. Therefore a role in checkpoint signalling would be consistent with its known functions. Components of the ATR-pathway have been observed to localise to the centrosome25
, most notably Chk126,27
. Chk1 centrosomal localisation, mediated by activating phosphorylation at Serine317 and 345 residues28
, is enhanced by DNA damage. At the centrosome, Chk1 is thought to delay G2/M transition through inhibiting activation of Cyclin B-Cdk1 by Cdc25 27
. Intriguingly, fusion of Chk1 with a PACT domain prevents G2/M mitotic progression27
, suggesting that endogenous pericentrin, one of only two PACT domain containing proteins, could provide the means of Chk1 localisation to mediate G2/M checkpoint arrest (model, ). Alternatively, PCNT may not be required for localisation of Chk1, but may be necessary for Chk1 to transmit its signal to downstream centrosomally localized components to effect G2/M arrest. Future investigations based on these possibilities will be valuable in understanding the mechanism of Chk1 relocalisation and activation at the centrosome.
Model of Pericentrin's role in ATR-dependent G2/M checkpoint arrest
The identification of Pericentrin mutations in Seckel syndrome also provides an interesting convergence between microcephaly genes implicated in ATR signalling (MCPH1, ATR) and those involved in centrosomal function (ASPM, CDK5RAP2, CENPJ). This suggests that further biochemical and developmental investigation is likely to be fruitful in delineating common cellular pathways responsible for the brain and body size phenotypes in these syndromes.