Mitotic Kinesin-Like Protein 1 (MKLP1/KIF23) in hereditary congenital dyserythropoietic anemia type III and in cancer

Abstract: A hereditary form of autosomal dominant congenital dyserythropoietic anaemiatype III (CDA III) has been reported in four families from Sweden, Argentina, Cuba and USA. CDA III patients might experience signs of mild anaemia and some of them need occasional blood transfusions. Other clinical features seen in CDA III patients are retinal angioid streaks, monoclonal gammopathy of undetermined significance and multiple myeloma. Their bone marrow is characterised by presence of giant erythroblasts with up to 12 nuclei. Previously, CDA III was mapped to a region on chromosome 15q21-q25.In this study we aimed to identify the genetic cause of CDA III, investigate the reasons why erythroid lineage in the patients’ bone marrow is mostly affected, and seek the explanation of increased rate of cancer in the Swedish CDA III family.We identified the genetic cause of CDA III using targeted next generation sequencing. A novel missense mutation c.2747C>G, p.P916R in kinesin familymember 23 gene (KIF23) segregated with the disease in both the American and Swedish family, and was absent in databases of sequence variants from healthy individuals. Knock-down and rescue experiments in HeLa Kyoto cells showed that the P916R mutation caused cytokinesis failure which resulted in large cells with several nuclei. This was consistent with the CDA III phenotype.To reveal interaction partners of wild-type and mutant KIF23 proteins, pull-down experiments followed by mass spectrometry and Western blot analysis were performed. This identified Coatomer Protein Complex I (COPI), a vesicle forming complex responsible for intracellular transport, as a KIF23 interactor. By using immunofluorescence and fluorescence microscopy, we showed that COPI subunits COPα and COPβ localize to the midbody during cytokinesis. These findings indicate involvement of vesicle transport proteins in mitosis and cytokinesis, though the significance of COPI-KIF23 interaction in cell division remains to be uncovered.To address the question if other cells are affected by the KIF23 P916R mutation, we created a knock-in mouse model with Kif23 c.2726C>G, p.P909R, which corresponds to the human KIF23 c.2747C>G. However, the mice did not developany phenotype indicating CDA III. This result was consistent with the studies ofother CDA subtypes where mouse models failed, suggesting that CDA occur only in humans. Our study of human and mouse KIF23/Kif23 expression revealed novel, previously not annotated transcripts, one in human and two in mice. Expression analysis of total mRNA using droplet digital PCR demonstrated an extensive variation of KIF23 and Kif23 expression levels in all tissues. The shortest Kif23 transcript lacking exon 17 and 18 was prevalent in mice, while corresponding transcript in human was the least expressed.Considering the importance of KIF23 in cytokinesis and KIF23 association with cancer, we hypothesized that somatic KIF23 mutations might be overrepresented in cancer. For this purpose, we screened KIF23 and its promoter in non-small cell lung cancer samples that previously demonstrated KIF23 overexpression. No pathogenic driving KIF23 variants were detected by Sangersequencing; however, subsequent genome-wide genotyping (SNP-array) detected gain of chromosome 15 in most cases that could possibly explain KIF23 overexpression.