The Lim1 oncogene as a new therapeutic target for metastatic human renal cell carcinoma

Metastatic clear cell renal cell carcinoma (CCC) remains incurable despite advances in the development of anti-angiogenic targeted therapies and the emergence of immune checkpoint inhibitors. We have previously shown that the sonic hedgehog-Gli signaling pathway is oncogenic in CCC allowing us to identify the developmental Lim1 transcription factor as a Gli target and as a new oncogene in CCC regulating cell proliferation and apoptosis, and promoting tumor growth. In this previous study, preliminary in vitro results also suggested that Lim1 may be implicated in metastatic spread. Here we investigated the potential pro-metastatic role of Lim1 in advanced CCC (1) in vitro using a panel of CCC cell lines expressing or not the von Hippel-Lindau (VHL) tumor suppressor gene either naturally or by gene transfer and (2) ex vivo in 30 CCC metastatic tissues, including lymph nodes, lung, skin, bone, and adrenal metastases, and (3) in vivo, using a metastatic model by intravenous injection of siRNA-transfected cells into Balb/c nude. Our in vitro results reveal that Lim1 knockdown time-dependently decreased CCC cell motility, migration, invasion, and clonogenicity by up to 50% regardless of their VHL status. Investigating the molecular machinery involved in these processes, we identified a large panel of Lim1 targets known to be involved in cell adhesion (paxillin and fibronectin), epithelial-mesenchymal transition (Twist1/2 and snail), invasion (MMP1/2/3/8/9), and metastatic progression (CXCR4, SDF-1, and ANG-1). Importantly, Lim1 was found constitutively expressed in all metastatic tissues. The H-score in metastatic tissues being significantly superior to the score in the corresponding primary tumor tissues (P value = 0.009). Furthermore, we showed that Lim1 silencing decreases pulmonary metastasis development in terms of number and size in the in vivo metastatic model of human CCC. Taken together, these experiments strengthen the potential therapeutic value of Lim1 targeting as a promising novel approach for treating metastatic human CCC.


INTRODUCTION
Renal cell carcinoma (RCC) is one of the most malignant tumors and clear cell RCC (CCC), among all subtypes of RCC, is the most common and aggressive one. At diagnosis, metastases are often already present, due to the asymptomatic nature of kidney cancer, and relapse after nephrectomy is common. 1 RCC metastases are surgically difficult to remove.
They are resistant to radiotherapy and systemic therapies including hormonotherapy, chemotherapy and immunotherapy based on interleukin-2 and interferon alpha. Several targeted therapies including the VEGF-neutralizing antibody bevacizumab, the receptor tyrosine kinase inhibitors sunitinib, sorafenib and axitinib, and more recently cabozantinib, and the mTOR inhibitors everolimus and temsirolimus have been approved for the treatment of metastatic RCC desease over the past 10 years. 2 However, to date, metastatic RCC remains largely incurable. 3 New therapeutic approaches are currently being evaluated based on immune checkpoint inhibitors anti-Programmed cell Death protein 1 (PD-1)/anti-Programmed Death-ligand 1 (PD-L1) and anti-Cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) antibodies, alone or in combination with targeted therapies; although gain have been observed with nivolumab (anti-PD-1) in terms of overall survival with these approaches, there are still no tumor regression and cure. 4 We have previously shown the oncogenicity of the developmental signaling pathway sonic hedgehog (SHH)-Glioma-associated oncogene (Gli) and its orchestral role in stimulating other signaling pathways and markers, such as the phosphatidylinositol-3 kinase (PI3K)/Akt pathway that we previously evidenced as an important pathway for kidney carcinogenesis. 5,6 In addition, in this previous work, we have identified various Gli targets, among them the Lim1 transcription factor, that we have subsequently shown to be a new oncogene in human CCC. 7 Lim1, a LIM homeobox protein, is an important actor during development as well, being already expressed at the gastrula stage and being involved thereafter in the head, in the 5 uterus and the vagina development. [8][9][10][11][12][13] In addition, Lim1 governs anterior mesendoderm, node, and midline development together with Otx2 (Orthodenticle homeobox 2), Foxa2 (Forkhead box A2), and its cofactor Ldb1 (LIM domain binding protein1), and is an important actor of kidney development. Indeed, it regulates the expression of genes from both proximal and distal kidney structures, 14,15 being for example involved in the definition of the renal proximal tubule, from which CCC arises. In the adult in human, expression of Lim1 is retricted to some regions of the brain, the thymus and in the endometrial epithelium and stroma. 8,16 Lim1 has been reported to be overexpressed in gastric and pancreatic cancers and in chronic leukemia [17][18][19] and to be reactivated in nephroblastoma and in medulloblastoma. 20,21 In addition, it has been identified as a susceptibility gene for hepatitis B virus-related hepatocellular carcinoma. 22 Interestingly, in medulloblastoma, Lim1 was identified as a functional genomic driver for metastatic dissemination. 21 In our previous work, we have shown that Lim1 is exclusively expressed in CCC tumors compared to normal corresponding tissues; the mechanisms underlying its oncogenic activities involved proliferative and antiapoptotic properties in vitro in cultured cells and on tumor growth in CCC-bearing nude mice models. Preliminary results also revealed that Lim1 may be involved in cell movements and thus that it may be part of the machinery leading to metastasis development. 7 The mechanism of such potential pro-metastatic activity of Lim1 has to be fully defined.
Research on RCC invasion and metastasis mechanisms has a high priority due to the poor prognosis of patients with metastatic RCC disease. Recent works described that matrix metalloproteinases (MMPs) or enzymes that degrade components of the extracellular matrix and basement membrane, such as heparanase, are involved in advanced tumoral stages and that factors involved in epithelial-mesenchymal transition (EMT), such as Snail and Slug, are also involved in metastatic stages through down-regulation of E-cadherin and up-regulation of MMPs. 23 We have shown that Lim1 is involved in the constitutive activation of oncogenic 6 signaling pathways, including the PI3K/Akt pathway, that has also been shown to be linked to metastatic desease. 6,[24][25][26] The present work focused on the effect of Lim1 in migration and invasion of CCC cells both in vitro and in vivo, in addition to its clinical relevance. we illustrated that Lim1 promoted metastasis development.

Lim1 silencing inhibits cell movements in a panel of human CCC cell lines
In our previous work, preliminary results showed that Lim1 is involved in cell movements in 2 CCC cell lines (786-0 and Caki1). We thus investigated this biological property in an additional panel of human CCC cell lines. No pharmacological inhibitors of Lim1 has yet been developed so that we silenced Lim1 expression using specific siRNAs.
In our previous study on Lim1 we used 3 siRNAs, siLhx1ex1, siLhx1ex2 and siLhx1ex3 targeting exon 1, 2 and 3 respectively. 7 Similar results were obtained in this previous work with the 3 siRNA sets. Consistent with this observations, the in vitro experiments herein were performed with the same 3 sets of siRNAs. Only results with siLhx1ex2 (abbreviated siLhx1) will be depicted.    Table 1). Lim1 expression was knockdown for 24 to 96h and many targets were dys-regulated in a time-dependent manner, with an expression profile quite similar from one cell line to another. Lim1 targets include: MMPs (1/2/3/8/9), proteins involved in EMT (snail1, Twist1/2) or cell adherence (fibronectin, paxillin, FAK), and proteins known as prometastatic (CXCR4, SDF-1, ANG-1). Similar results were obtained with the two other sets of siRNAs (data not shown). These results strongly suggest that Lim1 is involved in cell movements and regulates a large panel of proteins involved in invasiveness and EMT to achieve these effects. Clearly, these results argument the pro-metastastic potential of Lim1.

Lim1 induces clonogenicity in human CCC cells
We also performed colony forming assays to assess an other property required for the development of distant metastasis, i.e. clonogenicity. This assay provides a straightforward method to assess the transforming potential of an oncogene. Our experiments revealed that

DISCUSSION
Cancer metastasis is a multistep process in which malignant cells escape from the primary tumor to colonize distant, secondary sites. 27 At metastatic stages, human CCC remains largely uncurable. The genetic changes and the biological mechanisms that drive RCC metastasis are relatively unknown.
The hypothesis we have from our studies and from the investigations of other groups is that tumors hijack developmental signaling pathways and markers, for their own growth.
During renal tumorigenesis, Pax1/2 (paired box gene 2), Lim1, and WT1 (Wilms tumor protein) transcription factors, and the SHH, Notch and Wnt (Wingless integrated) signaling pathways are reactivated. 28 It has been shown that the blockade of the SHH-Gli signaling pathway induces cancer cell death. In addition, recent evidence suggests that a SHH paracrine mechanism mediating tumor-mesenchymal interactions may lead to metastasis. 29 Targeting the SHH-Gli pathway has shown promising in vivo results on tumor growth in gastric, pancreatic, prostatic and breast cancers and medulloblastoma. 30 We previously demonstrated that the SHH pathway is specifically re-expressed in human CCC and that its targeting results in tumor growth inhibition by inhibiting tumor cells proliferation and inducing apoptosis. 5 In this previous study, we also showed that the SHH-Gli signaling pathway interacts with various oncogenic and developmental pathways including the PI3K/Akt, the NF-kB (nuclear factor-kappa B), and the MAPK (mitogen-activated protein kinases) pathways which have been shown to be critical for CCC. 6,31 These results are reminiscent of what is observed during kidney development.
Importantly, we identified Lim1 as a Gli target in this disease. LIM homeobox genes are an important subfamily of homeobox genes which encode LIM-homeodomain proteins featuring two LIM domains in their amino termini and a centrally located homeodomain allowing 13 interactions with specific DNA elements in target genes. Among 12 human LIM homeobox genes, 10 LIM-homeodomain proteins have been reported to be associated with cancer, including Lim1. Thus, besides its primordial roles during development, it has been reported that Lim1 is reactivated or overexpressed in some cancer types. [17][18][19] No reports went deeper into its possible involvement in their pathogenicity and metastatic spread except studies in medulloblastoma. 21 We identified Lim1 as a new oncogene in human CCC, regulating cell proliferation and apotosis. Additionnally, preliminary results suggested that Lim1 blockade induced substantial inhibition of cell movements.
Here we extent our previous results showing that Lim1 is involved in cell motility, migration and invasion in vitro in 3 additional cell lines, i.e. ACHN, Caki2 and A498 cells.
Furthermore, through Lim1 depletion experiments we also show that Lim1 regulates clonogenicity of our panel of human CCC cell lines in vitro. All these effects were independent on VHL status as they were observed in cells expressing of not VHL,a nd clearly demonstrated in 786-0 cells either untransfected or transfected with VHL constructs. Various proteins belonging to the invasion machinery and EMT were shown to be regulated by Lim1.
Interestingly, the profile of proteins regulated by Lim1 from one cell line to another, although very similar, also showed some cell specificities. We expected such results since these cell lines were derived from the tumor of different patients, and because it is now approved that CCC tumors are heterogeneous from patient to patient and in a single tumor at genetic and molecular levels. 32 Clearly, such a result adds some levels of difficulties in the search of active therapies for human CCC. The fact that Lim1 depletion inhibits tumor cell movements of all cell lines tested using similar and probably different molecular pathways argue for a more widespread application of Lim1 targeting for therapeutic intervention in this pathology.
In addition, these observations pave the way for additional fields of investigation in the search of the molecular mechanisms accounting for metastasis development in human CCC. For 14 example, the observation that Lim1 silencing is associated with TIMP-1/MMPs inhibition simultaneously, surprised us initially, however, this unexpected balance between TIMP-1/MMPs has been found in several cancers. Moreover a high level expression of TIMP-1 is correlated with a poor prognosis in RCC. This observation demonstrates the complexity of the mechanisms involved in carcinogenesis. 33,34 The observations that Lim1 is expressed in all CCC metastases tested and the available corresponding primary tumors with an higher level in metastasis compared to the primary tumors (P value 0.009), and that the level of Lim1 expression in metastases was correlated with tumor stage argument further its role in metastasis development. Tumors containing sarcomatoid components have the worse prognostic with an expected life expectancy of one year, vs. 40 months for patients with tumor with no sarcomatoid components. The cohort we used did not allow us to assess whether Lim1 is part of the molecular signature of sarcomatoid areas in human CCC.
Lim1 involvement in metastasis development in vivo environment was clearly demonstrated in the in vivo metastatic model used in the present study. This is a well-known model to directly assess quantitatively and qualitatively the development of metastasis, and it has been used with success for human CCC as well. 35 In addition, it is suitable to assess the role of potential therapeutic targets against which there are no chemical inhibitors available yet. We show here that the metastatic phenotype of human CCC is regulated, at least in part,  37 caveolin-1, that could act also through the regulation of the β-catenin pathway, 38 the miR-193a-3p, that directly targets PTEN, 39 and the novel lncRNA termed metastatic renal cell carcinoma-associated transcript 1 (MRCCAT1), that acts via inhibiting NPR3 and activating p38-MAPK signaling. 40 In addition, comparing early metastatic and non-metastatic CCC samples, it was also discovered that certain differentilally expressed genes including JUN, TNF, RHOB, NR4A1, TGFβ2, LAMA1, LAMA2 and LAMA4 were potential target genes associated with early metastatic CCC. 41 In contrast, CD82/KAI1, acts as a metastasis suppressor in human CCC through the TGF-β1/Smad signaling, 42 as it is the case for KLF6, shown paradoxically as promoting metastasis in other cancer types, that acts via the transcriptional repression of E2F1. 43 It is interesting to note that although these metastasispromoting markers have been demonstrated experimentally to be involved in metastatic spread in human CCC, they were not identified at risk loci in genome-wide or meta-analysis studies. 44 In a recent report characterizing the long non-coding RNA transcriptome in CCC by next-generation deep sequencing, 4 clusters were identified, associated with distinct clinicopathological and genomic features of this disease. The cluster 4 was associated with tumor subtype arising from the distal tubules of the nephron. Lim1 was found to be overexpressed in this cluster compared to the others clusters, consistent with downregulation of its cis-acting lncRNAs RP11-283I3, and thus argumenting for a role for Lim1 in this kidney cancer subtype. Clearly, further studies will help to better characterize how Lim1, as well as other metastasis-promoting markers, act on metastasis development in this refractory disease, and using different but additional experimental approaches. In total, we have shown that Lim1 expression maintains the invasive potential of human CCC tumor cells in vitro and in vivo, which plays a pivotal role in the malignant progression of this cancer type and thus carries highly potent therapeutic implications. In this context, we recently developed in collaboration with chemists a very efficient tool to inhibit Lim1 expression in tumors, using new and innovative polydiacetylenic nanofibers able to vehicle siRNA in vitro and in vivo. 45 Next step will be to assess the efficiency of these tools in relevant models and ultimately to patients.

Cell lines and cell culture
The These patients developed one or more lymph nodes, visceral or bone metastasis (less than one years to 12 years after the diagnosis of the primitive tumors) and for one patient, the diagnosis of the RCC was done on bone metastasis. In total, 30 metastasis were analyzed by immunohistochemistry and/or by Western blot depending on the availability of the tissues, paraffin-embedded or frozen, or both. Thus, for 14 patients, tissues were only conserved in formol, for 1 patients tissues were only conserved in liquid nitrogen, while for 7 patients tissues were available in formol and liquid nitrogen. Sites of metastases were lung, bone, lymph node, adrenal gland, parotid gland, bronchus, pleura, trachea, skin and liver.

RNA isolation and quantitative real-time PCR analysis
Total RNAs were extracted and reverse transcribed as previously described. 47

Western blot analysis
Western blot assay was performed as previously detailed. 46 The mouse monoclonal antihuman Lim1 antibody and the mouse monoclonal anti-human VHL antibody were purchased from Abcam (Paris, France) and used at 1/3000 and 1/500 dilutions, respectively. The mouse anti-human housekeeping GADPH was purchased from Sigma-Aldrich/Merck and used at Wound healing assay Wound healing assay was performed as exposed previously. 7 Briefly, tumor cells were grown in 24-well plates to confluence and transiently transfected with siLhx1 or siCtl. After 24h of treatment, the wound was created in the center of the cell monolayer with a sterile plastic scratcher. The ability of the cells to migrate into the wound area was assessed after 12h and 24h by comparing the 0-12-and 24-h micrographs of several marked points along the wounded area. The percentage of recovered wound area was calculated by dividing the recovered area after 12h or 24h by the initial wound area at zero time.

Cell migration and invasion assays
Tumor cells were grown for 24h and transiently transfected for 24 to 48h with siLhx1 or siCtl.The migration and invasion assays cells were performed on Boyden chamber as detailed. 7,46 Colony formation assay Clonogenicity was assessed with a colony-forming assay. 48 Cells were seeded in 6-well plates at 50% confluence and then transfected with siLhx1 or siCtl at 50nM or untranfected. 24h post-transfection, cells were seeded in 6-well plates at 300 cells/well and cultured for 15 days.
Colonies on each plate were counted after fixation and coloration by crystal violet.
Histology and immunohistochemistry analysis 21 Tissues were fixed in 10% neutral-buffered formalin and processed for histologic examination including embedding in paraffin, sectioning, and staining with hematoxylin and eosin. Four µm sections from selected paraffin blocks for each specimen were used for immunohistochemical analysis. This analysis was realized on a Ventana Roche Benchmark XT and with the same antigen retrieval process, in a EDTA citrate buffer (pH 8, 3