KRASG12 mutant induces the release of the WSTF/NRG3 complex, and contributes to an oncogenic paracrine signaling pathway

It remains unclear how the signals of mutant KRASG12 in the transformed cells spread to the surrounding non-mutated cells and changes the microenvironment to promote tumor formation. We identified that Williams–Beuren syndrome transcription factor (WSTF), a non-secretory protein, was released in complex with secretory protein-neuregulin-3 (NRG3). The KRASG12 mutant activates the transcription of NRG3. The WSTF/NRG3 in extracellular space could activate oncogenic pathways in normal colon cells carrying wild type KRAS and endow them with the ability to express NRG3 and release WSTF/NRG3. Extracellular WSTF/NRG3 promotes the formation of colon tumors. Blockade of extracellular WSTF could restore cetuximab sensitivity of colon cancer cells with mutant KRAS. The appearance of WSTF/NRG3 in serum and urine correlates with a colon tumor carrying a KRASG12 mutant. In summary, our demonstration provides a new pathway to our understanding of the biological development of complex diseases.

wild type (WT) KRAS was detected by direct sequencing. A stable HIPEC line, which was introduced with KRASG12V mutation through transfecting pEGFP-N1human-H-Ras G12V plasmid, was designated as HIPEC KRASM . In HIPEC KRASM , the activity of the RAS-mitogen-activated protein kinase (MAPK) pathway was obviously enhanced with higher levels of phosphorylated extracellular signalregulated kinase 1/2 (P-ERK1/2) compared with normal HIPEC cells ( Figure 1A). The RAS-PI3K and -RalGEF pathways were activated to different levels ( Figure 1A).
With the intention to examine secreted proteins of HIPECs following the introduction of KRASG12V, we detected WSTF in the cell media unexpectedly, which was designed as a negative control ( Figure 1B). WSTF locates mainly in the nucleus and is involved in chromatin assembly [7,8]. No known signal peptide or transmembrane domain sequences of secretory protein were identified in WSTF [8]. Interestingly, P-WSTF (phosphorylated at Ser 158) was markedly increased intracellularly in HIPECs KRASM compared with original HIPECs, whereas no P-WSTF was detectable in either media ( Figure 1B). We speculate that WSTF in the media may be released actively from the living HIPECs KRASM , but not from fragmented cells.
To address whether the release of WSTF was induced by RAS signal, specific small interfering RNAs (siRNAs) were transfected into HIPECs KRASM to block MAPK, Ral-GEF, or PI3K pathway. As shown in Figure 1C, WSTF releasing was strongly decreased following the blockade of MAPK pathway by specific c-Raf siRNA, but not the blockade of the other two pathways. To further confirm this result, specific inhibitors were used to interrupt the transduction of MAPK or PI3K signal. As shown in Figure 1D, the release of WSTF was downregulated by MAPK pathway inhibitor U0126, but not by PI3K pathway inhibitor Wortmannin. preferentially activate the Ras pathways. The levels of P-ERK1/2, P-AKT and RalA-GTP were detected with specific antibodies to examine the activities of corresponding pathways. (B) Media containing WSTF and P-WSTF was detected using ELISA. Levels of intracellular P-WSTF, total WSTF protein and mRNA were detected as controls. (c) Different small interfering RNAs were transfected into HIPEC KRASM cells for 48 h. Two specific siRNAs for each gene were implicated in experiments. The secreted WSTF was detected by ELISA assay. (d) The media of HIPEC KRASM cells, which was cultured with U0126 (10 μM) or Wortmannin (10 μM) for 20 h, were tested by ELISA. Non-treated HIPEC KRASM cells were used as control. www.impactjournals.com/oncotarget secretion of WstF was not activated by exocytosis system Next, we wanted to explore the mechanism regulating the release of intracellular protein WSTF. Rabin8 is an important factor of the exocytosis system that stimulates vesicular trafficking to the plasma membrane, and is known to be activated by MAPK pathway [9]. Therefore, we investigated the phosphorylation level of Rabin8 in response to KRAS mutation, however, no obvious changes were detected ( Figure 2A). Further, knock-down or overexpression of Rabin8 had no greatly effect on the WSTF levels in media ( Figure 2B), so the release of WSTF does not appear to be due to Rabin8.
Next, to test the possibility of the involvement of lysosomal exocytosis, we checked the key component of this system, transcription factor EB (TFEB) [10]. The TFEB protein was increased or decreased with the introduction of expressing plasmid or RNAi. However, the level of WSTF in media did not fluctuate according to the change of TFEB level as shown in Figure 2C.

KrAs mutation activates the nrG3 protein expression to transport WstF
We then wondered whether WSTF was carried out of the cell by an ectopic secretory protein. To confirm this possibility, FLAG-WSTF was stably expressed in wildtype HIPEC and HIPEC KRASM cells and then subjected to affinity purification and mass spectrometry. NRG3, one of the neuregulin family members was identified to associate with WSTF following the introduction of KRAS G12V ( Figure 3A). Human NRG3 has been identified in the developing breast and central nervous system, but not in other normal tissues [11,12]. Ectopic expression of NRG3 was detected in a proportion of primary breast cancer biopsies (42%) with undefined function [13][14][15]. Further experiments were performed to test whether NRG3 in HIPECs KRASM is induced by KRAS mutation. NRG3 was not detected intracellularly and extracellularly under the circumstances of WT-KRAS, whereas the KRAS G12V mutation resulted in the expression and secretion of NRG3 ( Figure 3B-3C).
Besides the ectopic expression and secretion of NRG3, no significant changes of leukemia inhibitory factor (LIF) secretion were observed, which was induced by the RAS-MAPK pathway activation and was sufficient to induce growth arrest and differentiation of surrounding cells ( Figure 3C). These results demonstrate that the balance of paracrine signaling that is activated by the RAS-MAPK pathway may be broken in KRAS G12 mutation induced colon cancer.
Next, we sought to observe whether overexpression of NRG3 is sufficient to induce the secretion of WSTF in the absence of activated Ras mutation in HIPECs. Different amount of NRG3-expressing plasmids were transfected into wild-type HIPEC cells. As expected, secretory WSTF was detected in the media with NRG3 proteins ( Figure 3C). Moreover, WSTF was undetectable in the media of HIPECs KRASM following NRG3 knock-down ( Figure 3D).
To understand the possible reason of NRG3 silencing in colon cells, we analyzed the proteins binding at the NRG3 promoter through Chromatin Immunoprecipitation (ChIP). In HIPECs, active markers of transcription, such as RNA pol II and histone 3 lysine 4 dimethylation (H3K4me2), were absent, whereas markers of transcriptional repression, such as heterochromatin protein 1α (Hp1α) and H3K9me2, were detected at the NRG3 promoter region ( Figure 3E). Conversely, in HIPECs KRASM the promoter region of NRG3 showed signs of active transcription ( Figure 3E). We surmise that KRAS G12V induced changes in histone modifications and Hp1α alter the chromatin structure and then activate the transcription of NRG3, which is in agreement with similar RAS signaling events that we have previously identified [3].
To test whether the increased P-ERK1/2 is sufficient enough to induce NRG3 expression without KRAS G12 mutation, expression plasmid for ERK1/2, or c-Raf was transfected into HIPECs. NRG3 expression was not detected following the upregulation of P-ERK1/2, and the recruitment of RNA pol II, H3K4me2, Hp1α and H3K9me2 at the NRG3 promoter region were not strongly changed as measured by ChIP assay ( Figure 3F). This result revealed that P-ERK1/2 signaling alone is insufficient to initiate the expression of NRG3.

nrG3 directly binds WstF
To confirm the association between NRG3 and WSTF, HIPEC KRASM cells lysates or media were collected and co-immunoprecipitation were performed with antibody against WSTF followed by immunoblotting with antibody against NRG3. The data indicated that NRG3 binds WSTF ( Figure 4A, upper panel). Reciprocal co-immunoprecipitation with antibody against NRG3 and immunoblotting with antibody against WSTF also showed that WSTF binds NRG3 ( Figure 4A, middle panel). No co-immunoprecipitation of NRG3 and P-WSTF was detected ( Figure 4A, middle panel and lower panel).
Furthermore, to investigate the regions of WSTF binding with NRG3, glutathione S-transferase (GST) pull-down was conducted with NRG3-GST and wild type or serial truncated mutants of Myc-tagged WSTF, which were produced as previously reported [7]. The results revealed that WSTF could bind NRG3 directly and the region around amino acids (aa) 1-576 of WSTF is critical for binding with NRG3 ( Figure 4B). To detect the binding of endogenous WSTF to NRG3 in vivo, WSTF/NRG3 complex in HIPEC KRASM cells was detected by in situ proximity ligation assays (PLA). As shown in Figure 4C, in situ PLA signals were detected as green dots.
We were interested to investigate the binding of NRG3 with WSTF in other human cancer cell lines and found that the association between NRG3 and WSTF could be unique in colon cells. As shown in Figure 4D, NRG3 only was observed in MCF7 cells; nevertheless, NRG3 and WSTF were not detected in the MCF7 media ( Figure 4E).

secreted WstF/nrG3 activates the release of WstF/nrG3 from normal colon cells and increases the activities of oncogenic pathways
As NRG3 is not expressed in normal HIPECs, we wanted to explore the effects of WSTF/NRG3 release as a consequence of KRAS G12V . HIPECs KRASM were seeded in 6-well plates and after overnight incubation, the media was collected and half was analyzed for WSTF/ NRG3, while the other half was mixed with equivalent fresh media. This conditioned media was then used to culture wild-type HIPEC cells that were seeded in a 6-well plate for 20 hours followed by WSTF/NRG3 secretion analysis. The WSTF or NRG3 concentration in the media from wild-type HIPECs was almost equal to that in the HIPECs KRASM media ( Figure 5A). The recovery of WSTF/NRG3 concentration after dilution with equal amount fresh media suggest the normal HIPECs acquire the ability to release WSTF/NRG3 after cultured with the conditioned media.
The cell lysate from the wild-type HIPEC cells treated with conditioned media were also analyzed. The results demonstrated that the activities of RAS, NOTCH1 and JAK pathways become much higher in wild-type HIPECs following culture with the conditioned media ( Figure 5B). Importantly, the NRG3 expression was observed in the wild-type HIPECs cultured with the conditioned media ( Figure 5B). These results indicate that the WSTF/NRG3 released from KRAS G12V mutant www.impactjournals.com/oncotarget cells initiates the release of WSTF/NRG3 from normal colon cells and this function could be a consequence of enhancement of the oncogenic pathway activity.
To confirm the function of WSTF/NRG3 the media from HIPECs KRASM , which stably express HA-WSTF and MYC-NRG3, was collected and used to culture wild-type HIPECs with equal volume of fresh normal media. Then the media was collected for an ELISA assay. Endogenous and exogenous WSTF/NRG3 was both detected ( Figure 5C). HA-/MYC tagged WSTF/NRG3 was diluted after mixing with fresh media, whereas the total amount of WSTF/NRG3 was almost the same in the final media compared to the original media. Moreover, western blot was applied to analyze the endogenous levels of WSTF/ NRG3 in the wild-type HIPECs that were cultured with the conditioned media. Endogenous WSTF/NRG3 was detected without any tag signal ( Figure 5D). Same result was observed through in situ PLA assay (negative figures were not shown). These results indicate that WSTF/NRG3 in the media may initiate the release of endogenous WSTF/ NRG3 from the wild-type colon cells through specific receptor pathways rather than entering into the cells. This was supported by the fact that a western blot experiment demonstrated that HIPECs express endogenous NRG3 after cultured with fresh media, which contains an equal amount of WSTF and NRG3 ( Figure 5E).
Furthermore, after culturing with the media containing WSTF/NRG3 for 48 hours, the conditioned media was removed and the wild-type HIPEC cells were washed with PBS before culturing with fresh normal media for several passages. Surprisingly, we noticed the daughter generations of HIPECs inherit the capability to produce and release WSTF/NRG3 after culturing with the conditioned media ( Figure 5F).

extracellular WstF/nrG3 promotes cancer formation
To further explore the role of WSTF/NRG3 release in tumorigenesis in vivo, tumor formation experiments were performed. SW48 cells (human colon cancer cell line containing WT KRAS) in PBS or PBS containing WSTF/ NRG3 proteins (PBS + ) were injected into 6-to 8-weekold female BALB/C mice. Measurements of the tumor indicated that tumors of the PBS + group grew significantly faster than PBS group ( Figure 6A).
To investigate the underlying mechanisms of this effect, the tumor cells were collected for further analyses of signaling pathways and WSTF/NRG3 expression. The RAS-MAPK, NOTCH1 and JAK pathways had much increased activity in PBS + group in comparison to PBS group. The NRG3 protein expression was also detected in PBS + group, but not PBS group. Collectively, the results reveal that WSTF/NRG3 stimulates the growth of tumors, and this stimulation could be related with activation of oncogenic pathways ( Figure 6B).
We next sought to explore whether secreted WSTF/ NRG3 could be detected in serum. The serum samples of mice were collected once per day for 5 days during the course of the experiment followed by ELISA assays. From the fifth day after injection, WSTF and NRG3 could be detected in serum samples of the PBS + group and the levels were increased along with tumor growth ( Figure 6C). This result reveals that the SW48 cells acquire the ability to release WSTF/NRG3 after injection with the protein mixture. It is noteworthy that the appearance of WSTF and NRG3 in serum was earlier than any detectable tumor mass.

extracellular WstF/nrG3 could be developed as a diagnostic marker
Given the results that serum WSTF/NRG3 correlates with tumor formation, we sought to explore whether detection of the serum and urine WSTF/NRG3 could be developed into a clinical diagnostic test. We collected serum and 24-hour urine samples of 398 cases of colon cancer patients. Mutation of KRAS at G12 (G12V 58%, G12D 41%, others 1%) was identified in 109 of the 398 cases (KRAS M+ ). Furthermore, ELISA assay indicated that all the 109 cases were WSTF/NRG3 double positive (WSTF + /NRG3 + ) in serum and urine ( Figure 7A).
To confirm the relationship between KRAS mutation and WSTF/NRG3 secretion, the serum and urine samples of a separate cohort of 369 cases with suspected colon cancer were collected. WSTF + /NRG3 + of both serum and urine samples were identified in 116 out of the 369 patients. Furthermore, PCR and sequencing analysis were performed with the tumor tissue samples of the 116 patients after surgery. The results indicated that all the 116 cases contain KRAS mutation at G12 (G12V 61%, G12D 37%, others 2%). After pathological analysis, 100 of 116 were diagnosed as colon cancer and the other 16 were confirmed as atypical hyperplasia ( Figure 7B).

Blockade of WstF could reverse the drug resistance associated with KrAs mutant
As we mentioned previously, the release of WSTF activates NOTCH1 and JAK, in addition to RAS pathways. These three pathways are all relevant to drug resistance [16]. Patients with colon cancer who receive the epidermal growth factor receptor (EGFR) targeted antibodies cetuximab or panitumumab usually develop resistance within several months of initiating therapy. The emergence of mutations in KRAS is associated with acquired resistance to EGFR blockade [17,18].
To explore whether the blockade of WSTF was helpful in the treatment of EGFR drug resistance patients, we added cetuximab in combination with WSTF or β-actin antibodies to the media of SW48 or SW48 KRASM cells. As illustrated in Figure 8A, SW48 KRASM cells were not sensitive to cetuximab alone at doses up to 20 μg/mL The addition of WSTF antibodies (Abcam, ab50987 and ab51256) could inhibit the proliferation of SW48 KRASM cells, whereas the addition of a β-actin antibody (ab8226) did not have this effect ( Figure 8A). Furthermore, western blot results reveal that cetuximab treatment efficiently inhibited the phosphorylation of EGFR and downstream pathways activation ( Figure 8B), but not the ERK1/2, NOTCH1 and JAK pathways. However, the combination of cetuximab and WSTF antibody blocked these pathways remarkably in both SW48 and SW48 KRASM cells ( Figure 8B).
Next, we investigated whether a reduction in WSTF expression and secretion could restore cetuximab sensitivity. Intracellular and extracellular WSTF were decreased with WSTF or NRG3 siRNA ( Figure 8C). As show in Figure 8D, decreasing WSTF expression and release, in combination with cetuximab treatment, significantly inhibited SW48 KRASM cell growth. www.impactjournals.com/oncotarget dIscussIon Cells carrying KRAS mutations were only detected in a fraction of the tumor biopsies from resistant patients. One possibility is that a paracrine cross-talk driven by the resistant subpopulations may provide protection for surrounding sensitive cells that have wild type KRAS. Because colon cancer cells with wild type KRAS could be transformed by WSTF/NRG3, they would be not sensitive to the targeted inhibition of EGFR. Moreover, the WSTF/  signaling pathways were measured through western blot with the indicated antibodies. The HIPECs cultured with fresh media were used as control. (c) The same number of HIPECs KRASM and HIPECs were seeded in 6-well plates and cultured with normal or conditioned media respectively. After 24 hours culture, different media were collected for ELISA assay with the indicated tags or WSTF or NRG3 antibodies. (d) Cell lysate were collected from normal colon cells following cultured with conditioned media. WSTF/NRG3 and tags were detected with the antibodies indicated. The HIPECs and HIPECs KRASM cultured with fresh media were used as controls. (e) Normal colon cells was cultured with fresh media which contain equal amount of WSTF and NRG3 (indicated as conditioned media). The cell lysate were collected to perform immunoblot assay. (F) HIPECs were cultured with conditioned media (half fresh and half from HIPECs KRASM ) for 48 hours. Then the media was removed and the cells were washed with PBS and further cultured with fresh normal media for several passages in the next month, which were indicated as passaged cells. The HIPECs routinely cultured with fresh normal media were indicated as original cells.
NRG3 complex not only provides protection for the adjacent cells, the adjacent cells release their own WSTF/ NRG3 to further amplify this signal.
The event we demonstrate can potentially be applied beyond cancer to all human diseases containing gene mutations, even in tissue development. WSTF and NRG3 may be developed into diagnostic markers, although further analyses of more cases of patients are necessary. The problems still remaining are the following: (1) How WSTF and NRG3 enter into blood and urine? (2) Whether the release of WSTF/NRG3 induced by KRAS G12 mutations is tissue or cell-type specific. (3) How the complex of WSTF and NRG3 is formed. (4) Whether mutations occur more frequently in normal cells following the stimulation by WSTF/NRG3; and (5) whether all normal colon cells or just a subpopulation that expresses certain receptors could be affected by WSTF/ NRG3.    [6] (from a 35-year old male patient undergoing surgery at Kai-Luan General Hospital for ulcerative colitis) and cultured in F-12 medium (Gibco) containing epidermal growth factor (EGF) (5 ng/ml), transferrin (500 μg/ml), insulin (500 μg/ml), hydrocortisone (100 μg/ml), and retinoic acid (5 μg/ml). Experiments were performed according to the manufacturer's instructions and previous report [19]. All cell lines were routinely screened for the presence of mycoplasma (Mycoplasma Detection Kit, Roche Diagnostics).

ethics statement
All studies performed with human cancer specimen and mice were approved by the Ethics Committee and Animal Care Committee of North China University of Science and Technology, and informed consent was obtained from all patients.

Plasmid construction and sirnA
The coding regions of human KRAS, WSTF and NRG3 were amplified from 293T cDNA by polymerase chain reaction (PCR). The PCR products were subcloned into HA-tag or Myc-tag vectors and sequenced. The KRAS G12 mutant was constructed using the TaKaRa MutanBEST Kit (catalogue number R401), as recommend by the manufacturer. siRNAs were purchased from Shanghai GenePharma. Two specific siRNAs for each gene were implicated in experiments.

co-immunoprecipitation analysis
Co-immunoprecipitation assay was performed as described previously [19]. Briefly, cells were suspended with buffer and fragmented by sonication. Then, the cell lysates were reacted with normal IgG or different antibodies. The complex reacted with Protein A/G agarose beads. Next, the beads were washed with buffer and the deposited proteins were freed by boiling.

tail-vein collection of blood
Alcohol was used initially as a vasodilator, but it should not be used on broken skin. The tip of tail will be aseptically prepared or wiped with alcohol prior to the tail-snip or excision. The tail will be nicked with the sterile scalpel blade (< 0.5 mm) and the blood will be collected in a sterile capillary tubes. Pressure will be applied to the tail to stop bleeding and the mouse will be returned to the cage. If bleeding persists, the nicked tail will be cauterized using a silver nitrate stick.

statistical analysis
The statistical analyses were performed using the Student's t test. A p value of 0.05 was considered significant (*, p < 0.05; **, p < 0.01).