Primary Cilia Structure
The primary cilium is a microtubule-based organelle that projects from the surface of most vertebrate cells. Under most circumstances, each cell only has one primary cilium (Fig. 1a, b). Unlike motile cilia, such as those found on respiratory epithelial cells and eukaryotic flagella, primary cilia are not involved in motility. Rather, primary cilia are responsible for transmitting cellular signals, and have a specialized structure that facilitates signal transduction . Vertebrate primary cilia project from the surface of the cell, emerging from a single centrosomal centriole known as the basal body, and are surrounded by the cell membrane. Within the membrane is the axoneme, which consists of a ring of nine microtubule doublets, known as a 9+0 configuration. In contrast, motile cilia additionally have two central microtubule singlets and dynein arms attached to the outer microtubule doublets, altogether known as the 9+2 configuration (Fig. 1b).
The transition zone at the base of the cilium acts as a gate that controls protein entry and exit . Thus, primary cilia serve as a unique microenvironment where specific proteins and lipids can interact in a dynamic fashion to alter cell signaling, function, and fate. Consequently, dysfunctional primary cilia structure or signaling can lead to cancer and congenital conditions collectively known as “ciliopathies” . Thus, understanding primary cilia is crucial to understanding the pathogenesis and clinical manifestations of many diseases.
Cilia were initially observed in protozoa by Antony van Leeuwenhoek in 1677, and were defined by their motility as the first cellular organelle [21, 22]. Almost two centuries later, Alexander Kovalevsky perceived that many vertebrate cells had a single, immotile cilia . This discovery was soon corroborated by other scientists, including Karl Wilhelm Zimmermann, who named them “centralgeissel” (central flagella) in mammalian cells and suggested that they may have sensory potential [24, 25]. Despite the short burst of scientific interest in centralgeissel in the late nineteenth century, the organelle was largely neglected and dismissed as vestigial because of lack of any obvious function. In 1968, immotile centralgeissel on mammalian cells were renamed primary cilia because their expression could be identified first during fetal development, before the emergence of motile cilia .
Interest in primary cilia smoldered to life in 1993 with the seemingly unrelated discovery of intraflagellar transport (IFT) in the unicellular green alga Chlamydomonas . IFT is the method by which cells move cargo, such as proteins, along the microtubules within cilia and flagella. Specifically, scientists discovered that loss of the IFT88 gene in Chlamydomonas resulted in loss of flagella . Interestingly, mice that lack Tg737 (the murine homolog of IFT88) have deformed primary cilia in their kidney cells and develop polycystic kidney disease . This nascent connection between cilia and disease renewed scientific interest in the primary cilium and sparked an explosion of primary cilium investigation that has persisted to the present day. A more detailed timeline and summary of significant advances in the cilia literature can be found in Fig. 2 and Table 1, which are based on information from a review article by Bloodgood .
Since the connection to polycystic kidney disease, an entire class of disorders, known as “ciliopathies” and spanning almost every organ system, have been linked to dysfunction in the primary cilium . Ciliopathies are inherited genetic disorders that are associated with mutations in genes that are important for the function of primary cilia. Outside of an inherited context, dysfunction of primary cilia has also been implicated in sporadic cancers [31, 32]. The mechanisms underlying both ciliopathies and the primary cilium’s role in cancer have been linked to more recent discoveries implicating the primary cilium as a central hub for many gene expression programs that are crucial for development and adult stem cell homeostasis, such as the HH , Wnt , and Notch  pathways.
BCC and HH Signaling
BCC affects approximately two million people per year and is the most common cancer in the USA . BCC typically arises in adult patients as a result of ultraviolet radiation-induced genomic mutations. Indeed, BCCs are the most mutated human cancer, with about 65 mutations per megabase . Although one might expect these numerous mutations to be random, BCCs are unified by misactivation of the HH pathway in both sporadic and syndromic cases. With respect to the latter, individuals with inherited heterozygous mutations in HH pathway inhibitor genes develop nevoid BCC syndrome (also known as Gorlin syndrome), which is characterized by numerous BCCs and other HH-associated cancers such as medulloblastoma and rhabdomyosarcoma [37,38,39,40,41]. Primary cilia play a crucial part in the pathogenesis of BCC because of the primary cilium’s essential role in the HH signaling pathway.
The HH pathway is activated by HH ligands, which are secreted lipoproteins . The best studied HH ligand gene encodes Sonic hedgehog (SHH), a crucial regulator of embryonic development and adult tissue homeostasis [43, 44]. HH ligands stimulate the HH pathway, often in nearby cells, by binding to their transmembrane receptors, patched 1 (PTCH1) and patched 2 (PTCH2), which are present at the ciliary membrane [45,46,47] (Fig. 3). HH signaling is transduced by the primary cilium. When HH binds to PTCH proteins, another transmembrane protein, Smoothened (SMO), accumulates at the cilium to activate the downstream pathway [48, 49] (Fig. 3). SMO activates GLI2, the primary activator of the HH transcriptional program, which in turn regulates GLI1, a feed-forward amplifier of transcriptional activity [50, 51]. When the HH pathway is inactive, suppressor of fused (SUFU) binds to GLI family members and represses their transcriptional activity [52,53,54] (Fig. 3).
Primary cilia and the HH pathway play crucial roles in normal skin development and growth, thus their dysregulation in BCC is perhaps not surprising. In the developing skin, SHH secreted by epidermal placodes induces proliferation and hair follicle formation in the underlying dermal condensates . Without SHH, epidermal placodes form but fail to grow down into the dermis . Similarly, in postnatal skin, hair follicle growth also relies on signaling through the HH pathway. For example, SHH is expressed during anagen, the hair follicle phase during which there is downward growth of the follicle into the dermis, and is important for maintaining epidermal stem cells and regulating epidermal growth [57,58,59].
The most common HH pathway-activating mutations in BCC include biallelic loss-of-function mutations in pathway inhibitors (e.g., PTCH1, PTCH2, and SUFU); amplification of GLI2, the principal HH pathway transcriptional activator; and monoallelic activating mutations in pathway activators, such as SMO . BCC cells possess cilia (Fig. 1a), which transduce HH signals from SMO to GLI2 [60, 61]. In fact, in mouse models of BCC, scientists have shown that both the ciliary gene Intu and HH signaling are necessary for BCC tumorigenesis and progression [61, 62]. Thus, the primary cilium, through HH signaling, acts a crucial nexus in the pathogenesis of BCC.
BCCs are typically slow-growing and are most often effectively treated with local excision. However, many factors can prevent complete excision, such as number or size of tumors, or proximity to critical structures, including the eye, lip, and nose. In these cases, nonsurgical local treatments, such as topical cytotoxic agents, radiotherapy, photodynamic therapy, and cryotherapy, can be used . In the small subset of patients with locally advanced or metastatic BCC, systemic therapy is indicated. For such cases, HH pathway inhibition with SMO antagonists, such as vismodegib or sonidegib, has been shown to be more effective than chemotherapy [64,65,66]. Although the proportion of BCC patients who are eligible for molecular therapy is small, the tremendous incidence of BCC cases each year makes the absolute number of patients who may be considered for vismodegib or sonidegib large. Unfortunately, systemic inhibition of the HH pathway can lead to adverse events, such as nausea, muscle cramps, loss of taste, weight loss, and alopecia . Although relatively mild, these symptoms can cause patients to not adhere to treatment regimens, which may lead to BCC recurrence. Thus, the combination of radiotherapy with HH pathway inhibition may be used to achieve durable responses with cessation of systemic therapy for such patients .
In addition to recurrence due to lack of adherence, resistance to vismodegib and sonidegib has also been documented, typically via mutations in SMO, the target of both inhibitors [69, 70]. A frequent activating mutation in SMO is W535L, also known as SMOM2, which causes SMO to accumulate in the cilium even in the absence of HH ligands [71, 72]. In medulloblastoma, another HH-driven cancer where HH pathway inhibitors are used, there are examples of resistance that arise from amplification of targets downstream of SMO, such as GLI2 or cyclin D1 [73, 74]. Outside of alternative methods of HH pathway activation, rare examples of BCC resistance have been seen via loss of ciliation, loss of HH signaling, and subsequent activation of alternative signaling pathways, such as the Ras/MAPK pathway . Overcoming resistance to SMO antagonists in BCC is an active area of research, with some efforts focused on targeting downstream elements of the HH pathway. HH pathway-independent treatment options, such as cancer immunotherapy, have also been proposed for resistant tumors. Given BCC’s high mutational burden and the correlation between mutational burden and the success of immunotherapy, clinical trials with anti-PD1 therapy have been initiated (NCT03132636, NCT03521830).
There are diverse genetic changes and transcriptional programs that contribute to melanoma pathogenesis. Prominent activating mutations in key oncogenic driver genes, such as BRAF or NRAS, are present early in melanocytic nevi, but the cellular programs defining proliferative and invasive phenotypes are not defined by particular genetic changes, rather are governed by transcriptional master regulators [75,76,77,78]. It is the culmination of these complex changes that translates to biologic function, which is further refined by an individual’s immune system and microenvironment . A detailed review of the genetic and phenotypic changes occurring in melanomagenesis is outside the scope of this review, but can be found elsewhere [75, 76]. Briefly, the initiating mutations result in uncontrolled proliferation of a melanocyte, followed by approximately 5–10 additional pathogenic alterations spread over several signaling pathways to result in melanoma . It is also generally accepted that there are “intermediate” cell states between the “benign” and “malignant” categories of melanocytic neoplasms . Although different combinations of mutations in a variety of genes underlie the diversity of melanocytic neoplasms, they all share the common denominator of activating the mitogen-activated protein kinase (MAPK) pathway. Beyond DNA mutations, deletions and amplifications, DNA methylation , microRNAs , transcription , and translation  also play important roles in determining the malignant potential of a particular melanocytic neoplasm . Finally, the tumor’s microenvironment and interaction with an individual’s immune system also contribute to the clinical course of melanoma . Importantly, regardless of the variability of the genetic landscape, there is downstream convergence on a malignant phenotype.
Primary Cilia in Melanocytic Neoplasms
The presence of primary cilia on the cell surface of lesional melanocytes within conventional melanocytic nevi was first reported in 2011 by Kim et al. . This was in contrast to the near-complete loss of primary cilia on in situ, invasive, and metastatic melanoma. At that time, it was also investigated whether the observed loss of primary cilia might be due to ongoing cell cycle progression, as primary cilia reabsorption is a known physiologic response during the interphase portion of the cell cycle . In their study, staining for Ki-67, a protein present during the cell cycle and absent during G0, was performed and found to vastly under-represent the percentage of non-ciliated cells. This was the initial evidence that active cell cycle progression was unlikely to account for the almost complete loss of primary cilia observed in melanoma.
An independent group, Snedecor et al., validated the original findings by evaluating a cohort of melanocytic nevi compared to in situ, invasive, and metastatic melanoma . Interestingly, although still significantly higher than their melanoma cohort, the average percentage of ciliated melanocytes within the nevus cohort was 25%. This is notably lower than that of the Kim et al. study demonstrating an average of 94% ciliated melanocytes in nevi. A direct comparison of these results may be limited by variations in immunofluorescence staining methods used as well as possible variability in the types of melanocytic nevi included. However, Snedecor et al. additionally contributed to the hypothesis that cilia loss in melanoma was unlikely to be solely due to an elevated proliferation rate.
The underlying mechanism of primary cilia loss in melanoma remained a mystery until a recent study examining the role of “enhancer of zeste homolog 2” gene (EZH2) overexpression/amplification in melanoma . Zingg et al. uncovered that many cilia-associated genes were transcriptionally repressed by EZH2, which functions as the histone methyltransferase unit of polycomb repressive complex 2 (PRC2) . PRC2 is a complex that methylates histone H3 to promote transcriptional silencing. As a global regulator of gene transcription, amplification or gain-of-function of EZH2 has significant downstream effects that have been associated with the transition to a melanoma state . This study also experimentally demonstrated activation of the beta-catenin pathway as a consequence of primary cilia knockdown. These data provide insights into the intricacies of potentially targeting the Wnt/beta-catenin signaling pathway in melanoma therapy. Given the correlation of beta-catenin pathway activation and melanoma immune evasion , primary cilia-related proteins may also provide novel candidate targets for patients resistant to immunotherapies.
The function of primary cilia in the context of melanocyte biology and melanomagenesis is still in early phases of investigation; however, it appears to have an active role in suppressing the oncogenic activity of the Wnt/beta-catenin pathway. Importantly, linking the loss of primary cilia to upstream global effectors of melanomagenesis brings us closer to appreciating the potential value of using primary cilia staining as a diagnostic tool in ambiguous melanocytic neoplasms.
Melanoma Molecular Diagnostic Tools
The complexity and diversity of biologic checkpoints involved in the development of melanoma reflects the challenges that both clinicians and pathologists can face in making an accurate diagnosis. In particular, the “intermediate” lesions that do not meet histopathologic criteria for melanoma but have atypical features raise the question of their true biologic potential. Many of the most well-established adjunct diagnostic tools available to dermatopathologists for the diagnosis of ambiguous melanocytic neoplasms involve either genetic/genomic analysis or single protein detection by immunohistochemistry [13, 14]. There is now a greater appreciation that a purely genetic analysis may miss critical post-transcriptional and post-translational changes. Techniques such as mass spectrometry  and micro-RNA profiling  are emerging to address this gap. Each of these tests provides additional information that the pathologist can then use to integrate with the histopathology to reach their best estimate of malignant potential. The decision to initiate additional molecular analysis beyond routine histopathologic assessment largely rests on the ease of performing the test, the availability of sufficient tissue, and expertise for interpretation.
Performing primary cilia staining is no more technically challenging than standard immunohistochemical staining for single proteins, which is currently the mainstay of additional workup for histopathologically challenging melanocytic neoplasms. Immunofluorescence staining for primary cilia requires a single unstained standard thickness tissue section from a paraffin-embedded formalin-fixed tissue block (detailed methods found in Kim et al. ), and can be easily incorporated into a routine diagnostic workup. Three primary antibodies are combined in a single cocktail for the identification of melanocytes (SOX-10), the ciliary axoneme (acetylated alpha-tubulin), and centrioles (gamma-tubulin). The corresponding three secondary antibodies are conjugated to fluorophores with distinct excitation and emission wavelengths, which allows for concurrent identification of all three components using an immunofluorescence microscope attached to a camera. An example of this immunofluorescence stain is shown in Fig. 4, which highlights primary cilia in a melanocytic nevus and cilia depletion in melanoma.
As this cell surface structure is clearly absent in melanoma when compared to conventional melanocytic nevi, it has the potential for being used in a similar manner as immunohistochemical stains for p16 or PRAME are currently being used [79, 91]. The most common acquired genetic change distinguishing precursor lesions, such as melanocytic nevi or melanoma in situ (MIS), from invasive melanomas is loss of the CDKN2A locus, which is considered an important driver of melanoma . Loss of immunohistochemical staining for the p16 protein can act as a surrogate of the underlying genetic event; however, negative staining for p16 does not always correlate with an underlying mutation being present and conflicting data argues against its use . Conversely, preserved p16 staining does not exclude the possibility of melanoma, and in fact approximately 25% of metastatic melanoma can retain this tumor suppressor gene (TCGA Research Network). With respect to PRAME immunohistochemical staining, there has been rapid adaptation of this stain for clinical use, but as with any single protein, the results must be interpreted with caution in the context of all clinical and histopathological findings. Overall, the cumulative literature results support the need for additional biomarkers, such as primary cilia staining, to help in cases when distinguishing benign from malignant by current immunohistochemical staining practices is insufficient.