We investigated the genetic spectrum and clinical phenotype of CCNF variants through our large Chinese ALS cohort and literature review. First, the PP variant frequency of CCNF in our cohort (1.1%) was comparable to that in previous studies (0.8%). Second, the 11 novel PP missense variants identified in our cohort expand the genetic variant spectrum of CCNF. Third, our study found that nearly one-third of CCNF variant carriers had additional variants in other ALS-related genes in addition to CCNF variants. Finally, more than half of CCNF carriers in our cohort had cognitive impairment, but no patients met the diagnosis criteria of FTD.
The CCNF gene encodes cyclin F, which is a founding member of the F-box protein family that mainly includes three functional modules: a pseudocatalytic module containing the F-box domain and NLS1, a substrate complement module containing two cyclin domains, and a regulatory module containing NLS2 and PEST (a short amino acid rich in proline, glutamic acid, serine and threonine) [24]. Cyclin F is essential for genome stability and regulates deoxyribonucleotide triphosphate levels, centrosome duplication and spindle formation by targeting and regulating the levels of SCFcyclin F substrates [25, 26]. Recently, it was also found that the expression of CCNF in the motor neuron-like cell line NSC-34 changes after exposure to oxidative stress [27]. CCNF can also affect the activity of valosin-containing protein (VCP) in the cytoplasm [28].
In our cohort, a total of 41 ALS patients were carriers of 29 variants in the CCNF gene, but only 15 missense variants were identified as deleterious by software analyses. Therefore, the variant frequency of CCNF in our ALS cohort was 1.1% (15/1587), which was similar to the results of previous studies [5, 17,18,19,20]. A previous study found an enrichment for rare protein alterations in CCNF (MAF < 0.0001) among SALS patients compared to control individuals [5]. However, this study also identified a CCNF frameshift variant (p.L372fs) that did not segregate with disease, suggesting that a dominant gain in toxic function may be required for the pathogenicity of mutant CCNF [5]. One patient with the frameshift variant (p.Leu553fs) was also found in our cohort. However, because of the lack of genetic information on relatives, it was not possible to determine whether the variant segregated with the disease. Therefore, whether the functional consequences of CCNF variants lead to increased functional toxicity or loss of dominant negative function or haploinsufficiency remains to be determined by more studies.
Currently, studies have not found a clear relationship between the location of CCNF variants and the clinical phenotype. Two variants (p.Pro27Ser and p.Arg21Gln) were located in the NLS1 domain, which mediates interactions with other components of the SCF ubiquitin-protein ligase complex to achieve ubiquitination of target substrates [24]. This is the first report of a variant in the NLS1 domain. Both variants were considered deleterious by software (predictions of 4/8 and 5/8, respectively). In our cohort, there was also a newly discovered variant (p.Arg568Trp) located in the NLS2 domain, which was considered to be harmful with a prediction of 4/8. The patient carrying this variant had extremely rapid disease progression, with a progression rate of 1.28, and died of sputum asphyxia less than 15 months after onset. The p.Arg568Trp variant has not been reported previously. However, two variants (p.R574Q and p.R572W) located in the NLS2 domain have been reported in European populations and are also considered to be pathogenic variants [5, 20].
Ten variants were located in the cyclin domains in our cohort, eight of which (all except the p.Pro487Ser and p.Arg344Lys variants) were predicted to be PP (prediction ≥ 4/8). Interestingly, the benign CCNF p.Pro487Ser variant (prediction 2/8) was present in six patients in our ALS cohort, and this variant was previously reported in an Asian patient [17]. In our cohort, a benign p. Arg344Lys variant (prediction 2/8) identified in two patients has been previously reported once in both European and Asian populations [5, 19].
Seven variants located in the PEST sequence were all predicted to be benign by software analyses (prediction<4/8) in our study, and none have been reported previously. However, the p.S621G variant located in the PEST sequence has been reported repeatedly in Caucasian FALS/FTD patients [5, 22]. The p.S621G variant prevents the phosphorylation of casein kinase II (CK2) and increases the lys48-ubiquitin activity, resulting in dysfunction of the autophagy degradation pathway [5, 29, 30]. It was also found that zebrafish with the p.S621G variant had disrupted axonal growth, suggesting a toxic gain-of-function mechanism in the pathogenesis of ALS [31]. The p.S621G variant was not found in our cohort, and all variants identified in our cohort were not located at the identified phosphorylation sites [29]. Therefore, based on the results predicted by software and the current literature review, we considered these seven variants located in the PEST sequence in our cohort to be likely benign, but more basic experimental studies are needed to explore the pathogenicity and pathogenesis of CCNF variants located in the PEST sequence.
Four CCNF variants (p.Gly161Arg, p.Ser222Pro, p.Ala131Ser, and p.Arg123Cys) predicted to be PP in our study were not located in any functional domain. Among them, the p.Gly161Arg variant of CCNF has been reported previously in both Caucasian and Asian populations. In the Asian populations, the p.Gly161Arg variant was found in a 43-year-old man who presented with rapidly progressive left-hand weakness and died of respiratory failure two years after onset [5, 20]. In our cohort, the patient carrying the p.Gly161Arg variant also had a rapid progression rate of 1.5, but the survival time of this patient could not be obtained due to loss of follow-up. A previously reported Asian carrier of the p.Ser222Pro variant of CCNF had a relatively long survival time of more than 60 months [17]. The in vitro functional study found that the p.Ser222Pro variant of CCNF had a deleterious effect on cyclin F-mediated proteasome degradation, and UPS damage may occur upstream of the proteasome through abnormal ubiquitination or transport to the proteasome [5, 17]. The p.Ala131Ser and p.Arg123Cys variants of CCNF have not been reported previously. Although they are not located in any functional domain, they may also be related to the folding and aggregation of abnormal proteins. Further basic studies are needed to explore the structural or functional changes of these variants.
Currently, this study is the largest sample size of CCNF variant carriers and clinical characteristics analysis. The frequency of PP variants in CCNF in our study was similar to that in previous studies, but there were differences in the results of sex ratio, site of onset, survival, and cognitive function. The CCNF carriers in our ALS cohort had a late age of onset (59.67 years vs. 52.62 years) and a large proportion of bulbar onset (27.8 vs. 10.5%), especially among Asian patients, where bulbar onset has not been reported before. In addition, patients with PP variants of CCNF in our study had a shorter median survival than previously reported results (24.0 months vs. 43.0 months). The proportion of male patients was similar between our cohort and Caucasian patients. In addition to the ALS phenotype, a previous study on FALS and FTD patient cohorts found CCNF carriers with primary lateral sclerosis [5], which was not found in our cohort. This may be because the current cohort only enrolled ALS patients. No significant differences in clinical manifestations were found in patients carrying CCNF variants in different domains in our cohort, but CCNF variant carriers with variants located in cyclin domains tended to have a faster disease progression rate than carriers with variants located in PEST domains (0.92 vs. 0.49). However, the difference was not statistically significant. This was consistent with our finding that all variants located in the PEST sequence were likely benign, while the majority of variants located in the cyclin domains were probably pathogenic.
Our study has the following limitations. First, we did not have genetic information from the relatives of carriers of CCNF variants, so we cannot estimate whether these variants segregate with disease. Second, there are few studies on CCNF variants in ALS, and the correlation between CCNF carriers and clinical phenotypes cannot be clarified. Third, we assessed the pathogenicity of missense variants only by in silico tools, without further basic experiments, which could lead to misjudgments of pathogenicity.