Kinome-wide siRNA Screen Identifies a DCLK2-TBK1 Oncogenic Signaling Axis in Clear Cell Renal Cell Carcinoma

Summary

TANK-binding kinase 1 (TBK1) is a potential therapeutic target in multiple cancers including clear cell Renal Cell Carcinoma (ccRCC). However, targeting TBK1 in clinical practice is challenging. One approach to overcome this challenge would be to identify an upstream TBK1 regulator that can be targeted therapeutically in cancer specifically. In this study, we perform a kinome-wide siRNA screen and identify Doublecortin like kinase 2 (DCLK2) as a TBK1 regulator in ccRCC. DCLK2 binds to and directly phosphorylates TBK1 on Ser172. Depletion of DCLK2 inhibits anchorage-independent colony growth and kidney tumorigenesis in orthotopic xenograft models. Conversely, overexpression of DCLK2²⁰³, a short isoform that predominates in ccRCC, promotes ccRCC cell growth and tumorigenesis in vivo. Mechanistically, DCLK2²⁰³ elicits its oncogenic signaling via TBK1 phosphorylation and activation. Taken together, these results suggest that DCLK2 is a TBK1 activator and potential therapeutic target for ccRCC.

Graphical Abstract

eTOC blurb:

Hu et al. discovered that a kinase DCLK2 promotes tumorigenesis in ccRCC by activating TBK1 phosphorylation. The oncogenic role of DCLK2 is mainly elicited by a specific splice isoform. Targeting DCLK2 to selectively inhibit TBK1 activity in cancer cells could emerge as a potential therapeutic strategy in ccRCC.

INTRODUCTION

The von Hippel-Lindau (VHL) gene was identified as the target of germline mutations in patients with a clear cell renal cell carcinoma (ccRCC) predisposition syndrome¹. Additionally, inactivating VHL mutations play a major role in sporadic ccRCC². The VHL protein forms part of a complex that exhibits E3 ubiquitin ligase activity and loss of VHL leads to hypoxia inducible factor α (HIFα) accumulation, which contributes to the oncogenic phenotype of ccRCC^1,3,4. While HIF1α is thought to function as a tumor suppressor^5–7, HIF2α stabilization (as a result of VHL loss) is both necessary and sufficient for ccRCC tumor growth⁶. PT2399, a specific HIF2α antagonist, inhibits tumor growth in about 50% of ccRCCs^8,9. However, a significant portion of ccRCC are resistant to HIF2α inhibitors^8,9, highlighting the importance of identifying additional therapeutic vulnerabilities of VHL-deficient ccRCC.

We recently demonstrated that TBK1 is activated in VHL-deficient ccRCC¹⁰. We found that TBK1 binds to and is inactivated by VHL but does not get degraded (in contrast with VHL interactions with HIF and other factors). Loss of VHL leads to TBK1 phosphorylation at Ser172 leading to downstream phosphorylation and stabilization of p62/SQSTM1, an important driver of ccRCC tumorigenesis¹¹. Notably, TBK1 is critical for ccRCC cell proliferation, tumor growth and supports spontaneous metastasis, therefore emphasizing that targeting TBK1 may preferentially inhibit VHL-null cancer cells in this setting¹⁰. Although studied in several cancers, TBK1 and the closely related IKKε kinase have been most extensively implicated in the innate immune response¹². For example, in the widely studied STING signaling pathway, TBK1 is activated following binding of STING to cGAMP, a second messenger synthesized by cGAS in response to pathogen invasion. Subsequently, TBK1 phosphorylates the transcription factor IRF3, which translocates to the nucleus to drive interferon-β gene expression^13,14. Interestingly, despite clear TBK1 activation in ccRCC cells (dependent on VHL loss), we found no evidence of phosphorylation of IRF3 (or IKKε activation) in ccRCC cells¹⁰. Thus, TBK1 promotes cancer-related properties in ccRCC through a mechanism that is independent of its role regulating interferon.

In several biological settings, TBK1 is thought to transphosphorylate at Ser172 for its activation¹⁵. Nevertheless, Cohen and colleagues have proposed the existence of an upstream TBK1 kinase^16,17. However, a kinase that activates TBK1 by direct phosphorylation on Ser172 remains unclear in cancer cells. Although targeting TBK1 in cancer is a promising concept, this may suppress innate immunity and there are no TBK1 specific inhibitors. Currently available TBK1 inhibitors consistently inhibit the related IKKε kinase^18,19. To overcome these difficulties, one approach would be to identify a critical and specific upstream TBK1 regulator implicated in tumorigenesis.

Here, we report the results of a kinome-wide siRNA screen that led to the identification of Doublecortin like kinase 2 (DCLK2) as a TBK1 kinase that phosphorylates Ser172 in cancer cells supporting ccRCC tumorigenesis. Doublecortin like kinase 1 and 2 (DCLK1 and 2) belong to the doublecortin (DCX) protein family characterized by two evolutionary conserved DCX domains, which includes DCX among other members²⁰. The DCX domain is implicated in binding microtubules and DCX as well as DCLK1 and 2 are microtubule associated proteins (MAPs)²¹. Mutations in DCX result in lissencephaly or subcortical band heterotopia (SBH)^22,23, a neural migration disorder. Like DCX, DCLK1 and 2 also regulate neuronal migration and axon growth during development^24,25. Overexpression of DCLKs (especially DCLK2) can also promote neuronal survival and growth cone regeneration after injury²⁶. In cancer, DCLK1 is a cancer stem cell (CSC) marker. It is expressed in CSCs but not NSCs (normal stem cells), as determined by lineage tracing techniques in the murine intestine²⁷. In the pancreas, DCLK1 marks a rare population of long-lived, quiescent pancreatic progenitor cells involved in regeneration and tumorigenesis²⁸. In addition, DCLK1 is also involved in other cancer types including kidney cancer^29,30, breast cancer³¹, hepatocellular carcinoma^32,33 and cholangiocarcinoma³⁴. Both DCLK1 and DCLK2 have multiple splicing variants and some DCLK1 variants have been linked to tumor promotion^29,35,36. Compared to DCLK1, very little is known about DCLK2 in tumorigenesis, but their high similarity (68% AA identity) suggests that DCLK2 may be similarly involved. In this study, we identify DCLK2 as a TBK1 activator in ccRCC and establish its role in tumorigenesis and as a potential therapeutic target.

RESULTS

Kinome-wide siRNA screen to identify TBK1 upstream kinases in kidney cancer cells

The mechanism of TBK1 activation in innate immune signaling has been well established but the upstream regulators for TBK1 autocrine signaling in cancer remain understudied (Figure S1A)¹⁵. To search for potential upstream kinases contributing to TBK1 phosphorylation in ccRCC, we performed a kinome-scale siRNA screen in kidney cancer cells with VHL loss (Figures 1A and S1B). 709 pooled siRNAs (3 independent siRNAs within each pool) targeting either kinases or kinase-associated proteins were individually transfected into a representative ccRCC cell line with VHL loss (UMRC6). Phosphorylation of TBK1 at Ser172 (pTBK1), the key site in the activation loop responsible for TBK1 activation, was examined by western blot. To confirm that the custom siRNA library contains efficient siRNAs that can silence gene expression, we transfected an siRNA pool for TBK1 and IKKε and found that both led to efficient gene silencing (Figure 1B). We measured the ratio of pTBK1/TBK1 in transfected cells and plotted pTBK1/TBK1 ratios. By applying a cut off value of −1.2 for Log2(pTBK1/TBK1), we narrowed the potential positive hits to 42 different genes (Figure 1C). These included kinases, non-kinase genes such as kinase-associated scaffold proteins or other less characterized targets. Interestingly, our screen retrieved 3 protein kinase C subtypes: PRKCD, PRKCG and PRKCH. Notably, protein kinase C has been shown to activate TBK1 through phosphorylation at Ser716 in lung cancer³⁷. ULK1 phosphorylates TBK1 on Ser172 in adipocytes³⁸, but ULK1 was not identified in our screening with ccRCC cells.

Figure 1. Kinome-wide siRNA screen to identify TBK1 activators in ccRCC.

A, Schematic model of the kinome-wide siRNA screen.

B, Immunoblots of lysates from UMRC6 cells transfected with Ctrl siRNA or pooled TBK1 (or IKKε) siRNA as indicated.

C, Relative pTBK1/TBK1 level of the potential candidates selected from the siRNA screen for further validation.

D-E, Immunoblots of lysates from UMRC6 cells transfected with individual and pooled (M) siRNA of indicated candidate genes. D, kinase genes whose depletion can decrease pTBK1 level; E, non-kinase genes whose depletion can decrease pTBK1 level.

Next, we obtained 3 independent siRNAs from the library against each gene and also made individual pools (M). We transfected these siRNAs into UMRC6 cells and performed western blot analyses. Based on these results, we narrowed down the potential TBK1 regulators to 8 kinases (DCLK2, WNK2, SRPK3, STK3, MAPKAPK3, TWF2, ADCK5, CAMK1G) and 1 pseudokinase (PRAG1) (Figure 1D). Both the pools (M) as well as the individual siRNAs decreased TBK1 phosphorylation. We also identified 5 non-kinase whose depletion decreased pTBK1 (Figure 1E). Since this screen was performed in one cell line, the result could be likely cell-line specific effect rather than cancer-type specific. However, the candidates may be important for TBK1 phosphorylation in the representative ccRCC cell line, which motivated us to perform additional experiments to examine their role in additional ccRCC cell lines. Since our aim was to identify direct TBK1 activators, we focused on the kinases.

DCLK2 interacts with TBK1 and promotes TBK1 phosphorylation on serine 172

Next, we examined whether these candidates would bind TBK1 by co-transfecting these proteins in 293T cells. Among these candidates tested, SRPK3, MAPKAPK3 and DCLK2 displayed binding with FLAG-TBK1 (Figure 2A). In addition, we performed in vitro kinase assays with in vitro translated (IVT) TBK1 mixed with ATP and recombinant kinases. Only DCLK2 and STK3 promoted strong TBK1 phosphorylation (Figures 2B and 2C). However, in the following experiments we found STK3 depletion did not consistently lead to decreased TBK1 phosphorylation in other ccRCC cells including 786-O and UMRC2 (Figure S2A). Thus, we decided to focus on characterizing the role of DCLK2 in controlling TBK1 activity in ccRCC.

Figure 2. DCLK2 interacts with TBK1 and directly phosphorylates TBK1 on Ser172.

A, Immunoblots of whole cell lysates (Input) and immunoprecipitations (IP FLAG) from 293T cells transfected with indicated plasmids.

B-F, Immunoblots of in vitro kinase assay samples as indicated. TBK1 protein in F is shown by Coomassie Bright Blue staining.

G, Potential phosphorylation sites on TBK1 protein and intensity identified by MS. Samples are from F lane 2 and lane 3.

H, Quantification of the intensity of TBK1 protein and Ser172 phosphorylation from MS result. Data are presented as mean ± SEM. ***p < 0.001. n.s., no significance.

I, MS spectra of both phosphorylated and unphosphorylated peptides containing TBK1 Ser172.

The role of DCLK2 in cancer has not been evaluated until recently³⁹. Previous research showed that TBK1 undergoes autophosphorylation at Ser172^16,40. To examine whether the effect with DCLK2 on TBK1 phosphorylation was independent of autophosphorylation, we made a catalytically inactive TBK1 mutant (D157A) which displays minimal autophosphorylation¹⁷. We found that DCLK2 could still promote TBK1 phosphorylation, albeit to a significantly lesser extent (Figure 2D). We also used CMPD1, a TBK1 inhibitor¹⁸, to inhibit TBK1 autophosphorylation in kinase reaction and found that DCLK2 still promoted TBK1 phosphorylation (Figure 2E). We then repeated kinase assay with a purified TBK1 protein (1-657AA) that contains D135N, another kinase dead mutation⁴¹ and analyzed phosphorylation status of TBK1 in the absence or presence of DCLK2 by mass spectrometry (MS) (Figure 2F, Lane 2 and 3 respectively). Three potential phosphorylated sites were identified by MS (Figure 2G, Table S1). Among these sites, only TBK1 S172 phosphorylation was detected in the presence of DCLK2 but not in the absence of DCLK2 (Figures 2G–2I). Overall, these data suggest that DCLK2 is able to phosphorylate TBK1 on Ser172, but the effect is significantly amplified by a catalytically active TBK1. As control, IKKε, the close related IKK-family kinase, was not phosphorylated by DCLK2 on the corresponding Ser172 site (pIKKε, Figure S2B), suggesting DCLK2 specifically phosphorylates TBK1. We also overexpressed DCLK2 in the ccRCC cell lines UMRC2 and UMRC6 and found that DCLK2 promotes TBK1 phosphorylation as well as p62 phosphorylation on Ser366, a reported TBK1 target in ccRCC¹⁰ (Figure S2C). We then examined how DCLK2 bound TBK1 and found that TBK1 could be recovered with both the DCX1 and DCX2 domains, but not the kinase domain (Figures S2D and S2E). We also examined pTBK1 and DCLK2 in a panel of RCC cell lines and found that RCC cell lines generally expressed higher levels of pTBK1 than normal renal epithelial cell lines such as RPTEC, HKC as well as 293T cells, and that this was accompanied by high levels of DCLK2 in A498 and 769-P cells (Figure S2F). Taken together these data suggest that DCLK2 serves as a kinase that phosphorylates TBK1 on Ser172 in ccRCC.

DCLK2 depletion reduces TBK1 phosphorylation and tumor cell proliferation in vitro and in vivo

Next, we examined the effect of DCLK2 siRNA on TBK1 phosphorylation in multiple ccRCC cell lines, including UMRC6 (original cell line used for siRNA screen), 769-P and A498. Consistently, DCLK2 depletion led to decreased TBK1 phosphorylation, which in turn likely accounts for decreased p62 phosphorylation on Ser366 (Figures 3A–3C). In several samples even p62 protein level is decreased, which could be because that Ser366 phosphorylation governs p62 protein stability ¹⁰. Next, we sought to determine whether the regulation of TBK1 phosphorylation is specific for DCLK2, and whether it applies to the closely related DCLK1 protein. In contrast to DCLK2, DCLK1 depletion using three different siRNAs did not consistently decrease pTBK1 (Figure S3A). Additionally, DCLK1 did not promote TBK1 phosphorylation using an in vitro kinase assay (Figure S3B). Therefore, DCLK1 does not appear to regulate TBK1 phosphorylation on Ser172.

Figure 3. Depletion of DCLK2 inhibits pTBK1 level as well as ccRCC growth both in vitro and in vivo.

A-D, Immunoblots and quantification of pTBK1/TBK1 (A-C), representative soft agar images and corresponding quantification of colony number (D) of indicated cells transfected with either Ctrl siRNA or DCLK2 siRNA. All quantification data were from three biological replicates.

E-H, Immunoblots and quantification of pTBK1/TBK1 (E-G), representative soft agar images and corresponding quantification of colony number (H) of indicated cells infected with lentivirus encoding either Ctrl sgRNA or DCLK2 sgRNA. All quantification data were from three biological replicates.

I-K, Immunoblots (I), representative soft agar images (J) and corresponding quantification of colony number (K) of A498 cells infected with lentivirus expressing indicated genes.

L, Immunoblots of lysates from A498-Luciferase stable cells infected with lentivirus encoding either Ctrl sgRNA or DCLK2 sgRNA (sg1) and then treated with doxycycline to induce FLAG-Cas9 expression.

M-O, Representative bioluminescence images (M), corresponding quantification data (N) and tumor weight (O) from kidney orthotopic tumors derived from A498-Luciferase stable cells expressing indicated genes. Two-way ANOVA analysis was performed for N.

P-R, Image of kidney orthotopic tumors derived from A498-Luciferase stable cells expressing indicated genes (P), immunoblots of representative tumor samples (Q) and quantification of pTBK1/TBK1 (R). All WB signal quantification was performed by Image J. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. n.s., no significance.

We then examined 3-D soft agar growth of ccRCC cells upon DCLK2 depletion. We found that DCLK2 knockdown led to decreased 3-D colony formation in all tested ccRCC cell lines (Figure 3D) indicating that DCLK2 is necessary for ccRCC proliferation under these conditions. Next, we expanded these results using a newly generated ccRCC cell line from a patient-derived tumorgraft (XP258) that displays DCLK2 protein expression⁴². Consistent with the results in established ccRCC cell lines, depletion of DCLK2 by siRNA in XP258 reduced pTBK1 and decreased colony formation in 3-D soft agar assay (Figures S3C–S3E). The effects of DCLK2 depletion on XP258 are less dramatic compared to other ccRCC cells tested. Since XP258 cells grow much faster than established ccRCC cell lines in soft agar, we postulated that there may be multiple growth-promoting signaling pathways that contribute to colony growth in XP258. The effect of DCLK2 depletion could be compensated by other growth-promoting signaling pathways, which remains to be investigated for our future research. Next, we used CRISPR-Cas9 to inactivate DCLK2 using several independent sgRNAs (sg1, sg2, sg3). Consistent with DCLK2 siRNA data, DCLK2 editing decreased pTBK1 and pS366 on p62 (Figures 3E–3G). In addition, DCLK2 editing decreased 3-D colony growth in these ccRCC cell lines (Figure 3H). To determine whether the effect of DCLK2 depletion on ccRCC growth is dependent on TBK1 Ser172 phosphorylation, we found that reduced colony formation by DCLK2 depletion was rescued by a constitutively active TBK1-S172D mutant (Figures 3I–3K). Our results suggest that TBK1 is the major effector downstream of DCLK2 involved in ccRCC cell proliferation and potentially tumorigenesis.

Next, we aimed to determine the effect of DCLK2 depletion on ccRCC tumorigenesis. For these experiments, we labeled ccRCC cell lines with firefly luciferase and then generated stable cell lines expressing either sgCtrl or sgDCLK2 (Figure S3F). We orthotopically injected these cells into the kidney capsule of NSG mice and monitored tumor growth over time through bioluminescence imaging. As shown, DCLK2 depletion led to decreased kidney tumor formation in xenograft models (Figures S3G–S3I). In order to prevent the potential compensatory signaling pathway induced by stable DCLK2 depletion, we also generated ccRCC cell lines expressing a doxycycline-inducible Cas9 and either sgCtrl or sgDCLK2. This system enabled the conditional depletion of DCLK2 (Figure 3L). As shown, DCLK2 depletion led to decreased kidney tumor formation, which corresponded with reduced tumor weight and tumor size (Figures 3M–3P). We also examined DCLK2 and pTBK1 in these tumors and found that the DCLK2-depleted tumor group displayed decreased DCLK2 and pTBK1 (Figures 3Q and 3R) suggesting that decreased tumor progression was due to DCLK2 depletion and decreased pTBK1. Cumulatively, our results suggest that DCLK2 is essential for kidney tumorigenesis.

Existence of multiple DCLK2 gene splice isoforms in ccRCC

To study the clinical relevance of DCLK2 in ccRCC patients, we first analyzed DCLK2 mRNA expression from the TCGA dataset using UALCAN^43,44. We found increased DCLK2 mRNA levels in 2 RCC subtypes (KIRC, clear cell RCC and KICH, chromophobe RCC) (Figure 4A). We then analyzed RNA-seq data from our own UT Southwestern RCC cohort^42,45, where we also found that DCLK2 mRNA was significantly increased in 2 RCC subtypes including ccRCC and tRCC (Figure 4B left panel). One limitation of studies in patient tumors is that they are composed of both tumor and stromal cells and thus the results are confounded. To determine whether DCLK2 mRNA levels were specifically induced in tumor cells, we analyzed tumorgrafts (TGs), patient tumors implanted in mice. In TG, the stroma is replaced by the host enabling thereby analyses specifically of tumor cells. Not only were increased DCLK2 mRNA levels observed in patient tumors, but also TGs, indicating that DCLK2 levels are induced in tumor cells (Figure 4B right panel).

Figure 4. Existence of multiple DCLK2 gene splice isoforms in ccRCC.

A, DCLK2 mRNA level in three RCC subtypes and corresponding normal tissues. DCLK2 mRNA level was determined by UALCAN. KIRC, kidney renal clear cell carcinoma. KIRP, kidney renal papillary cell carcinoma. KICH, kidney chromophobe.

B, DCLK2 mRNA level in three RCC subtypes and corresponding normal tissues. Data was from UTSW renal cancer cohort. T, tumor. TG, tumorgraft.

C, Normalized expression level of DCLK2 isoforms in 3 RCC subtypes. Data was from TCGA.

D, Normalized expression level of DCLK2 isoforms in A498 cells.

E, Normalized expression level of DCLK2 isoforms from normal kidney tissue (GTEx_kidney) and three RCC subtypes.

F, mRNA and protein structure of DCLK2²⁰¹ and DCLK2²⁰³. Red arrowhead indicates the stop codon in the additional exon in DCLK2²⁰³. Dashed lines indicate exons shared by both isoforms.

We noticed the existence of multiple DCLK2 gene splice isoforms (Ensembl Genome Browser, Release 109)⁴⁶. They share most exons but with slight differences. To determine if all DCLK2 isoforms express in RCC, we analyzed their particular expression in TCGA (KIRC, KIRP and KICH). Only DCLK2-201, 202 and 203 isoforms were detected (Figure 4C). In all three tumor types, the expression level of DCLK2-203 was higher than the other two isoforms (Figure 4C). We also analyzed the A498 cell line and found that DCLK2-201 and DCLK2-203 were the two dominant isoforms (Figure 4D). We then made a comparison between normal kidney tissues and RCC. In ccRCC, both DCLK2-201 and DCLK2-203 were increased compared to normal tissues and the increase of DCLK2-203 was more significant (Figure 4E). Taken together, these results suggest that DCLK2-201 and DCLK2-203 may be potential oncogenic drivers in ccRCC. Thus, in the following studies we focused on the DCLK2-201 and DCLK2-203 isoforms (DCLK2²⁰¹ and DCLK2²⁰³ hereafter).

DCLK2²⁰¹, the canonical isoform, contains 16 exons, and differs from DCLK2²⁰³ in that in the later (i) exon 1 has partial deletion; (ii) there is an additional exon inserted between exon 15 and 16; and (iii) a stop codon in the additional exon precludes the expression of exon 16 causing an early termination (Figure 4F). The protein product of DCLK2²⁰³ contains 694AA, having the same N-terminus as DCLK2²⁰¹ since the deletion of exon 1 is non-coding, but losing about 10% of the C-terminus compared to DCLK2²⁰¹ (766AA). (Figure 4F). This truncation retains the DCLK2 kinase domain as well as the two DCX domains. As the commercially available DCLK2 kinase used in the kinase assays in Figure 2 just contains 1-690AA, the common region of DCLK2²⁰¹ and DCLK2²⁰³, both isoforms are likely competent for TBK1 phosphorylation. Since all DCLK2 siRNAs and sgRNAs used in Figure 3 target both DCLK2²⁰¹ and DCLK2²⁰³, the isoform-specific function could not be distinctively defined. Therefore, we aimed to determine the function of DCLK2 isoforms in ccRCC.

DCLK2²⁰³ is essential and sufficient for TBK1 phosphorylation and ccRCC growth in vitro and in vivo

Next, we evaluated various DCLK2 antibodies to detect DCLK2 isoforms since the DCLK2 antibody used previously (i.e. Figure 3) was raised against the C-terminus of DCLK2²⁰¹ and cannot detect DCLK2²⁰³. Among several antibodies tested, an antibody generated with the DCLK2 N-terminus (106515) detected both DCLK2²⁰¹ and DCLK2²⁰³ isoforms with predicted molecular weight. We validated this antibody by using lysates derived from ccRCC cells transfected with DCLK2 siRNAs or infected with DCLK2 sgRNA vectors, and we found that both DCLK2²⁰¹ and DCLK2²⁰³ were reduced compared to controls (Figures S4A and S4B).

To study DCLK2²⁰¹ and DCLK2²⁰³ separately, we designed specific qPCR primers on the isoform specific regions as shown in Figure 5A. We also designed specific siRNAs targeting either DCLK2²⁰¹ or DCLK2²⁰³ isoform separately by siDirect2.1^47,48. By transfecting A498 cells, we found these siRNAs could knock down their targets respectively (Figure 5B), suggesting that these isoform-specific siRNAs and qPCR primers are effective and specific. We then examined siDCLK2 samples by RT-qPCR and confirmed depletion of both DCLK2²⁰¹ and DCLK2²⁰³ isoforms (Figure S4C).

Figure 5. DCLK2²⁰³ is essential and sufficient for TBK1 phosphorylation and ccRCC growth in vitro and in vivo.

A, Schematic model of specific siRNA targets and qPCR amplicon regions of DCLK2²⁰¹ and DCLK2²⁰³ isoforms.

B-D, RT-qPCR quantification (B), Immunoblots (C), representative soft agar images and corresponding quantification of colony number (D) of A498 cells transfected with Ctrl siRNA or siRNA targeting either DCLK2²⁰¹ or DCLK2²⁰³. The arrowhead in C indicates DCLK2²⁰³.

E-G, RT-qPCR quantification (E), Immunoblots (F), representative soft agar images and corresponding quantification of colony number (G) of A498 cells infected with lentivirus expressing Ctrl shRNA or shRNA targeting either DCLK2²⁰¹ or DCLK2²⁰³.

H-I, Representative bioluminescence images (H) and corresponding quantification data (I) for kidney orthotopic tumors derived from A498-Luciferase stable cells expressing indicated genes. Two-way ANOVA analysis was performed for I.

J-L, Immunoblots (J) and quantification of pTBK1/TBK1 (K), representative soft agar images and corresponding quantification of colony number (L) of A498 and 769-P cells infected with lentivirus expressing either EV, inducible HA-DCLK2²⁰¹ or HA-DCLK2²⁰³ and then treated with doxycycline.

M-P, Image (M), size quantification (N), immunoblots (O) and quantification of pTBK1/TBK1 (P) of tumors derived from A498 cells expressing indicated genes. Two-way ANOVA analysis was performed for N.

Q, Immunoblots of lysates from indicated 10 pairs of ccRCC patients normal (N) and tumor (T) tissues. The complete blots of Ab106515 were shown and arrowheads indicate DCLK2²⁰¹ (blue) and DCLK2²⁰³ (red) bands.

R, Quantification of DCLK2²⁰¹ and DCLK2²⁰³ protein level in paired ccRCC patients normal and tumor tissues (n=25).

S, Case number of different combinations from Figure 5Q and Figure S4L. The change of each protein is defined as comparing tumor sample with normal sample. All WB signal quantification was performed by Image J. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. n.s., no significance.

By transfecting these siRNAs into A498 and 769-P cells, we found that depletion of DCLK2²⁰¹ did not affect TBK1 phosphorylation while depletion of DCLK2²⁰³ led to decreased TBK1 phosphorylation (Figure 5C and Figure S4D). Functionally, while depletion of DCLK2²⁰¹ did not affect colony growth in soft agar, depletion of DCLK2²⁰³ inhibited 3-D colony growth significantly (Figure 5D and Figure S4E). These results suggest that DCLK2²⁰³ is the critical isoform required for TBK1 phosphorylation and ccRCC cell growth. We then made shRNAs for DCLK2²⁰¹ and DCLK2²⁰³ isoforms by using the same targeting sequences as siRNAs specified above (siDCLK2²⁰¹-1 and siDCLK2²⁰³-1). A498 cells expressing these shRNAs showed the same phenotype as siRNAs, where depletion of DCLK2²⁰³, but not DCLK2²⁰¹, led to decreased TBK1 phosphorylation and decreased ccRCC colony growth (Figures 5E–5G). Of note, the reduced TBK1 phosphorylation and growth defect induced by shDCLK2²⁰³ can be rescued by re-expressing an shRNA-resistant DCLK2²⁰³ (Figures S4F and S4G), suggesting the phenotype of shDCLK2²⁰³ is on-target. Motivated by the results of cell-based assay, we further investigated the effect of either isoform depletion in vivo by using an orthotopic kidney tumor model. Our result suggests that depletion of DCLK2²⁰³, but not DCLK2²⁰¹, suppressed kidney tumor progression significantly (Figures 5H and 5I).

We then explored the potential gain-of-function effect of DCLK2²⁰¹ and DCLK2²⁰³ isoforms in ccRCC. By overexpressing both isoforms in A498 and 769-P cells, we found that DCLK2²⁰³, but not DCLK2²⁰¹, induced TBK1 phosphorylation (Figures 5J and 5K). Consistent with pTBK1 regulation, DCLK2²⁰³ promoted colony formation to a greater extent than DCLK2²⁰¹ (Figure 5L). Motivated by these findings, we injected these cells subcutaneously into NSG mice and observed that DCLK2²⁰³ promoted tumor growth (Figures 5M and 5N). Upon necropsy, we also observed increased pTBK1 in the DCLK2²⁰³ group (Figures 5O and 5P), suggesting that pTBK1 induction by DCLK2²⁰³ may be responsible for tumor growth in these mice. To examine whether this regulation was dependent on VHL, we overexpressed DCLK2²⁰¹ and DCLK2²⁰³ in A498 cells restored with VHL. We found that DCLK2²⁰³ also promoted tumor growth in this context (Figures S4H–S4K), suggesting that DCLK2²⁰³ can promote ccRCC tumor growth in a VHL-independent manner.

To evaluate DCLK2²⁰¹ and DCLK2²⁰³ in ccRCC tumors from patients, we examined 25 pairs of tumor and adjacent normal tissues. As shown, increased protein levels of both DCLK2²⁰¹ and DCLK2²⁰³ were observed in a significant portion of ccRCC (Figures 5Q, 5R and S4L). The extent of increased DCLK2²⁰³ protein level is more significant compared to DCLK2²⁰¹ in these patients by using normal tissues as isogenic controls (Figure 5R). To investigate the potential connection between DCLK2²⁰³ and TBK1 phosphorylation, we also examined pTBK1 and TBK1 levels in these samples. 15 out of 25 tumors showed increased DCLK2²⁰³ level compared to normal tissues. Among them, 12 tumors (12/15, 80% patients) also displayed increased pTBK1 (Figure 5S). These results suggest that elevated DCLK2²⁰³ may promote TBK1 phosphorylation in ccRCC.

Until now, the upstream signaling that drives DCLK2 expression in ccRCC remains unknown. We found several transcription regulators (ZFX, CTCF, EP300 and ZMYND8) which are predicted to regulate DCLK2 expression in ccRCC or normal kidney epithelial cells by Cistrome Data Browser via analyzing ChIP-seq data (Figure S5A, upper panel)⁴⁹. As a control, HIF2α (EPAS1) doesn’t display an obvious binding peak on DCLK2 gene promoter region. ZHX2 is a transcription factor which is upregulated in ccRCC upon VHL loss reported by our lab previously⁵⁰. We noticed that ZHX2 displays binding peak on DCLK2 gene promoter as well (Figure S5A, lower panel). We then knocked down these five genes individually in both A498 and 769-P cells and found that depletion of ZHX2 and ZFX, but not others, led to decreased DCLK2 expression (Figures S5B–S5K). ZHX2 protein level is increased in ccRCC due to VHL loss. The function of ZFX in ccRCC is currently not well studied except for one report suggesting that ZFX knockdown could suppress RCC growth and induce apoptosis⁵¹. Taken together, increased DCLK2 expression in ccRCC could be at least partially driven by ZHX2. Whether ZFX regulates ccRCC growth through DCLK2 needs further investigation.

Based on Ensemble Gene Annotation, DCLK2²⁰³ transcript is an alternative splice variant predicted to undergo nonsense mediated decay (NMD), a mechanism that cells exploit to degrade mRNA that contains premature termination codon (PTC)^52,53. It is important to consider that NMD is not an “all-or-none” phenomenon. A transcript can be regulated by NMD and still be expressed at functional levels in a given cell type⁵⁴. To test if DCLK2²⁰³ isoform is regulated by NMD as predicted, we knocked out the expression of UPF1, a factor essential for NMD, by CRISPR/Cas9 in A498 and 769-P cells and then examined mRNA level of DCLK2 isoforms. Two SRSF2 splice variants were used as positive controls here: one variant with PTC (SRSF2 PTC+) was reported to be regulated by the NMD pathway but another variant without PTC (SRSF2 PTC−) was not regulated by NMD^55,56. Upon UPF1 depletion, mRNA level of DCLK2²⁰¹ remained the same but mRNA level of DCLK2²⁰³ was increased (Figures S6A and S6B). We then knocked out another NMD factor SMG6 in A498 and 769-P cells and got similar results (Figures S6C and S6D). These results suggest that DCLK2²⁰³ is indeed regulated by the NMD pathway. Recently the NMD pathway has been found to be a critical modulator in tumorigenesis^57,58. Interestingly, we noticed that both UPF1 mRNA and protein level was decreased in ccRCC compared to normal tissue (Figure S6E). And UPF1 depletion in normal kidney epithelial cells HKC and HK-2 could selectively promote DCLK2²⁰³ expression (Figures S6F and S6G). Taken together, we propose that decreased UPF1 expression may partially explain the high level of DCLK2²⁰³ in ccRCC.

DCLK2²⁰³ regulates ccRCC tumorigenesis by controlling TBK1 phosphorylation

To examine whether the kinase activity of DCLK2²⁰³ is important for TBK1 phosphorylation, we made a catalytic dead mutant of DCLK2²⁰³. We mutated a conserved lysine residue critical for the phosphotransfer reaction (K423A)^59,60 and found that WT DCLK2²⁰³, but not the K423A mutant, promoted TBK1 phosphorylation in an in vitro kinase assay (Figure 6A). Expressed in ccRCC cell lines, whereas WT DCLK2²⁰³ promoted pTBK1, the K423A mutant failed to do so (Figures 6B and 6C). Finally, the K423A mutant failed to efficiently promote anchorage independent growth or xenograft tumor growth (Figures 6D–6H).

Figure 6. DCLK2²⁰³ regulates ccRCC tumorigenesis by controlling TBK1 phosphorylation.

A, immunoblots of in vitro kinase assay samples as indicated.

B-C, Immunoblots (B) and quantification of pTBK1/TBK1 (C) of A498 and 769-P cells infected with lentivirus expressing either EV, inducible HA-DCLK2²⁰³-WT or HA-DCLK2²⁰³-K423A and then treated with doxycycline. WB signal quantification was performed by Image J.

D-E, Representative soft agar images (D) and corresponding colony number quantification (E) of A498 and 769-P cells infected with lentivirus expressing indicated genes.

F-H, Image (F), size quantification (G) and weight quantification (H) of tumors derived from A498 cells expressing indicated genes. Two-way ANOVA analysis was performed for G.

I, Immunoblots of lysates from indicated A498 stable cell lines transfected with either Ctrl siRNA or TBK1 siRNA and then treated with doxycycline.

J-M, Immunoblots (J and L), representative soft agar images and corresponding quantification of colony number (K and M) of A498 cells infected with lentivirus expressing indicated genes and then treated with doxycycline. Arrowhead in J indicates endogenous DCLk2²⁰³.

N-P, Immunoblots (N), representative soft agar images (O) and corresponding quantification (P) of A498 and 769-P cells infected with lentivirus expressing indicated genes and then treated with doxycycline.

Q-R, Image (Q) and size quantification (R) of tumors derived from A498 cells expressing indicated genes. Two-way ANOVA analysis was performed for R.

S, Sequence alignment between DCLK1, DCLK2²⁰¹ and DCLK2²⁰³. The conserved Threonine site (T693 on DCLK2²⁰¹) is highlighted by green.

T-W, Immunoblots (T and U), representative soft agar images (V) and corresponding quantification of colony number (W) of A498 and 769-P cells infected with lentivirus expressing either EV, inducible HA-DCLK2²⁰¹, HA-DCLK2²⁰¹-T693A or HA-DCLK2²⁰³ and then treated with doxycycline.

X, Schematic model of the DCLK2-TBK1 signaling in ccRCC. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. n.s., no significance.

To determine whether DCLK2²⁰³ induced p62 phosphorylation on Ser366 (pS366) via its regulation on TBK1, we overexpressed DCLK2²⁰³ while simultaneously depleting TBK1 with siRNA. Consistent with our previous study¹⁰, TBK1 depletion decreased pS366 and DCLK2²⁰³ failed to promote pS366 in the absence of TBK1 (Figure 6I), arguing that DCLK2²⁰³-induced p62 phosphorylation depends on TBK1 activity.

To examine whether the effect of DCLK2 depletion on 3-D colony growth can be rescued by DCLK2²⁰¹ or DCLK2²⁰³, we expressed them in DCLK2-depleted ccRCC cells. While wild-type DCLK2²⁰³ fully rescued TBK1 phosphorylation and colony growth, the DCLK2²⁰³ K423A mutant failed to do so (Figures 6J and 6K). Additionally, re-expression of DCLK2²⁰¹ did not rescue TBK1 phosphorylation and colony growth efficiently in DCLK2-depleted cells (Figures 6L and 6M). Taken together, these results indicate that DCLK2²⁰³ drives the oncogenic phenotype in ccRCC cells in a catalytically-dependent manner.

To examine whether the effect of DCLK2²⁰³ on kidney cancer growth is mediated through TBK1, we generated cell lines expressing a doxycycline inducible TBK1 shRNA. Our results showed that DCLK2²⁰³ effects on cell and tumor growth were abrogated upon TBK1 depletion (Figures 6N–6R). Previous research showed that the DCLK1 C-terminal residue Threonine 688 restricted its activity by preventing aberrant hyperphosphorylation within its microtubule-binding domain⁶¹. Interestingly, this residue is present in DCLK2²⁰¹ (Threonine 693) but not in DCLK2²⁰³ (Figure 6S), which may explain why DCLK2²⁰³ possesses higher activity compared to DCLK2²⁰¹. To test this possibility, we mutated Threonine 693 to Alanine on DCLK2²⁰¹ and measured the activity of T693A on TBK1 phosphorylation and soft agar colony formation. We found that the T693A mutation potentiated DCLK2²⁰¹ ability to phosphorylate TBK1, almost to the same extent as the DCLK2²⁰³ isoform (Figures 6T and 6U), which also corresponded with increased cell proliferation on 3-D soft agar assay (Figures 6V and 6W). We conclude that the greater activity of DCLK2²⁰³ over DCLK2²⁰¹ in promoting TBK1-mediated tumorigenesis is due, at least in part, to the absence of an auto-inhibitory C-terminal residue that dampens its activity.

Discussion

In this study, we identified DCLK2 as a TBK1 kinase that promotes oncogenesis in kidney cancer. Specifically, we established that the DCLK2²⁰³ isoform, which is the predominant isoform in ccRCC, is responsible for TBK1 phosphorylation on Ser172 with consequent activation (Figure 6X). This process is due, at least in part, to the absence of a C-terminal autoinhibitory threonine residue (T693) in DCLK2²⁰³ compared to other DCLK2 isoforms. Our study identifies a mechanism of TBK1 regulation in kidney cancer opening up an opportunity for selective therapeutic intervention.

Our kinome-wide siRNA screen identified an unexpected kinase, DCLK2, as a critical regulator of TBK1 in kidney cancer cells. Although some studies have probed the role of TBK1 in cancer^62,63, underlying mechanisms by which TBK1 is activated in cancer cells remain largely unknown. Here we show that DCLK2 directly phosphorylates TBK1 on Ser172 in the activation loop (Figures 2D–2I). This phosphorylation event seems very specific for both DCLK2 and TBK1. DCLK1 and DCLK2 belong to the same family and display high similarity in protein structure. However, DCLK1 did not promote TBK1 activity in vitro and in kidney cancer cells (Figures S3A–S3B). In the other hand, IKKε and TBK1 are highly conserved on the residues surrounding Ser172. However, DCLK2 does not phosphorylate IKKε on its corresponding serine site (Figure S2B). This specific regulatory mechanism may help cells achieve precise regulation on highly conserved TBK1 and IKKε in response to different physiological conditions. By targeting DCLK2, we may avoid the issue of TBK1 inhibitors being cross reactive with IKKε. Since TBK1 is a central player in innate immune signaling, whether DCLK2 may affect innate immunity through TBK1 should be considered. According to the established model¹⁴, phosphorylation of IRF3 by TBK1 requires STING dimerization induced by the second messenger cGAMP. Therefore, DCLK2-induced TBK1 activation would not activate IRF3 signaling under this condition in ccRCC. This is consistent with our recent publication showing that p-IRF3 was not active despite high TBK1 activity in ccRCC¹⁰. Therefore, we may selectively inhibit TBK1 activity in cancer by targeting DCLK2.

The role of DCLK2 in cancer was poorly understood, and this study for the first time implicates DCLK2 in kidney cancer. We noticed that DCLK2²⁰³ is expressed predominantly in multiple cancer types (Figure S7), suggesting that it may play oncogenic roles in a broad range of human cancers. It is not clear whether DCLK2²⁰³ can promote TBK1 activity in other cancers besides ccRCC. The detailed function of DCLK2 in human cancers remains to be fully investigated. We have identified that the expression of DCLK2²⁰³ isoform is regulated by the NMD pathway (Figure S6). However, how DCLK2²⁰³ isoform is generated remains an outstanding question. The truncation on the first exon in DCLK2²⁰³ suggests that it may have different transcription initiation or mRNA processing compared to DCLK2²⁰¹(Figure 4F). The differential expression of DCLK2²⁰¹ and DCLK2²⁰³ in ccRCC also implies potential distinct post-translational regulatory mechanisms (Figures 5Q and S4L).

The NMD pathway has been found to be a critical modulator of tumorigenesis^57,58. However, how NMD pathway regulates kidney cancer progression remains largely unknown. Our results suggest that reduced NMD pathway activity may promote ccRCC by selectively upregulating DCLK2²⁰³ expression. It remained to be determined on whether there are other specific gene isoforms regulated by NMD pathway contributing to ccRCC.

Future studies will focus on leveraging these discoveries for therapeutic application. While a selective inhibitor of DCLK1 has been previously reported⁶⁴, there are currently no selective DCLK2 inhibitors. There is another inhibitor called LRRK2-IN-1 that displayed potent activity against LRRK2, DCLK2, MAPK7 and PLK4 in nano molar range⁶⁵. In preliminary experiments, we found that LRRK2-IN-1 decreases TBK1 phosphorylation and inhibits proliferation of ccRCC cells, with or without LRRK2. Accordingly, LRRK2-IN-1 may serve as a promising lead for the development of more selective DCLK2 inhibitors. Further experiments will be required to characterize the potential therapeutic role of LRRK2-IN-1 in kidney cancer.

Limitations of the study

Our study does carry a few limitations. For example, we are still in need of high-quality antibodies that can detect both DCLK2²⁰¹ and DCLK2²⁰³ isoforms robustly in cells and in tumor tissues. In addition, the physiological role of DCLK2²⁰³ in kidney cancer needs to be investigated with some tissue specific transgenic mice. Lastly, the therapeutic potential of targeting DCLK2 in kidney cancer remains to be investigated, preferably with selective DCLK2 inhibitors that are currently not available.

STAR METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Qing Zhang (qing.zhang@utsouthwestern.edu).

Materials availability

All materials generated in this study are available from the Lead Contact with a completed Materials Transfer Agreement.

Data and code availability

Original western blot and soft agar colony images have been deposited at Mendeley and are publicly available as of the date of publication. Mass spectrometry data are presented in Table S1.

This paper does not report original code.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental Model And Study Participant Details

Cell lines

UMRC6, A498, 786-O, UMRC2, UMRC6, HEK293T, HKC and HK-2 cells were cultured in DMEM medium (Gibco) containing 10% fetal bovine serum (FBS, Gibco), 1% penicillin/streptomycin (Gibco). 769-P cells were cultured in RPMI-1640 medium (Gibco) containing 10% FBS and 1% penicillin/streptomycin. All cells were cultured at 37 °C with 5% CO₂. All cells used for experiments were within 10-20 passages from thawing. Cells were mycoplasma-free as determined by a mycoplasma detection kit (Lonza, LT07–218).

Orthotopic kidney tumor xenograft

Eight-week-old NSG mice were used for orthotopic xenograft studies. Approximately 5x10⁵ viable A498-Luciferase stable cells expressing sgCtrl/sgDCLK2 or shCtrl/shDCLK2²⁰¹/shDCLK2²⁰³ were resuspended in 20 μL PBS containing 50% matrigel (Corning 354234) and injected orthotopically into the left kidney of each mouse as described previously (8 male and 4 female were injected for each group)⁵⁰. For tumor cells with doxycycline inducible vectors, after injection and following bioluminescence imaging to make sure tumor successfully implanted in kidney, mice were then fed with Purina rodent chow #5001 with 2000ppm doxycycline (Research Diets Inc.) to induce target gene expression. Tumor growth was monitored by bioluminescence imaging once a week with SPECTRAL AMI-HTX imaging system. After the indicated number of weeks, mice were euthanized. Tumors were taken out and weighed, the weight of tumor was calculated by subtracting the weight of right normal kidney from the weight of left tumor kidney. All animal experiments were following NIH guidelines and were approved by the Institutional Animal Care and Use Committee (IACUC) of UT Southwestern Medical Center.

Subcutaneous tumor xenograft

To generate subcutaneous xenograft, approximately six-week-old male NSG mice were injected subcutaneously with 1.5x10⁶ A498 cells (resuspended in 100 μL PBS containing 50% matrigel) in the left flank. For tumor cells with doxycycline inducible vectors, one week after injection, mice were fed with Purina rodent chow #5001 with 2000ppm doxycycline (Research Diets Inc.) to induce target gene expression. Tumor size was measured with calipers once a week and tumor volume was calculated as V = 0.5 x L x S². L and S indicate long measurement and short measurement. After the indicated number of weeks, mice were euthanized, and tumors were taken out and weighed. All animal experiments were following NIH guidelines and were approved by the Institutional Animal Care and Use Committee (IACUC) of UT Southwestern Medical Center.

METHOD DETAILS

Kinome-wide siRNA screen

Silencer^™ Select human kinase siRNA library V4 was purchased from Thermo Fisher Scientific. ddH₂O was used to dissolve siRNA to reach final concentration of 10μmol/L. Before screen, siRNA mixture for each target was prepared by mixing 3 independent siRNA with the ratio of 1:1:1. For siRNA screen, UMRC6 cells were seeded in 12-well plate as 8x10⁴ cells per well. The next day, every 11 siRNA mixtures along with a non-target control siRNA (10pmol siRNA for each well) were transfected by Lipofectamine RNAi Max (Invitrogen) in each 12-well plate according to instruction. 16hr post-transfection, fresh medium was added to replace old medium. 60hr posttransfection, cells were harvested and proteins were extracted by EBC lysis buffer (50mM Tris-HCl pH8.0, 120mM NaCl, 0.5% NP40, 0.1mM EDTA and 10% glycerol). Protein samples were resolved by SDS-PAGE and Tubulin, TBK1 and pTBK1 were detected by Western Blot. Quantification of western blot signal was performed by Image J.

To validate selected targets, UMRC6 cells were seeded in 12-well plate as 8x10⁴ cells per well. For each target, the 3 independent siRNAs and one mixed siRNA plus a non-target control siRNA (10pmol siRNA for each group) were transfected. 60hr post-transfection, cells were harvested and proteins were extracted by EBC lysis buffer. Western blot was performed to detect Tubulin, TBK1 and pTBK1. The sequence of all individual siRNA used in this study can be found in Table S2.

In vitro kinase assay

To perform in vitro kinase assay, 3μL in vitro translated TBK1 product (generated with T7 Quick Coupled Transcription/Translation System, Promega L1170) was mixed with 25 μL kinase reaction buffer (20 mmol/L Tris-HCl pH 7.4, 500 mmol/L β-glycerol phosphate, 12 mmol/L magnesium acetate), 1 μL 10mmol/L ATP and 1 μL purified recombinant kinase (SignalChem) to reach a 30 μL reaction system. Reaction system was then incubated at 30 °C, 600rpm for 1 hour. Then 7.5 μL 5x SDS loading buffer was added to stop reaction. After boiling at 95 °C, Western Blot was performed to detect TBK1, pTBK1 and the kinase proteins. Kinases used in this study can be found in key resources table.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit anti-TBK1	Cell Signaling Technology	Cat# 3504; RRID: AB_2255663
Rabbit anti-phospho-TBK1(Ser172)	Cell Signaling Technology	Cat# 5483; RRID:AB_10693472
Rabbit anti- HIF2α	Cell Signaling Technology	Cat# 7096; AB_10898028
Rabbit anti-p62	Cell Signaling Technology	Cat# 39749; RRID:AB_2799160
Rabbit anti-DCLK1	Cell Signaling Technology	Cat# 62257; RRID:AB_2799622
Rabbit anti-VHL	Cell Signaling Technology	Cat# 68547; RRID:AB_2716279
Rabbit anti-FLAG-Tag	Cell Signaling Technology	Cat# 14793; RRID:AB_2572291
Rabbit anti-HA-Tag	Cell Signaling Technology	Cat# 3724; RRID:AB_1549585
Rabbit anti-V5-Tag	Cell Signaling Technology	Cat# 13202 RRID:AB_2687461
Rabbit anti-GST	Cell Signaling Technology	Cat# 2625; RRID:AB_490796
Mouse anti-α Tubulin	Cell Signaling Technology	Cat# 3873; RRID:AB_1904178
HRP-linked anti-rabbit IgG	Cell Signaling Technology	Cat# 7074; RRID:AB_2099233
HRP-linked anti-mouse IgG	Cell Signaling Technology	Cat# 7076; RRID:AB_330924
Rabbit anti-DCLK2	Abcam	Cat# ab106639; RRID:AB_10887397
Rabbit anti-DCLK2	Thermo Fisher Scientific	Cat# PA5-82538; RRID:AB_2789696
Rabbit anti-DCLK2	Thermo Fisher Scientific	Cat# PA5-106515; RRID:AB_2854184
Mouse anti-Vinculin	Sigma-Aldrich	Cat# V9131; RRID:AB_477629
Rabbit anti-phospho-p62 (Ser366)	Affinity Biosciences	Cat# AF7374; RRID:AB_2843814
Rabbit anti-STK3	Abclonal	Cat# A9036; RRID:AB_2863645
Mouse anti-ZFX	Cell Signaling Technology	Cat# 5419; RRID:AB_10705453
Rabbit anti-CTCF	Cell Signaling Technology	Cat# 2899; RRID:AB_2086794
Rabbit anti-ZHX2	GeneTex	Cat# GTX112232; RRID:AB_1952653
Rabbit anti-ZMYND8	Bethyl Laboratories	Cat# A302-089A; RRID: 1604282
Bacterial and virus strains
MAX Efficiency™ DH5α Competent Cells	Thermo Fisher Scientific	Cat# 18258012
One Shot™ Stbl3™ Chemically Competent E. coli	Thermo Fisher Scientific	Cat# C737303
Biological samples
Human ccRCC paired tissues	UTSW Tissue Management Shared Resource Center	N/A
Chemicals, peptides, and recombinant proteins
Puromycin	GIBCO	Cat# A11138-03
G418 Sulfate	Thermo Fisher Scientific	Cat# 10-131-027
Lipofectamine 3000 transfection reagent	Thermo Fisher Scientific	Cat # L3000015
RNAiMAX transfection reagent	Thermo Fisher Scientific	Cat# 13778150
Polybrene	Santa Cruz Biotechnology	Cat# sc-134220
Doxycycline	Sigma-Aldrich	Cat# D9891
D-Luciferin-potassium salt	Goldbio	Cat# LUCK-1G
Iodonitrotetrazolium chloride	Sigma-Aldrich	Cat# I8377
Agarose	Fisher Scientific	Cat# BP165-25
Recombinant human DCLK2	SignalChem	Cat# D15-11G-10
Recombinant human DCLK1	SignalChem	Cat# D14-10G-10
Recombinant human ADCK5	SignalChem	Cat# A13-11G-20
Recombinant human CAMK1G	SignalChem	Cat# C10-10BG-05
Recombinant human MAPKAPK3	SignalChem	Cat# M41-10G-05
Recombinant human SRPK3	SignalChem	Cat# S41-10G-05
Recombinant human TWF2	SignalChem	Cat# P98-30G-20
Recombinant human WNK2	SignalChem	Cat# W03-11G-05
Recombinant human STK3	SignalChem	Cat# S24-10G-05
Critical commercial assays
Silencer™ Select human kinase siRNA library V4	Thermo Fisher Scientific	Cat# 4397918
RNeasy mini kit	Qiagen	Cat# 74106
iScript™ cDNA Synthesis Kit	Bio-Rad	Cat# 1708891
KOD-Plus Mutagenesis Kit	Toyobo	Cat# SMK-101
T7 Quick Coupled Transcription/Translation System	Promega	L1170
Experimental models: Cell lines
A498	ATCC	Cat# HTB-44
769-P	ATCC	Cat# CRL-1933
UMRC6	Sigma-Aldrich	Cat#08090513
UMRC2	Sigma-Aldrich	Cat#08090511
786-O	ATCC	Cat#CRL-1932
293T	ATCC	Cat# CRL-3216
XP258	J. Brugarolas Lab	N/A
HKC	W. K. Rathmell Lab	N/A
HK-2	P. Ly Lab	N/A
Experimental models: Organisms/strains
NOD SCID Gamma mice	UTSW Animal Resource Center (ARC)	N/A
Oligonucleotides
DCLK2 sgRNA-1: TAGACTCAGGAGTCGTCAAG	This paper	N/A
DCLK2 sgRNA-2: TGATATTGGGATGTTTCACT	This paper	N/A
DCLK2 sgRNA-3: TCTTCTTCCAATGTAAACGG	This paper	N/A
UPF1 sgRNA: TCAACTGGGACAGCTCGCAG	Zhu et al.56	N/A
SMG6 sgRNA: GCCCGTATAGGGATACTGGG	Zhu et al.56	N/A
ZMYND8 sgRNA: GGAGCGCGGCATATCCGACA	Chen et al.66	N/A
All siRNA targeting sequence, see Table S2	This paper	N/A
All shRNA targeting sequence, see Table S3	This paper	N/A
All RT-qPCR primers sequence, see Table S4	This paper	N/A
Recombinant DNA
pLX304-DCLK2-V5	Horizon	Cat# OHS6271-213587493
pcDNA-HA-TBK1	This paper	N/A
pcDNA-HA-TBK1-D157A	This paper	N/A
pcDNA-FLAG-IKKε	This paper	N/A
pLenti6-FLAG-TBK1	This paper	N/A
pLenti6-FLAG-TBK1-S172D	This paper	N/A
pInducer-DCLK2203-WT	This paper	N/A
pInducer-DCLK2203-K423A	This paper	N/A
pInducer-DCLK2201-WT	This paper	N/A
pInducer-DCLK2201-T693A	This paper	N/A
pLenti6-FLAG-DCLK2	This paper	N/A
pLenti6-FLAG-DCLK2-F1	This paper	N/A
pLenti6-FLAG-DCLK2-F2	This paper	N/A
pLenti6-FLAG-DCLK2-F3	This paper	N/A
pInducer20 EV	Addgene	Cat# 44012
Software and algorithms
Fiji, ImageJ	Schneider et al.67	https://imagej.net/Downloads
GraphPad Prism 9	GraphPad	https://www.graphpad.com/scientific-software/prism/
MaxQuant version 2.3.0.0	Max Planck Institute	https://www.maxquant.org/
Other
Purina rodent chow doxycycline	Research Diets	Cat# C11300-2000i
Tet system approved FBS	Takara Bio	Cat# 631105
Matrigel	Corning	Cat# 354234
cOmplete™, EDTA-free Protease Inhibitor Cocktail	Millipore-Sigma	Cat# 11873580001
PhosSTOP Phosphatase inhibitor	Millipore-Sigma	Cat# 4906837001
Anti-FLAG M2 Affinity Gel	Millipore-Sigma	Cat# A2220
Bradford protein assay dye	Bio-Rad	Cat# 5000006
Gateway™ BP Clonase™ II Enzyme mix	Invitrogen	Cat# 11789020
Gateway™ LR Clonase™ II Enzyme mix	Invitrogen	Cat# 11791100

Analyzing TBK1 phosphorylation site by mass spectrum

Sample preparation.

TBK1 with or without DCLK2 in 50 mM Tris (pH 8.0) containing 6.4 M urea was reduced and alkylated followed dilution with 25 mM Tris to make final 2 M urea and tryptic digestion. The peptides were desalted on home-made C18 stage-tips and dried in speed-vac.

Mass spectrometry analysis.

The clean peptides were dissolved in 0.1% formic acid and analyzed on a Q-Exactive HF-X coupled with an Easy nanoLC 1200 (Thermo Fisher Scientific, San Jose, CA). Peptides were loaded on to a nanoEase MZ HSS T3 Column (100Å, 1.8 μm, 75 μm x 250 mm, Waters). Analytical separation of peptides was achieved with 50-min gradient. A linear gradient of 5 to 30% buffer B over 29 min and 30% to 40% buffer B over 6 min was executed at a 250 nl/min flow rate followed a ramp to 100%B in 1 min and 14-min wash with 100%B, where buffer A was aqueous 0.1% formic acid, and buffer B was 80% acetonitrile and 0.1% formic acid.

LC-MS experiments were performed in a data-dependent mode with full MS with a resolution of 60,000 at m/z 200 followed by high energy collision-activated dissociation-MS/MS of the top 15 most intense ions with a resolution of 15,000 at m/z 200. High energy collision-activated dissociation-MS/MS was used to dissociate peptides at a normalized collision energy of 27 eV in the presence of nitrogen bath gas atoms. Dynamic exclusion was 20 seconds. There were three technical LC-MS replicates for each sample.

Raw data processing and analysis.

Mass spectra were processed, and peptide identification was performed using the MaxQuant software version 2.3.0.0 (Max Planck Institute, Germany). The database searches were performed against TBK1 sequence (Q9UHD2, UniProt human protein sequence database UP000005640). A false discovery rate (FDR) for both peptide-spectrum match (PSM) and protein assignment was set at 1%. Search parameters included up to two missed cleavages at Lys/Arg on the sequence, phosphorylation of Ser, Thr and Tyr, oxidation of methionine, and protein N-terminal acetylation as a dynamic modification. Carbamidomethylation of cysteine residues was considered as a static modification. Peptide identifications are reported by filtering of reverse and contaminant entries and assigning to their leading razor protein.

Immunoprecipitation and western blot

For immunoprecipitation, cells were lysed in EBC buffer supplemented with proteinase and phosphatase inhibitor (Roche Applied Bioscience). The lysates were clarified by centrifugation and then lysates containing equal amount of proteins were incubated with anti-FLAG antibody conjugated beads (Sigma A2220) for 6-8hr at 4°C. After incubation, beads were washed with ice-cold EBC buffer for three times and bound protein complexes were eluted with SDS loading buffer and then subject for western blot analysis. For direct western blot, proteins of whole cell lysate were extracted by EBC buffer (supplemented with proteinase and phosphatase inhibitor). Protein concentration was measured by Bradford protein assay dye (Bio-Rad). Equal amount of proteins were resolved by SDS-PAGE for western blot analysis. Chemiluminescence signal was detected by ChemiDoc^™ Touch Imaging System (Bio-Rad).

Real-time qPCR (RT-qPCR)

Total RNA was extracted by RNeasy Mini Kit (Qiagen), followed by cDNA synthesis using iScript^™ cDNA Synthesis Kit (Bio-Rad). RT-qPCR was performed in a CFX384 Real-Time PCR System (Bio-Rad). Relative expression level of target genes was calculated using the 2^−ΔΔCt method. The sequence of RT-qPCR primers can be found in Table S4.

DCLK2 overexpression plasmids

A pLX304-DCLK2-V5 plasmid was purchased from Horizon, which encodes DCLK2²⁰³. To get the CDS of DCLK2²⁰¹, an N-terminal fragment was cloned by PCR from pLX304-DCLK2²⁰³-V5 (Fw primer: AACCACCGGTGCCAGCACCAGGAGTATCGAGCTGGAG; Rv primer: CATGATGACGGAGACCCCGGTGG). A C-terminal fragment was cloned by PCR from cDNA library of A498 cells (Fw primer: ATGCGCTCCCCAAACAGAACAG; Rv primer: TCTGCAGGAGGCTCAGTCTC). A third PCR was performed to generate full length DCLK2²⁰¹ using both N- and C-fragments as templates. Finally, Q353 was inserted to make full length DCLK2²⁰¹ complete by site-directed mutagenesis kit (Toyobo SMK-101, Fw primer: CAGATTTCTGCTCATGGCAGATCTTCTTC; Rv primer: CTTTAATCCTCTGAAACTTCCTGGAC).

To generate doxycycline inducible DCLK2 expression vectors, DCLK2 CDS was first cloned into pDONR223 vector through BP reaction system (Invitrogen 11789020), and then cloned into pInducer2.0 vector through LR reaction system (Invitrogen 11791020). Site-directed mutagenesis kit (Toyobo) was used to generate all DCLK2 mutation forms in this study.

Anchorage independent 3-D soft agar growth assay

UMRC6 and A498 cells were plated at a density of 1.5x10⁴/mL in DMEM medium containing 0.4% agarose (Fisher Scientific, BP165-25) onto bottom layer composed of medium with 1% agarose. 769-P cells were 2.5x10⁴/mL in RPMI-1640 medium containing 0.4% agarose. All soft agar assays were performed in 12-well plates. Every 3 days, fresh medium was added onto each well. After 3-4 weeks incubation, cell colonies were stained with medium containing 100mg/mL iodonitrotetrazoliuim chloride (Sigma, I8377-1G) for 2-3 days in incubator and then foci number was counted.

Virus production and infection

HEK293T cells were used for generating lentivirus particles. Virus packaging plasmids were transfected with lipofectamine 3000 (Invitrogen). 8-12hr post transfection, fresh medium was added to replace old medium. Then, viruses were collected twice at 24hr and 48hr post-changing medium. After filtering through 0.45μm filters, appropriate amount of viruses was used to infect target cells in the presence of 8 μg/mL polybrene (Santa Cruz). Subsequently, target cell lines were cultured in the presence of puromycin (2 μg/ml) or neomycin (0.8mg/ml) depending on the vectors.

Quantification and statistical analysis

Unless indicated, the unpaired two-tailed Student t test was used for experiments comparing two sets of data. All graphs depict mean ± SEM unless otherwise indicated. Graphs were generated by GraphPad Prism. *, **, *** and **** denote P value of <0.05, 0.01, 0.001 and 0.0001 respectively; n.s. denotes no significance.

Supplementary Material

2

Table S1. Mass spectrometry data for identifying TBK1 phosphorylation site (Related to Figure 2)

Highlights:

1. An unbiased kinome-wide siRNA screen identifies DCLK2 as a TBK1 activator in ccRCC

2. Depletion of DCLK2 inhibits ccRCC growth both in vitro and in vivo

3. DCLK2 gene produces multiple isoforms, among which DCLK2²⁰³ predominates in ccRCC

4. DCLK2²⁰³ is essential for TBK1 phosphorylation and ccRCC growth in vitro and in vivo