Primary cilia presence and implications in bladder cancer progression and invasiveness

Urothelial bladder cancer is the tenth most common cancer worldwide. It is divided into muscle and non-muscle invading bladder cancer. Primary cilia have been related to several cancer hallmarks such as proliferation, epithelial-to-mesenchymal transition (EMT) or tumoral progression mainly through signaling pathways as Hedgehog (Hh). In the present study, we used immunohistochemical and ultrastructural techniques in human tissues of healthy bladder, non-muscle-invasive bladder cancer (NMIBC) and muscle-invasive bladder cancer (MIBC) to study and clarify the activation of epithelial-to-mesenchymal transition and Hedgehog signaling pathway and the presence of primary cilia. Thus, we found a clear correlation between EMT and Hedgehog activation and bladder cancer stage and progression. Moreover, we identified the presence of primary cilia in these tissues. Interestingly, we found that in NMIBC, some ciliated cells cross the basement membrane and localized in lamina propria, near blood vessels. These results show a correlation between EMT beginning from urothelial basal cells and primary cilia assembly and suggest a potential implication of this structure in tumoral migration and invasiveness (likely in a Hh-dependent way). Hence, primary cilia may play a fundamental role in urothelial bladder cancer progression and suppose a potential therapeutic target.


Introduction
Bladder cancer is the tenth most common cancer worldwide (Global Cancer Observatoty 2018), being three times more frequent in men than women (Antoni et al. 2017). In the Western countries, the most common histological subtype is urothelial cancer, reaching 90% of cases (Bryan 2015).
Carcinomas progression and invasiveness are closely related to epithelial-to-mesenchymal transition (EMT) (Prieto-García et al. 2017). EMT consists of the dynamic and reversible switch from an epithelial non-motile to a mesenchymal motile phenotype (Zeisberg and Neilson 2009;Lamouille et al. 2014;Prieto-García et al. 2017). During EMT, epithelial tumor cells undergo transient changes resulting in a mesenchymal phenotype characterized by a loss of cellular junctions, changes in cell-matrix adhesion, loss of apicobasal polarity, cytoskeletal changes and an increased 1 3 contractility and motility (Lamouille et al. 2014). These features allow tumor cells to release from neighbor cells and migrate to distant areas. Thus, EMT is a key process in tumor progression, invasiveness and metastasis (Zeisberg and Neilson 2009;Prieto-García et al. 2017). EMT is induced by several transcription factors (remarkably Snail1, Snail2 or Slug, Twist, ZEB1 and ZEB2) (Lamouille et al. 2014). These transcription factors repress the expression of epithelial-related proteins such as cytokeratin and E-cadherin (which mediates cell to cell junctions) (Lamouille et al. 2014). Conversely, they induce the transcription of genes such as Vimentin and N-cadherin, which confer a mesenchymal phenotype (Lamouille et al. 2014). Regarding bladder cancer, Twist is the best-studied EMT-related transcription factor and its expression has been related to smoking, poorer prognosis and increase metastasis (Fondrevelle et al. 2009;Yun and Kim 2013). EMT can be triggered by different signalling pathways such as TGFβ, Notch, Wnt or Hedgehog (Hh) (Lamouille et al. 2014). Most of these pathways have been related to primary cilia (Wheway et al. 2018). Primary cilium is an organelle composed of a basal body (originated from a modified centriole) which gives rise to an axoneme protruding outside the cell. It has been related to different cellular functions including signaling pathway activation, cytoskeletal functions, cell cycle control or protein homeostasis (Malicki and Johnson 2017). Moreover, this structure regulates several cancer hallmarks (Fabbri et al. 2019). However, the presence and importance of primary cilia in bladder cancer is still not known and it has just started to be explored (Du et al. 2018).
Hh is the best-studied cilia-dependent pathway and it has been closely related to several types of cancer (Rohatgi et al. 2007;Amakye et al. 2013;Castiella et al. 2013;Chahal et al. 2018;Iruzubieta et al. 2019). Hh proper activation depends on primary cilia (Rohatgi et al. 2007). Thus, when Hh ligands bind Patched receptor, located in the ciliary membrane, it is removed from the cilia allowing Smoothened (Smo) entrance to the membrane and Glioma associated oncogene 1 (Gli1) activation and translocation into cell nuclei (Rohatgi et al. 2007;Ramsbottom et al. 2016). Gli1 activates target genes transcription, including genes related to EMT, angiogenesis and cancer stem cell maintaining (Katoh and Katoh 2009;Amakye et al. 2013;Chahal et al. 2018).
The exact relationship between Hh pathway and bladder cancer remains elusive. Thus, Hh has been proposed both as an inductor of EMT and invasiveness (Islam et al. 2016;Nedjadi et al. 2019) and as a protector element that restrains bladder cancer progression (Shin et al. 2014b).
In the present study, we study through immunohistochemical techniques EMT activation, Hh signalling pathway and primary cilia presence in bladder cancer. Besides, we performed a detailed ultrastructural study of tumoral cell features and EMT process. Our results emphasize the importance of primary cilia in cancer metastasis and recurrence, proposing the relevance of primary cilia as a potential target in bladder cancer.

Materials
Six MIBC (all G3 except for one G2-G3, five pT2NxM0 and one pT3bN0M0), seven NMIBC (five G2 and two G1, all of them pTaNxM0) and two healthy bladder controls were used for the study. They were retrieved from the files of the Pathology Department at the University Clinic Hospital of Zaragoza. All protocols developed were approved by the Human Research Ethics Committee "Comité Ético de Investigación Clínica de Aragón".
Semi-thin sections (1.5 μm) were cut with a diamond knife, lightly stained with 1% toluidine blue and examined by light microscopy (Olympus BX51 microscope, Olympus Imaging Corporation; Tokyo, Japan). Later, ultrathin (0.05 μm) sections were cut, collected on Formvar coated single-slot grids counterstained with 1% uranyl acetate and Reynold's lead citrate for 1 min.
They were examined under a JEOL 1010 TEM. The images were captured with Gatan Bioscan Camera with pixel size 6.45 µm × 6.45 µm and spatial resolution of 1024,1024, 1, 1 (x, y, z, t) and automatic gain.

Immunohistochemistry
Immunohistochemical staining was performed on 4-μm-thick formalin-fixed paraffin-embedded sections using EnVision® (DAKO) method. The primary antibodies used in this study were mouse anti-Cytokeratin Clone MNF116 The tissue sections were deparaffinised in xylene for 20 min and rehydrated in graded ethanol (see above). Prior to all assays, a heat-induced antigen retrieval step was performed for E-Cadherin, Gli1 and Smo immunolabeling using buffer citrate (pH6, DAKO S2031) by treatment for 5 min in microwave at 800 W and 3 additional minutes at 360 W. For cytokeratin and patched staining, an antigen retrieval step using TBS (pH9) was performed at 96 °C for 20 min. Twist unmasking was performed by heating in Tris-EDTA buffer (pH9) at 98 °C for 40 min.
After washing twice with PBS for 3 min, endogenous peroxidase was blocked using peroxidase blocking reagent (DAKO, S2001) for 10 min and sections were washed in distilled water and PBS 3 min, twice.
Afterwards, sections were incubated with primary antibodies at 4 °C overnight in a humidified chamber. Sections were washed in PBS three times for 5 min and incubated with Labelled Polymer EnVision-HRP (DAKO K5007) or Rabbit Anti-Goat Immunoglobulins/HRP (1:200, DAKO, P016002-2) for 1 h. To confirm the presence of immunocomplexes, 3,3′ diaminobenzidine was used as the chromogen. The samples were contrasted with Mayer's haematoxylin 7 min, washed in tap water for 15 min, dehydrated in graded ethanol (2 min in 70°, 2 min in 96° and 5 min in 100°), cleared in xylene and coverslipped with Eukitt (03,989 Sigma-Aldrich; St. Louis, MO, USA). Negative controls were based on samples to which no primary antibody was added. All antibodies were commercial and supposedly validated.
Whole samples were analysed and digital images from representative areas were captured by Olympus BX53 and Olympus UIS UPLFLN Series objectives. FN: 26.5. 20 × AN 0.5 and 40 × AN 0.75. Images were captured with Olympus DP72 Digital Camera. Pixel size 6.45 µm × 6.45 µm, total number of pixel 1.5 million. Exposition mode: SFL Auto and spatial resolution of 4140 × 3096 × 1 × 1 (x, y, z, t).

Immunofluorescence microscopy
Four μm formalin-fixed paraffin-embedded sections were cut and mounted on glass slides. The sections were deparaffinised in xylene for 20 min and rehydrated in graded ethanol (4 min in 100°, 4 min in 96°, 4 min in 70° and 5 min in distilled water). Tissues were permeabilised with 0.1% Triton X-100 in PBS for 8 min. Afterwards, an antigen retrieval step using Tris-buffered saline (TBS, pH9) was performed at 96 °C for 20 min.
Sections were incubated overnight at 4 °C with the following primary antibodies: mouse anti-Acetylated tubulin (1:4000, Sigma-Aldrich, T7451; St. Louis, MO, USA) and rabbit anti-Pericentrin (1:100, Abcam, ab4448; Cambridge, UK) in a dark humidified chamber. Sections were washed in PBS three times for 5 min and incubated for 1 h at RT in a dark humidified chamber with secondary antibodies: donkey anti-mouse Alexa Fluor 594 (1:1000, ThermoFisher, R37115; Waltham, MA, USA), donkey anti-rabbit Alexa Fluor 488 (1:1000, ThermoFisher, A-21206; Waltham, MA, USA). After washing in PBS, DAPI (1 μg/mL, Sigma-Aldrich) was added for 1 min for nuclei counterstaining. Sections were washed in PBS and the slides were covered with fluorescence mounting medium (DAKO, S3023). Negative controls were based on samples to which no primary antibody was added. On the other hand, we used a ciliated tissue (Fallopian tube) as a positive control for cilia immunofluorescence. All antibodies were commercial and supposedly validated.
Images were analysed with DP Controller Software. Every fluorescent channel was individually photographed using automatic autoexposure and channels were merged using FIJI Image_J software, where contrast and brightness were adjusted (Schindelin et al. 2012).
Statistical analyses were performed counting ciliated cells per High Power Field (HPF = × 400). The areas analysed were representative of the samples. A non-parametric test (Kruskal-Wallis test) was used.

Histological study
Haematoxylin-eosin staining was performed in all samples ( Fig. 1). Healthy bladder controls showed four to eight cellular layers with the characteristic umbrella cells on top (Fig. 1a). NMIBC presented abundant papillae composed of a thin vascularized stromal core surrounded by an increased number of cellular layers (Fig. 1b). On the other hand, MIBC samples show abundant heterogenous cells with atypia features such as enormous nuclei with prominent nucleoli, pale vacuolized cytoplasm, numerous mitosis and muscle invasion by tumoral cell cords (Fig. 1c). In some cases, necrotic areas were found, indicative of uncontrolled proliferation (Fig. 1d).

Epithelial-to-mesenchymal transition (EMT) in bladder cancer progression
As EMT is a predominant mechanism in tumor progression and invasiveness, we analyzed the expression of epithelial markers E-cadherin and Cytokeratin and the EMT-transcription factor Twist in bladder cancer stages regarding normal tissue (Fig. 2).
Our results showed that control samples showed prominent staining for cytokeratin ( Fig. 2a) that was progressively reduced in NMIBC ( Fig. 2b) and insignificant in MIBC (Fig. 2c). Similar results were found regarding E-cadherin expression ( Fig. 2d-f). Although it was not consistently presented in the whole sample, there were specific areas in NMIBC tissues where E-cadherin and Cytokeratin expression was clearly reduced in basal cell layers and progressively increased in distal layers ( Fig. 2b and e).
Accordingly, Twist nuclear staining was inexistent in the healthy bladder (Fig. 2g) and progressively increased and more intense in NMIBC (Fig. 2h) and MIBC (Fig. 2i). Twist expression was limited to tumoral cells while surrounding tissue showed no staining. The nuclear presence of Twist in these samples points to the activation of EMT program in bladder cancer cells, especially in MIBC. Once again, in NMIBC we found subtly predominant staining in basal layers (Fig. 2h).
Hence, our experiments confirmed that epithelial markers are reduced in bladder cancer while nuclear translocation of the EMT transcription factor Twist is increased. These findings confirmed that EMT activation correlates with tumoral grade. Moreover, we found that NMIBC samples show EMT indicators predominantly in basal layers.

Hh signalling pathway in bladder cancer progression
Afterwards, we studied the presence of different Hh signalling pathway components in different tumor stages to clarify the role of this pathway in bladder cancer.
Thus, we performed immunohistochemical experiments for Patched receptor, Smo and the main pathway effector Gli1 (Fig. 3). We found that Hh pathway components were practically no present in healthy controls while they show intermediate staining in NMIBC and they were prominently expressed in MIBC.
Smo expression was absent in healthy controls (Fig. 3a) and subtle expression was found in NMIBC (Fig. 3b). However, in MIBC, Smo staining was present in cell membrane and intense cytoplasm accumulations (Fig. 3c). Accordingly, Patched was not expressed in control bladders (Fig. 3d). Conversely, it was found in a localized paranuclear area both in NMIBC (Fig. 3e) and MIBC (Fig. 3f), being especially intense and abundant in the latter.
Gli1 nuclear staining is an indisputable sign of Hh signaling pathway activation. No Gli1 staining was found in healthy controls (Fig. 3g), but light presence appeared in several NMIBC cells (Fig. 3h). Some isolated positive cells were found beyond the basement membrane. Gli1 nuclear expression was intensely present in MIBC cells (Fig. 3i).
These results show that Hh activation is directly related to tumoral grade. This supports the hypothesis of Hh signalling pathway role in bladder cancer progression and invasiveness showing, consequently, an intense expression in high-grade tumours.

Primary cilia in bladder cancer
Canonical Hh signalling pathway is mediated by primary cilia both in physiology and cancer. Once we found Hh activation in bladder cancer, we studied the presence of primary cilia in these tumours through immunofluorescence techniques. Hence, using specific markers for both primary cilia axoneme (Acetylated Tubulin) and centrioles (Pericentrin), we analyzed the presence, number and proportion of primary cilia in healthy bladders, NMIBC and MIBC (Fig. 4).
Primary cilia were found in controls (Fig. 4a-c), NMIBC ( Fig. 4d-f) and MIBC (Fig. 4g-i). Interestingly, both healthy controls and NMIBC, showed ciliated cells located fundamentally (though not uniquely) in basal layers ( Fig. 4a and  d). Furthermore, several ciliated cells were found across the basement membrane in NMIBC (Fig. 4d). Quantitative analyses were performed counting ciliated cells per high power field and a non-parametric test (Kruskal-Wallis) was used (Fig. 4j). The number of ciliated cells showed a clear tendency to progressively increase from controls to MIBC, although no statistical significance was found (p = 0.084). As these differences could be secondary to the increased In non-invasive tumours (NMIBC), some areas with characteristic staining were found. These areas showed a reduced cytokeratin expression in basal layers. c In muscle-invasive bladder cancer (MIBC), however, only a weak expres-sion was found. d-f Similar findings appeared regarding E-cadherin expression, which was mainly located in cell membranes. g Twist nuclear staining was fundamentally negative in control bladders, h whereas it was weakly positive in NMIBC especially in basal layers, although also distally and i clearly positive in numerous tumoral cells of MIBC but no in the surrounding tissue. Scale bar = 100 μm 1 3 number of cells found in cancer, we also studied the proportion of ciliated cells regarding the total number of cells per high power field (Fig. 4k). No apparent nor statistical significant differences were found neither (p = 0.4264).
In conclusion, primary cilia were found in healthy controls, NMIBC and MIBC with a non-statistically significant tendency to be increased in high-grade tumours.

Ultrastructural study
To better characterize the cells and the presence of primary cilia in bladder cancer, ultrastructural studies using transmission electron microscopy (TEM) were performed in tumoral tissues.

NMIBC
In low-grade tumours, hyperplasic epithelial cells were found. They appeared organized in perpendicular layers to the papillae axe. The number of cellular layers was increased comparing healthy controls (Fig. 5a). Two cellular subtypes were clearly identified regarding cytoplasm electron density: clear and dark cells (Fig. 5b).
On the one hand, clear cells were less numerous and they showed less electron dense cytoplasm. These cells were characterized by scarce rough endoplasmic reticula cisterns close to dictyosomes of Golgi apparatus, free polyribosomes and paranuclear location of centrioles. Converse to dark cells, clear cells showed abundant mitochondria with e Subtle expression was present in NMIBC and f intense positivity was found in MIBC, showing paranuclear aggregates. g Regarding Gli1, no nuclear staining was found in controls. h Some subtly positive cells were found in NMIBC, Isolated Gli1 + cells were found beyond the basement membrane. i Intense nuclear expression was identified in MIBC tumoral cells but not in surrounding cells. Scale bar = 100 μm thicker crests (Fig. 5c). Moreover, they showed electron dense nuclei with abundant condensed heterochromatin, a sign of low transcriptional activity.
On the other hand, dark cells were the most frequent cell type in NMIBC. They showed voluminous oval nuclei with very decondensed euchromatin, suggesting active transcription, and a fine band of marginal heterochromatin surrounding the nuclear envelope. Nuclear envelopes characteristically presented deep invaginations (Fig. 5d) which sometimes generates interesting structures called nuclear envelope-limited chromatin sheets (ELCS) (Sup. Fig. 1a). Furthermore, they usually showed a prominent fibrillar nucleolus (Fig. 5d). Dark cells had a medium electron dense cytoplasm which contained similar organelles than clear cells (Fig. 5e). Especially relevant is the presence of multivesicular bodies fulfilled by vesicles of different  cells (b, c). Primary cilia were especially found in basal layers. d-f Primary cilia in NMIBC. Some ciliated cells were found across the basement membrane (dashed line). g-f Primary cilia in MIBC. j Quantitative analyses of total number of ciliated cells per High-Power Field. Apparent but not statistically significant differences could be detected by Kruskal-Wallis test (p = 0.084). k Proportion of ciliated cells regarding the total number of cells. No differences were found (p = 0.4264). Scale bar = 50 μm 1 3 sizes (Fig. 5f). Abundant filaments originate from the characteristic darker cytoplasm.
A detailed ultrastructural study identified EMT features in NMIBC (Fig. 6). Hence, urothelial cells located in basal layers lost their apicobasal polarity (Fig. 6a) due to desmosomes disassembly (Fig. 6b) and formation of filopodialike structures (Fig. 6c), which allow cellular migration. As a consequence, intercellular space was wider. This mesenchymal phenotype acquisition was related to an increased number of mitochondria and centrioles activation.
Thus, in some dark cells, activation of centrioles was found, characterised by the loss of their perpendicular disposition, the presence of ciliary vesicles and migration to the cellular surface (Fig. 7a). There, they bind the cell membrane and formed primary cilia (Fig. 7b, c). This structure was fundamentally found in basal cells (Fig. 7c). Primary cilia subcomponents such as 9 + 0 axoneme, basal body, transition fibres binding the cell membrane and subdistal appendages were identified. Primary cilia base was ensheathed by a membrane invagination called cilliary pocket.
Although diagnosed as NMIBC, some cells were found invading the basement membrane, showing a clear deformation of the cellular shape and a nuclear hourglass Activated centrioles originate primary cilia that emerges to the extracellular space (inset). Characteristic nuclear invaginations are identified in these cells. Scale bar = 2 μm (inset 250 nm) c A tumoral dark cell located on top of basement membrane showing an activated centriole binding cell membrane through transition fibres (tf) (inset). Scale bar = 5 μm (inset 500 nm) d Some tumoral cells crossed the basement membrane (bm) by adapting the cellular shape through an hourglassmorphology nucleus (arrows). Scale bar = 2 μm ax axoneme, cv ciliary vesicle, ct centrioles, cp ciliary pocket, bb basal body, dc daughter centriole morphology (Fig. 7d). Some epithelial dark cells were found in lamina propria, located near blood vessels (Figs. 6a and  8a, c). Interestingly, these cells showed primary cilia projecting to the extracellular space, near Golgi apparatus (Fig. 8b,d). These cells kept a typical disposition of organelles in perinuclear area. However, their cytoplasm seemed extremely vacuolated and loss of cytoplasm fragments was found (Fig. 8).

MIBC
High-grade tumours showed an enormous cellular heterogeneity, although previously described clear and dark cells were still identifiable (Fig. 9a). Abundant mitosis (Fig. 9b) and some apoptotic cells were found (Sup. Fig. 1b). Moreover, tumoral infiltration of muscles, blood vessels and nerves was identified. Clear cells kept their cytological properties, highlighting irregular nuclei and deep invaginations of nuclear envelope (Fig. 9c). These cells established new contacts with clear cells by small punctiform desmosomes and to dark cells through interdigitations generating tumoral cords ( Fig. 9c and d). The so-called dark cells were more numerous than in low-grade tumours and, occasionally, they showed a fusiform-mesenchymal morphology (Sup. Fig. 1c). Some cells showed abundant autophagosomes containing rough endoplasmic reticula (Sup. Fig. 1d). Although the exact nature of these structures needs further research, it may point to autophagy impairment in these cells. Moreover, angiogenesis of new blood vessels was identified (Sup. Fig. 1e) and, as pointed above, necrosis and necroptosis were also found in some tumoral cells (Sup. Fig. 1f).
Activated centrioles and primary cilia were also found in MIBC cells (Fig. 9e, f).
Hence, ultrastructure showed that two types of cells were present in bladder cancer: clear and dark cells. The latter showed primary cilia and appeared in NMIBC crossing basement membrane and next to blood vessels. In MIBC, cells show an increased heterogeneity, abundant mitosis, invasiveness and some cellular alterations. These cells present a progressive vacuolization (v) that finally produce cytoplasm release. Interestingly, these cells showed primary cilia. Scale bar = 5 μm b and d Magnification of squared areas in a and c (respectively) showing a detailed image of primary cilia and their components: basal body (bb), subdistal appendages (sa), axoneme (ax). Near primary cilia, Golgi apparatus is usually present (g). Scale bar = 500 nm dc daughter centriole, n nuclei, tf transition fibres

Discussion
The present study shows new roles and potential implications of primary cilia in bladder cancer, pointing to the potential importance of this structure in tumor progression and invasiveness, likely in relation with epithelial-to-mesenchymal transition (EMT) and the cilia-dependent Hedgehog (Hh) signalling pathway.
EMT is a key element in epithelial tumoral cells, which acquire features related to stemness, invasiveness and metastasis (Wever et al. 2008;Lamouille et al. 2014;Prieto-García et al. 2017). In bladder cancer, the EMT-activator Twist has been related to tumor progression and epithelial markers downregulation (Fondrevelle et al. 2009;Yun and Kim 2013). Accordingly, we found increased Twist expression in MIBC and lower in NMIBC and health controls while the epithelial markers E-cadherin and cytokeratin were downregulated in MIBC. These results confirmed that EMT program is activated in bladder cancer, especially in invasive tumors.
Several studies have demonstrated the importance of Hh signaling pathway in bladder cancer. However, the exact influence of Hh activation remains controversial. Most studies point that ligand-dependent Hh signaling activation results in increased tumorigenicity, stemness, EMT and migration in bladder cancer (Fei et al. 2012;Islam et al. 2016;Kitagawa et al. 2019). Furthermore, inhibiting Hh has shown to reduce these effects in bladder cancer cells (Islam et al. 2016;Miao et al. 2018;Kitagawa et al. 2019) and to decrease cancer stem cell population (Wang et al. 2017). Conversely, Shin et al. have shown that Hh expression by urothelial basal cells plays a protective role in bladder cancer by stimulating urothelium differentiation through a feedback loop with stromal cells (Shin et al. 2014a). Consequently, they showed Hh reduction in a mouse model of bladder cancer and human samples (Shin et al. 2014a, b). Our results are in accordance with former works, showing an enhanced expression of Patched, Smo and Gli1 in MIBC. Besides, we did not find Hh pathway components expression in stromal cells or blood vessels as was described in Shin's work (Shin et al. 2014b).
Nevertheless, our experiments interestingly showed that basal cells in some areas of NMIBC presented Hh pathway activation and a reduction of epithelial markers. Our findings suggest that urothelial basal cells are importantly related to EMT and Hh activation in low-grade tumours and probably related to cancer initiation, recurrence and future progression; an idea also shared by other authors (Shin et al. 2014b).
Primary cilia have been straightly related to cancer although their exact role is not completely understood. Hence, it has been related to tumor progression, drugresistance (Jenks et al. 2018), stemness and EMT activation (Guen et al. 2017) especially through Hh signalling (Chahal et al. 2018;Eguether and Hahne 2018). Moreover, primary cilia have been closely related with cell migration both in physiology and pathology (Schneider et al. 2010;Veland et al. 2014), orientating parallel to the movement direction and sensing different signals that regulate cell migration and interacting directly with the extracellular matrix (Veland et al. 2014). Furthermore, this structure also plays an important role in cellular polarity (Tang and Marshall 2012) and its impairment in tumoral epithelial cells.
Primary cilia presence in urothelium has not been defined. Here, we have described for the first time primary cilia in the healthy bladder, NMIBC and MIBC through immunofluorescence and electron microscopy techniques.
Our results show that primary cilia appear on basal cells of healthy controls and NMIBC. In fact, we have interestingly shown that in low-grade tumours, ciliated cells cross basement membrane and invade stroma, which may explain the high recurrent and progression rates found in NMIBC and may point to primary cilia as an important target in these tumours.
The absence of statistical differences in the proportion of ciliated cells in the samples studied would indicate the importance of primary cilia in every tumoral stage in urothelial carcinoma pointing out a specific subpopulation. Thus, we believe this ciliated cellular subpopulation might be related to both, physiologic and cancer stem cells in urothelium as they are in other tissues (Lyu and Zhou 2017). These ciliated basal cells would be related to EMT, tumorogenesis, migration and invasiveness, probably through Hh pathway activation regardless of the ciliated cell proportion.
One study described a reduction of primary cilia in highgrade bladder cancer regarding non-cancerous tissues (Du et al. 2018). They described that they practically found no ciliated cells in NMIBC and MIBC. Conversely, we found an important number of ciliated cells in tumoral tissues. These opposite results could be a consequence of the heterogeneity found in human tumours, differences in sample processing or the short-length axoneme of urothelial cells which make cilia identification difficult. This short-length axoneme, generally embedded in the ciliary pocket, is in opposition to the long protruding cilia found in other tumours, such as GIST (Castiella et al. 2013). This difference might be in relation to the characteristic location of primary cilia near the cell membrane in epithelial cells (Sorokin 1968;Bernabé-Rubio and Alonso, 2017) and may affect ciliary signalling sensing.
On the other hand, we have described the main ultrastructural elements of EMT in urothelial carcinoma showing loss of cellular unions, formations of filopodia-like structures (which probable mediate cell migration), loss of polarity and enlargement of intercellular space. These processes are in parallel to primary cilia assembly, which remains during advanced stages (probably regulating tumoral progression). To the best of our knowledge, no one has previously described urothelial carcinoma EMT ultrastructural features.
Other ultrastructural findings showed in the present work point out important features of bladder cancer, i.e. ELCS, multivesicular bodies and autophagic vacuoles. ELCS are composed of chromatin fibers sheathed by inner nuclear membrane. Interestingly, they have been related to quiescent brain stem cells and they have been found in different tumours (Olins and Olins 2009;Cebrián-Silla et al. 2017). No one has described this structure in bladder cancer before. Bladder cancer exosomes play an important role in EMT activation and invasiveness (Franzen et al. 2015). Besides, we found abundant autophagosomes containing different organelles, especially rough endoplasmic reticula. Autophagy plays an important role in bladder cancer viability and drug resistance and its inhibition has been shown to increase tumoral cell apoptosis (Lin et al. 2017a, b).
Moreover, it has been described that primary cilia activate autophagy through Hh signaling (Pampliega et al. 2013), suggesting an additional mechanism for primary ciliadependent bladder cancer formation and maintaining.
However, the aim of this work was descriptive and focused on human biopsies to analyse the ultrastructure of these tumours, to confirm previous studies of EMT and Hh in bladder cancer lines (Islam et al. 2016;Miao et al. 2018;Kitagawa et al. 2019) and to give new evidence of primary cilia presence and potential implications. A limitation in this kind of study is the reduced number of samples.
In conclusion, our work sheds light on the importance of primary cilia in bladder cancer showing for the first time the presence of this structure in invasive cells of bladder cancer. Especially interesting is the presence of these ciliated invasive cells in NMBIC, which could explain the high recurrence index of these tumours. Primary cilia may play a fundamental role in basal cells which would initiate tumoral progression and invasiveness even in early tumoral stages probably through EMT and Hh activation. Furthermore, electron microscopy techniques have shown its importance to identify the ultrastructural nature of invading cells, giving evidence of potential targets for bladder cancer treatments.