A Group of ent-Kaurane Diterpenoids Inhibit Hedgehog Signaling and Induce Cilia Elongation

The Hedgehog (Hh) signaling pathway plays important roles in the tumorigenesis of multiple cancers and is a key target for drug discovery. In a screen of natural products extracted from Chinese herbs, we identified eight ent-Kaurane diterpenoids and two triterpene dilactones as novel Hh pathway antagonists. Epistatic analyses suggest that these compounds likely act at the level or downstream of Smoothened (Smo) and upstream of Suppressor of Fused (Sufu). The ent-Kauranoid-treated cells showed elongated cilia, suppressed Smo trafficking to cilia, and mitotic defects, while the triterpene dilactones had no effect on the cilia and ciliary Smo. These ent-Kaurane diterpenoids provide new prototypes of Hh inhibitors, and are valuable probes for deciphering the mechanisms of Smo ciliary transport and ciliogenesis.


Introduction
The Hedgehog (Hh) pathway plays fundamental roles in embryonic development, and aberrantly activated Hh signaling drives the formation of multiple cancers, particularly basal cell carcinoma (BCC) and medulloblastoma (MB) [1]. In the past decade, major progress has been made in Hh pathway targeted cancer therapies with Hh pathway antagonists (HPAs). Several Smoothened (Smo) antagonists are in clinical trials for treating a variety of cancers. For example, vismodegib (GDC-0449, Roche) is the first HPA approved by the FDA for treating BCC [2]. However, acquired resistance to vismodegib was soon observed in MB and BCC patients, resulting in cancer relapse [3,4]. Genomic analysis revealed that the predominant mechanism of resistance was Smo mutations that disrupted vismodegib binding with the pocket of Smo [5,6]. Thus, it has been widely suggested to develop new HPAs that target different molecules.
In mammals, Hh signal transduction depends strictly on the primary cilium. Many of the crucial components in the mammalian Hh pathway are concentrated within cilia, and most of the pivotal signaling transduction events take place in the cilia. Thus, the normal structures and functions of cilia are essential for Hh signal transduction [1,7,8]. In the absence of Hh ligands, the Hh receptor Patched (Ptch) locates to the cilia and prevents ciliary accumulation of Smo to repress the pathway. When Hh binds Ptch, Ptch is internalized and removed from the cilia, allowing Smo to enter and accumulate in the cilia, where it activates the transcriptional effector Gli by relieving its inhibition by Suppressor of Fused (Sufu) [1,7]. The primary cilia can either promote or inhibit BCC and MB development depending on the oncogenic events that have occurred, indicating that cilia could be a potential diagnostic marker and therapeutic target in Hh-related cancers [9,10].
Currently, most HPAs are synthetic compounds with similar structural skeletons and limited diversity [11]. Plant natural products, which possess enormous structural and chemical diversity, have been a continuing source for drug discovery, often showing surprising bioactivities [12]. The first Hh pathway inhibitor, Cyclopamine (Cyc), is a natural product from corn lilies that binds to the 7-transmembrane domain of Smo [13,14]. To discover new HPAs, we conducted a screen of natural products from herbs and identified a group of ent-Kaurane diterpenoids that inhibit the Hh pathway and promote cilia elongation.

Reagents
SAG, cytochalasin D, 5Z-7-Oxoeaneol, GANT58, EGCG and MG132 were purchased from Sigma, vismodegib from Selleck Chemicals, 20(S)-hydroxycholesterol (20(S)-OHC) from Cayman Chemicals, Bodipy-Cyclopamine from Toronto Research Chemicals, and vinblastine and AZ-TAK1 from Santa Cruz Biotech. measured using the Bright-Glo reagents (Promega) on a Fluoroskan Ascent (Thermo Fisher). All of the samples were conducted in triplicate. The Gli-firefly/Renilla luciferase ratio represented the Hh pathway activity. For Wnt reporter analysis, HEK 293W cells were seeded on 96-well plates at 60% confluence, treated with the compounds for 1 day, and then lysed to measure the reporter activities.

Bodipy-Cyclopamine Competition Binding Assay
Bodipy-Cyclopamine competition binding assays were conducted as previously described [13] with modifications. HEK 293 cells were seeded onto 24-well plates and transfected with Smo-mCherry using Fugene HD (Roche) at 70% confluence. After 2 days of Smo expression, the cells were incubated with 10 nM Bodipy-Cyc and 10 or 20 μM of various HPAs for 1 hour (100 nM of SAG was used as a positive control). Then, the cells were fixed and stained with Hoechst 33342 to visualize the nuclei. All of the images were captured under the same exposure conditions using a 10x objective with an Olympus FV1000 confocal microscope. At least five images were taken from each sample with similar Smo-mCherry expression levels. Fluorescence intensities of each image were quantitatively measured using ImageJ (NIH).

Real-time PCR Analysis of Hh Target Gene Expression
NIH 3T3 cells and Sufu -/-MEFs were grown to confluency and their medium were changed to 0.5% FBS medium diluted with various HPAs for 30 hr. Total RNA was extracted and purified using TRIzol (Invitrogen) according to the standard protocol. Next, 1 μg of total RNA from each sample was reversely transcribed to cDNA with random hexamer primer (Fermentas). The mRNA levels of mouse Gli1, Ptch1and Gapdh were quantified using LightCycler 480 SYBR Green I Master (Roche) on a LightCycler 480 system (Roche). The following primers were

Cilia and Ciliary Smo Analysis
For de novo ciliogenesis and ciliary Smo analysis, NIH 3T3 cells were grown to confluence, and then their medium was changed to serum starvation medium (0.5% FBS/DMEM) with various compounds for 30 hr. To analyze the effects of ent-Kauranoids on mature cilia, the confluent cells were first incubated in serum starvation medium for 1 day to fully induce cilia formation before the compounds were added for an additional 30 hr. Cilia and Smo immunostaining was performed according to previously reported procedures [25]. The following antibodies were used for the immunostaining: rabbit polyclonal anti-Arl13b (17711-1-AP, Proteintech), mouse monoclonal anti-Smo (sc-166685, Santa Cruz), mouse monoclonal anti-acetylated α-tubulin (T7451, Sigma), and donkey AlexaFluor 488 or 594 conjugated secondary antibodies (Invitrogen).
Images were acquired with identical setting parameters, using 6-plane Z-stacks (1 μm step size) to encompass all of the cilia and Smo in the field with a 100X oil immersion objective on an Olympus FV1000 laser confocal microscope. Images of the Z-stacks were summed as final micrographs for quantitative assessment with the method of maximal intensity projections using Image-Pro Plus 6 (Media Cybernetics). To measure the length of the cilia, images were calibrated with scale ruler, the cilia contour were outlined using the segmented line tool, and the lengths of cilia were measured with ImageJ. To quantitatively analyze the ciliary Smo, the cilia were manually outlined with a mask in the Arl13b channel and the Smo fluorescence intensities were then measured in the Smo image channel. The total signal for a single cilium was taken as one data point. Approximately 100 cilia and ciliary Smo were measured from ten images for each group.

Statistical Analysis
Calculation of the mean, SD and IC 50 with the Student's t test and one-way ANOVA analysis were performed with GraphPad Prism 6.

Tubulin and Actin Polymerization Assays
The effect of kamebakaurin on tubulin and actin polymerization was performed using a Tubulin Polymerization Assay Kit (BK011P, Cytoskeleton) and an Actin Polymerization Biochem Kit (BK003, Cytoskeleton), respectively, with the manufacturer's guidance.

A group of ent-Kaurane Diterpenoids and Triterpene Dilactones Inhibit Hh signaling
To search for novel HPAs, we conducted a screen of 500 natural products extracted from medicinal plants with a cell-based reporter assay using the Smo agonist SAG [26] to activate Hh signaling (Fig 1). During this screen, a group of ent-Kaurane diterpenoids and triterpene dilactones were identified. The 8 ent-Kaurene diterpenoids, kamebakaurin, kamebakaurinin [18], phyllostachysin H [19], calcicolin A [20], tenuifolin A, tenuifolin I [21], adenanthin C and adenanthin G [21], were isolated from plants of the genus Isodon. The 2 triterpene dilactones, longipedlactone L [23] and longipedlactone H [24], were isolated from Kadsura ananosma and Kadsura longipedunculata, respectively (Fig 2A). Plants from the genus Isodon are rich with diverse highly oxygenated ent-Kauranoids that have been used as anti-tumor, antibacterial and anti-inflammatory agents in Chinese folk medicine [27]. Triterpenoids from plants from the genus Kadsura have been reported to have anti-tumor and anti-HIV activities [21]. The chemical structures of the ten HPAs (Fig 2A) are different from all known Hh antagonists and represent novel prototypes of Hh inhibitors.
The 8 ent-Kauranoids share an ent-Kaurane skeleton with different hydroxyl and acetoxyl substituents (Fig 2A) and are categorized into C-14, 20-oxygenated and non-oxygenated groups, with the oxygenated compounds having stronger activities (Fig 2B, S1 Fig). These results reflect the structure-activity relationships of ent-Kauranoids that inhibit the Hh pathway. The 2 triterpene dilactones, longipedlactones L and H, had similar structures and activities (Fig 2A and 2B, S1 Fig).
Gli1 and Ptch1 are key components of the Hh pathway and are also Hh pathway target genes [1]. The HPAs significantly inhibited Gli1 and Ptch1 expression stimulated by SAG ( Fig  2C). In contrast, these HPAs had no clear inhibitory effect on Wnt signaling (S2 Fig). The HPAs also repressed Hh signaling induced by the Smo agonist 20(S)-OHC, which is different from SAG and binds to the cysteine-rich domain of Smo [28] (Fig 2D). The evidence suggests that HPAs act at the level or downstream of Smo. Cyclopamine inhibits Hh signaling by binding to the heptahelical bundle of Smo, which can be competed away by most known Smo-targeting HPAs [11,13]. In a competition experiment, the ent-Kauranoids and triterpene dilactones showed no effect on the binding of Bodipy-labeled Cyc to Smo-expressing cells ( Fig  2E and 2F), suggesting that they do not bind the Cyc-binding pocket of Smo. Sufu is a negative regulator of the Hh pathway that is downstream of Smo and Sufu knockout results in activated Hh signaling [16,29]. Real-time PCR results showed that the ent-Kauranoids and dilactones had no clear inhibitory effect on Gli1 and Ptch1 expression in Sufu -/-MEFs (Fig 2G), suggesting that they likely work upstream of Sufu. Given that kamebakaurin has been reported to inhibit the nuclear factor κB (NF-κB) pathway [30], we tested whether established NF-κB inhibitors, AZ-TAK1 [31] and 5Z-7-Oxoeaneol [32], could repress Hh signaling. As they both showed no clear effect on the expression of the Hh reporter stimulated by SAG (S3 Fig), we hypothesize that ent-Kauranoid HPAs do not indirectly inhibit Hh by repressing the NF-κB pathway.

ent-Kauranoids Elongate Primary Cilia and Inhibit Ciliary Accumulation of Smo
Smo enrichment in the cilia is critical for the activation of Hh pathway [7,8]. Thus, we evaluated the effects of the HPAs on the ciliary accumulation of Smo induced by SAG. The ent-Kauranoids, but not the triterpene dilactones, remarkably reduced the levels of ciliary Smo   Fig 1. The activity distribution of the herbal natural products in the screen for Hh pathway antagonists. The primary screen for Hh pathway antagonists was performed in Shh light II cells treated with 10 μM natural products (stock concentration in DMSO at 10 mM) with 150 nM SAG for 30 hr. The selected primary hits displayed an inhibitory effect on the Gli-reporter of more than 70% and Renilla activity of less than 30%. Activities of the compounds were normalized to SAG.   Fig 3A and 3B). Unexpectedly, the cilia were dramatically elongated in cells treated with the ent-Kauranoids but not the dilactones (Fig 3A and 3C). The intraflagellar transport (IFT) complex proteins affect the Hh pathway downstream of Smo and upstream of Gli, which are essential for the assembly and maintenance of primary cilia and ciliary transportation [7,33]. We speculate that the ent-Kauranoids likely act at similar levels as IFT, by affecting ciliogenesis and ciliary transport of cargo.
In the above experiments, the ent-Kauranoids were directly added after the cells reached confluence, when de novo ciliogenesis was to start. We further investigated their effects on well-formed mature cilia. The confluent cells were starved for one day to fully induce cilia formation before the drugs were added. Under such conditions, the ent-Kauranoids showed more prominent and variable effects on the elongation of the cilia as are shown by quantitative analyses of staining for the ciliary axoneme marker acetylated α-tubulin (Fig 4A). For example, cilia in kamebakaurin-and kamebakaurinin-treated cells were 12.59 ± 4.87 μm and 10.92 ± 5.16 μm in length, which were 3.92-and 3.40-fold of normal cilia, respectively. Cells treated with other ent-Kauranoids also had elongated cilia, ranging from 6.06 to 8.74 μm, while the cilia in cells treated with triterpene dilactones were normal (Fig 4A and 4B). Moreover, cilia with abnormal morphology, such as cilia with twisted and wave-like axoneme or bulged tips or cells with 2 or 3 cilia, were frequently observed (Fig 4C). KIF7 is a cilia-associated protein belonging to the kinesin family that plays a role in the hedgehog signaling pathway [34][35][36][37]. Cells with mutated Kif7 grow long and twisted cilia, similar to the cells treated with ent-Kauranoids. However, Kif7 mainly regulates the Sufu-Gli complex without affecting Smo [34][35][36][37], making it unlikely to be the target of the ent-Kauranoids.

Kamebakaurin Disturbs Spindle Assembly and Chromosome Congression
Kamebakaurin is the most potent ent-Kauranoid that we identified in Hh inhibition and cilia elongation experiments. Interestingly, when treated with kamebakaurin, cells with binuclei or micronuclei were frequently observed to undergo abnormal mitosis (S4 Fig). Thus, we examined the mitotic spindle structures in kamebakaurin-treated cells, which were enriched to metaphase by MG132 [25]. Indeed, kamebakaurin treatment led to defective (distorted, multipolar) spindles with poor spindle microtubule assemblies and spindles lacking of the fusiform shape (Fig 5A). The congression of chromosomes often failed when the chromosomes misaligned with no attachment to the microtubules. In control groups, cells incubated with vehicle alone had normal spindle morphology and chromosomes congression (Fig 5A and 5B). The distribution of the pericentriolar matrix protein pericentrin became more diffused (Fig 5A, S5  Fig). At a lower concentration (25 μM), kamebakaurin induced clear but less severe spindle malformation in treated cells. (Fig 5A, S5 Fig). We next examined whether kamebakaurin inhibits spindle formation by interfering with the polymerization of microtubules and actin. Kamebakaurin showed no effects on either microtubule or actin dynamics in vitro (Fig 5C and  5D), suggesting that it likely targets protein(s) other than tubulin and actin.

Discussion
In the last decades, major progress has been made in the development of small molecules that specifically inhibit the Hh signaling pathway. Currently, most identified HPAs target Smo, several of which are being tested in clinical trials for the treatment of multiple types of cancer [2]. Unfortunately, acquired resistance to Smo inhibitor has already been reported in patients with advanced BCC and MB [3,4]. Thus, Hh pathway inhibitors acting downstream of Smo could be more promising for such patients. In our study, we identified two triterpene dilactones and a group of ent-Kauranoids as potent Hh pathway inhibitors with distinct mechanisms of  Smo and promote cilia elongation, whereas the triterpene dilactones have no such effects. Kamebakaurin, the most potent compound identified in our screen, also causes mitosis defects, but does not directly target tubulin or actin. The ent-Kauranoids might target common protein (s) involved in Smo ciliary transport, ciliogenesis and mitosis. Similar to the ent-Kauranoids, ciliobrevin (HPI-4) has been shown to inhibit Hh signaling while simultaneously affecting cilia length, Gli2 localization and spindle formation [25,38]. Subsequent work demonstrated that it targets the cytoplasmic motor protein dynein, which is involved in ciliary trafficking, mitotic spindle formation and organelle transport [24].
In addition to dynein, there are other proteins that are reported to participate in both cilia formation and mitosis, such as the IFT motor protein Kif3a [39,40], the IFT complex component IFT88 [40,41], and the oncogenic Aurora A kinase [42,43]. Specifically, the Ran importin system was recently found to regulate ciliogenesis [44], which forms a diffusion barrier at the base of cilia to selectively control the entry of proteins into cilia [45,46]. The Ran importin system is also critical for the accurate progression of mitosis, playing roles in spindle microtubule assembly, kinetochore-microtubule attachment and chromosome congression [47]. Interestingly, ent-15-oxokaurenoic acid, a chemical with similar backbone to the ent-Kauranoids, has been reported to cause mitotic arrest and defective chromosome congression by targeting RanBP2 [48], a Ran binding protein from the Ran importin system. Additionally, RanBP9 (RanBPM) and RanBP10 have been identified as Smo-binding proteins, with RanBP9 being involved in the cilia transportation of Smo [49]. Thus, the Ran binding proteins are attractive candidate targets for ent-Kauranoids, a possibility that awaits further investigation.
Maintenance of normal ciliary length is critical for the diverse functions of cilia including the Hh signal transduction, which has been reported to be affected by many proteins and compounds [25,33,50,51]. The ent-Kaurane diterpenoids we identified might provide new tools for deciphering the mechanisms of ciliogenesis and cilia-dependent regulation of the Hh pathway.