Ultrastructural Characterization of Primary Cilia in Pathologically Characterized Human Glioblastoma Multiforme (GBM) Tumors

Joanna J Moser; Marvin J Fritzler; Jerome B Rattner


BMC Clin Pathol. 2014;14(40) 

In This Article


The expression of a primary cilium relies on two main events: 1) activation of ciliogenesis and 2) orderly progression through a series of developmental stages so that a structurally and functionally competent mature cilium is formed.[27–29] Our study illustrates that ciliogenesis was activated in all the GBM samples examined but cilium morphogenesis beyond stage 1 was rare in the majority of tumors. These findings support our previous examination of several astrocytoma/glioblastoma cell lines.[17] Thus, cells from each of these sources (cell line or tumor tissue) express a similar defect or set of defects that targets the earliest stages of ciliogenesis and does not inhibit the cells ability to proliferate.

These findings are compatible with previous studies of melanoma, renal cell carcinoma and pancreatic cancer, which found that primary cilia loss was independent of Ki67 staining (cell proliferation marker) suggesting that cilia loss is not the result of altered cellular proliferation rates but rather may be due to aberrations in another mechanism that is inherent to ciliogenesis.[30–32] Yang and colleagues (2013) recently showed that cell cycle-related kinase (CCRK) and its substrate intestinal cell kinase inhibited ciliogenesis in a glioblastoma cell line.[33] Specifically, they showed that dysregulated high levels of CCRK are present in U-251 MG glioblastoma cells whereby knockdown of CCRK led to the formation of primary cilia indicating that CCRK depletion restored primary ciliogenesis.[33] Furthermore, it was demonstrated that the inhibition of ciliogenesis by over-expression of CCRK in U-251 MG glioblastoma cells promoted cell proliferation capacity.[33]

From our ultrastructural studies in astrocytoma/glioblastoma cell lines and GBM tumor tissues it is interesting to note that profiles occasionally displayed centriole/basal bodies with structural abnormalities (i.e. altered length or microtubule integrity). This suggests that it is possible for such structural alterations to be tolerated by the cycling cell, perhaps by being repaired, or that these defects underlie further aberrant cancer cell behaviour.

It is important to note that in the majority of previously published studies, IIF alone was used to evaluate ciliogenesis status. This technique alone does not allow for the precise identification or characterization of the earliest stages of ciliogenesis. Thus, truncated cilia such as that seen in a few patients within our study may be more common that previously indicated. Our ultrastructural data not only reveals a defect in early ciliogenesis but also shows that this defect specifically affects the initial elaboration of the distal surface of the basal body and its ability to associate with Golgi derived vesicles. There have been a number of proteins shown to act at the distal end of the basal body, particularly at the distal appendage region, and they include; Cep170,[34] ninein,[8,35] ε-tubulin,[36] cenexin/ODF2[37] (likely cenexin1[38]) and by association Rab8a,[39] centriolin/Cep110,[8,40] Cep164[41] and Cep123[42] reviewed in.[1] For example, in neuronal primary cilia, B9-C2 containing proteins have been shown to collect at the base of the primary cilium in the transition zone[43–47] and physically interact and with ciliary protein localization[48] (and reviewed in[49]). One B9-C2 family gene in particular named Stumpy (or B9d2) is required for mammalian ciliogenesis where knockout mutants displayed near-complete loss of neuronal primary cilia with remaining cilia displaying dysmorphic stump-like ultrastructures.[43]

Of particular interest, a distal appendage protein, Cep 123, has recently been shown to be required for initiation of ciliogenesis by modulation of capping the distal end of the mother centriole with a ciliary vesicle.[42] Sillibourne and colleagues (2013) showed that Cep123 is required for assembly of a primary cilium but not the maintenance of the axoneme in human retinal pigment epithelial (RPE1) cells.[42] Depletion of Cep123 using Cep 123 siRNA perturbed ciliary vesicle formation at the distal end of the basal body which suggests that distal appendage proteins are critical for progression of cilia beyond the early stages of ciliogenesis.[42] These knockdown studies are captured in Figure 6B by Sillibourne et al. (2013)[42] and are very similar in appearance to the abnormal early stages of ciliogenesis seen in our GBM tumors. Given this high degree of ultrastructural similarity, Cep123 may be the best candidate to explain the defects we observed in GBM tumors. Although a review of the glioblastoma literature does not highlight Cep123 as being defective in patients with GBM tumors, our study suggests that it is a reasonable target for future expression studies and ultrastructural analysis in GBM tumors.

Limitations of the current study are small sample size and lack of normal brain tissue for comparison. Given the small incidence of malignant gliomas per year, we collected samples over a 5 year period and eliminated those samples that were not grade IV glioblastomas/GBM. To respect the ethics of collection of normal human brain tissue from our patients, we compared the GBM patient results to those previously established in normal human astrocyte cells that were used between passages 3–5 in culture.[17] Although 10 grids were examined for each patient sample, there were noticeable differences between tumor samples in terms of cellularity and patient heterogeneity. It is important to emphasize that this is a complex mixture of cells and extracellular matrixof GBM brain tissue and that some cell types are ciliated whereas others are not ciliated. Because of the tissue complexity and heterogeneity, it is impossible to identify all the cells which have the potential to undergo ciliogenesis. When we do see profiles, we can quantitate the number of cells undergoing abnormal ciliogenesis (including profiles at stages 3/4/5) which is summarized as follows. Patient #1: 20 cells with profiles, 2 cells with normal cilia, 0 cells with abnormal cilia. Patient #2: 9 cells with profiles, 0 cells with normal cilia, 2 cells with abnormal cilia. Patient #3: 20 cells with profiles, 0 cells with normal cilia, 2 cells with abnormal cilia. Patient #4: 22 cells with profiles, 0 cells with normal cilia, 0 cells with abnormal cilia. Patient #5: 20 cells with profiles, 0 cells with normal cilia, 0 cells with abnormal cilia. Patient #6: 30 cells with profiles, 0 cells with normal cilia, 3 cells with abnormal cilia. Patient #7: 15 cells with profiles, 0 cells with normal cilia, 0 cells with abnormal cilia. Taken together, only one patient expressed morphologically normal cilia. As a whole, we did see the same types of abnormalities in the early stages of ciliogenesis amongst tumor samples which suggests that early ciliogenesis defects are a generalized problem in GBM tumors. In summary, we can say that the large majority of grade 4 glioblastoma/astrocytomas (i.e. GBM tumors) are likely to express abnormal immature primary cilia suggesting that this defect may be a hallmark of GBMs. We have not detected a clear correlation between abnormal ciliogenesis and the 3 main molecular characterizations examined in these patient samples. It must be kept in mind that our patient sample size may not be sufficient to reveal correlations with molecular markers and this might require a larger study.

In summary, we found that ciliogenesis is activated in GBM tumors but the normal development of a mature cilium is perturbed at early stages of ciliogenesis. The aberrant ultrastructural profiles observed in our survey of GBM tumors and a review of the current ultrastructural profiles present in the literature suggest the possibility that at present the best possible candidate protein underlying defects in the early stages of ciliogenesis within GBM tumors might involve Cep123.