HN1L promotes migration and invasion of breast cancer by up‐regulating the expression of HMGB1

Abstract Recent reports showed that haematological and neurological expressed 1‐like (HN1L) gene participated in tumorigenesis and tumour invasion. However, the expression and role of HN1L in breast cancer remain to be investigated. Here, bioinformatics, western blot and immunohistochemistry were used to detect the expression of HN1L in breast cancer. Wound healing, transwell assay, immunofluorescence assay and mass spectrum were used to explore the role and mechanism of HN1L on the migration and invasion of breast cancer, which was confirmed in vivo using a nude mice model. Results showed that HN1L was significantly over‐expressed in breast cancer tissues, which was positively correlated with M metastasis of breast cancer patients. Silencing HN1L significantly inhibited the invasion and metastasis of breast cancer cells in vitro and lung metastasis in nude mice metastasis model of breast cancer. Mechanistically, HN1L interacted with HSPA9 and affected the expression of HMGB1, playing a key role in promoting the invasion and metastasis of breast cancer cell. These results suggested that HN1L was an appealing drug target for breast cancer.

about HN1L cellular function. In 293T cells, HN1L mainly activates cell cycle-related signalling pathways such as E2F, Rb and P53. Overexpression of HN1L promotes cell malignant proliferation in nonsmall cell lung cancer. 11 Besides, inhibition of HN1L reduces tumour cell proliferation, cancer stem cell self-renewal and migration, tumour growth and metastasis and is associated with improved patient survival. 12 Here, we describe the essential role of HN1L in metastasis of breast cancer and explore its mechanism of action in this context.
We found that the expression of HN1L was up-regulated in breast cancer tissues, and patients with high HN1L expression have a poor prognosis. Knockdown of HN1L not only inhibited the invasion and metastasis of breast cancer cells MDA-MB-231 and BT-549, but also inhibited lung metastasis in nude mice metastasis model of breast cancer. Silencing HN1L effectively down-regulated the expression of HMGB1. Moreover, the expression of HMGB1 was likely to be regulated by the HN1L/HSPA9 interaction. These findings revealed that HMGB1 may be a key protein in HN1L promoting invasion and metastasis of breast cancer and suggested that HN1L was an appealing drug target for breast cancer.

| Tissue samples
Fresh human tissue samples including breast cancer tissues and adjacent mammary tissues were collected from Henan Cancer Hospital. A cohort of 115 paraffin-embedded archived breast cancer specimens were used to determine the clinical significance of HN1L, which were clinically diagnosed as breast cancer from 2013-2014. In this study, adjacent tissues were defined as 2.0 cm from the tumour margin. Histological diagnosis and tumour-node-metastasis (TNM) stages of cancers were determined in accordance with the American Joint Committee on Cancer (AJCC) manual criteria breast cancer. 13

| HN1L expression and patient survival analysis
We selected pairs of sample data of RNA-seq and RNAseqV2 from TCGA database for analysis. Data standardization uses the TMM (Trimmed Mean of M-values) method. We used Mann-Whitney U test to analyse the significant difference of HN1L expression at different levels in different clinical data. We used Spearman's test to analyse the correlation between the expression level of HN1L in cancer tissues and clinical data. The differential expression of the original data of HN1L in 106 pairs of TCGA RNA-seq samples is represented by a line chart. The screening criteria for significant differences must be met | Fold change | should be greater than 2, and the P-value value should be less than .05. According to the median value of HN1L expression as a cut-off value, HNL1 was marked as high expression and low expression. Finally, the relationship between the expression of HNL1 and the overall survival of breast cancer patients was analysed by Kaplan-Meier survival curve.

| Immunohistochemistry (IHC)
Tissue paraffin blocks from 115 breast cancer patients were selected and sectioned for immunohistochemistry. Briefly, the tissue sections (4 μm) were dehydrated and subjected to peroxidase blocking. Primary antibodies were added and incubated at 4°C overnight in a humidified chamber, the slides were incubated with horseradish peroxidase-conjugated secondary antibody and stained with the DAB substrate. The stained slides were observed under microscopy, and images were acquired. 14 The percentage of HN1L-positive cells was scored as 0, <5%; 1, 5%-25%; 2, 25%-50%; 3, 50%-75%; 4, 75%-100%. The intensity of HN1L-positive staining was scored as 0, negative; 1, weak; 2, moderate; 3, strong. The total score was determined by the following formula: Staining index = positive percentage × intensity. The cut-off value for HN1L level was 4, so score ≤4 was considered as low level, and score >4 was considered as high level.

| Cell proliferation assay
Cell viability was performed by MTT assay according to the manufacturer's instructions. Briefly, approximately 3 × 10 3 cells were seeded in 100 μL per well DMEM/RPMI 1640 in a 96-well plate for 96 hours. Cells were then incubated with MTT (5 mg/mL) for another 4 hours at 37°C. After the medium was carefully removed, 100 μL of dimethyl sulfoxide was added and agitated to dissolve the formazan crystals. The absorbance at 490 nm was measured on a spectrophotometric plate reader. Each group was repeated in three different wells.

| Generation of gene knockdown stable cell lines
Stable gene knockdown cell lines were generated as previously described. 15 Briefly, the short hairpin RNA sequences of HN1L were amplified and cloned into the GV115 vector (GeneChem) between Age I and EcoR I sites for expression driven by the hU6 promoter.
The positive bacterial colony was transferred to 150 mL of LB liquid medium containing Amp antibiotics and cultured overnight at 37°C shaker. The plasmid was extracted according to the EndoFree Maxi Plasmid Kit instructions (TIANGEN, DP117), and the right gene sequence plasmid was used for viral packaging. MDA-MB-231 and BT549 cells were plated in six-well plate and infected with lentivirus containing shRNA. And then cells were cultured in a 5% CO 2 incubator at 37°C for another 72 hours. Cells were harvested and total RNA and protein were extracted to determine the knockdown efficiency using real-time PCR and western blot. The sequences of the shRNA are as follows: shHN1L: 5′-CTAATAGGATGGCATCTAA-3′; shHN1L-2 12 : 5′-CGCCTGTATTTGGAAGATTTAA-3′; shHSPA9: 5′-ACATTGTGAAGGAGTTCAA-3′; The negative control was 5′-TTCTCCGAACGTGTCACGT-3′.

| Western blotting
Lentivirus-infected MDA-MB-231 and BT549 cells were collected, lysed and centrifuged, the supernatant was collected for western blot analysis. 16 The equal amount of protein was separated by 12% SDS- HSPA9 (Abcam, ab2799) and GAPDH (Santa Cruz, sc-32233), followed by incubating with secondary antibodies labelled with horseradish peroxidase for 2 hours at room temperature. At last, proteins were detected by using the ECL plus reagents (Beyotime). Densitometric analysis for the quantification relative to GAPDH was performed using the Image J software.

| Cell migration and invasion assays
The migration and invasion assays were performed in a 24-well Boyden chamber with 8 μm pore size polycarbonate membrane. For migration assay, 100 μL of serum-free medium (containing 1 ×

| Co-immunoprecipitation assay
The 3 × FLAG-HN1L interacting complex was purified using the anti-FLAG magnetic beads (Sigma) according to the manufacture's protocol. In brief, 3 × FLAG-HN1L expressing cells and the control cells were lysed respectively with lysis buffer (20 mmol/L Tris-HCl pH7.5, 150 mmol/L NaCl, 1% Triton X-100) supplemented with protease inhibitor cocktail and phosphatase inhibitor cocktail. Equal amount of protein derived from each cell pool was incubated with anti-FLAG magnetic beads at 4°C overnight. After immunoprecipitation, the anti-FLAG beads were washed with TBS buffer (50 mmol/L Tris-HCl, pH7.4, 150 mmol/L NaCl) to eliminate non-specific binding. Immunoprecipitates were then eluted by 150 μg/mL of 3 × FLAG peptide (Sigma).
Eluates were either detected with HSPA9 and Flag antibodies for the interaction of HSPA9 with HNIL or separated by 12% SDS-PAGE gel and stained with Coomassie Blue staining (CBB) for the LC-MS/ MS analysis.

| LC-MS/MS analysis
The experiments were performed on an EASY-nLC 1000 system (Thermo Fisher Scientific) connected to an Orbitrap Fusion mass spectrometer (Thermo Fisher Scientific). The peptide was resuspended with 10 μL solvent A (water with 0.1% formic acid) and was loaded onto the trap column (Thermo Scientific Acclaim PepMap C18, 100 μm × 20 mm) and subsequently separated on the analytical column (Acclaim PepMap C18, 150 μm × 12 cm) with a flow rate of 600 nL/min for 3 minutes. Elution gradient consists of mobile phase B (99.9% acetonitrile, 0.1% formic acid) and mobile phase A. Elution gradient solutions were added as follows: B was increased from 5% to 10% in 16 minutes, 10%-30% in 55 minutes, 30%-95% in 1 minutes and 90% for 6 minutes.

MS data were acquired by Orbitrap Fusion Tribrid Mass
Spectrometer with an ESI nanospray source. 17 The instrument was set to run in top speed mode with 3 seconds cycles for the sur-

| Immunofluorescence assay
Immunofluorescence assay was performed to evaluate the expression of HN1L and HSPA9. Breast cancer cells were plated on a 96well plate. The cells were fixed in 3.7% formaldehyde for 10 minutes and permeabilized with 0.2% Triton X-100 in phosphate-buffered saline (PBS). The cells were labelled with antibodies for 1 hour at room temperature, rinsed in PBS and incubated with a fluorescein isothiocyanate-conjugated secondary antibody for 45 minutes at room temperature. The nuclei were stained with DAPI for 10 minutes. Finally, cells were visualized using immunofluorescence microscope and photographed with a camera.

| RNA extraction and quantitative realtime PCR
Total RNA of breast cancer cells and tissues were isolated using TRIzol Reagent (Invitrogen), and the first strand cDNA was syn-

| In vivo animal experiments
Tumour formation assay was done according to previous reports. 18 Small living animal imaging technology was used to detect the occurrence of tumour metastasis. 19

| GeneChip prime view
The total RNA of the samples (Control vs shHN1L) were extracted by Trizol method, and the total RNA was tested by NanoDrop 2000 and Agilent Bioanalyzer 2100, and the qualified samples were put into the chip experiment. First, the cDNA, was synthesized by one strand, and then the double DNA template was synthesized by two strands, and then the biotin-labelled aRNA (amplified RNA) was obtained by inversion in vitro. The aRNA was purified and then fragmented and hybridized with the chip probe. After the completion of hybridization, the chip was washed and dyed, and finally scanned to get the picture and the original data.

| Statistical analysis
Statistical analysis was performed using the SPSS 19.0 statistical software package (SPSS Inc). All experiments were carried out at least 3 times, and the results were presented as the mean ± standard deviation. The unpaired 2-tailed t test was used for the comparison of parameters between groups. The Mann-Whitney test was used for data that are not of normal distribution by SPSS software. The standard deviation (SD) value was calculated by Excel software. P values of <.05 were considered statistically significant.  (Tables S1 and S2).

| HN1L is over-expressed in breast cancer tissues and closely related to the invasion and metastasis of breast cancer
Next, the expression of HN1L was evaluated using immunohistochemistry (IHC) staining from 15 adjacent breast tissues and 115 breast cancer tissues collected in Henan Cancer Hospital from 2013-2014.
Statistical analysis showed that the positive expression rate of HN1L was 71.3% in breast cancer tissues and 20% in adjacent normal tissues (Table S3). HN1L was over-expressed in breast cancer tissues compared with adjacent normal tissues ( Figure 1B). Meanwhile, HN1L was also expressed in the highly aggressive breast cancer cell lines MDA-MB-231, BT549 and T47D ( Figure 1C). These results were consistent with the results of 106 pairs of breast cancer and paracancerous tissues in the TCGA database ( Figure 1D). Besides, further analysis found that the expression of HN1L was closely related to Androgen receptor (AR), Ki-67, Clinical nodal stage and TNM stage (Table S4, P < .05). Taken together, these findings suggested that the expression of HN1L was up-regulated in breast cancer tissues, which was closely related to the invasion and metastasis of breast cancer. The survival analysis results in the TCGA data show that breast cancer patients with high HN1L expression have a poor prognosis ( Figure 1E).

| Knockdown of HN1L inhibits the proliferation of breast cancer cells
In order to determine the function of HN1L in the invasion and metastasis of breast cancer, RNA interference technique was used to knock out the expression of HN1L. We selected highly invasive  Figure 2B,C). In a cell growth assay, we found that the ability of cell proliferation was drastically decreased after HN1L knockdown in both cell lines ( Figure 2D).

| Knockdown of HN1L inhibits the invasion and metastasis of breast cancer cells
To further assess the role of HN1L on the invasion and metastasis of breast cancer, stable HN1L knockdown cell line was generated and wound healing assay and transwell analysis were examined. As shown in Figure 3A

| Knockdown of HN1L inhibits lung metastasis in vivo
After demonstrating the inhibition efficacy of HN1L silencing in vitro, we further examined the effect of HN1L knockdown in vivo. Tumour growth and metastasis were monitored in NOD-SCID mice implanted with control and HN1L knockdown MDA-MB-231 cells. The tumour weight showed no statistic difference between Control and HN1L knockdown groups ( Figure S1). However, knockdown of HN1L resulted in a 70% decrease of lung metastases ( Figure 4A,C). Furthermore, small living animal imaging technology data analysis showed that the expression of whole body fluorescence was decreased in HN1L knockdown group (P < .05) ( Figure 4B,D), suggesting that silencing HN1L reduced the metastasis rate of breast cancer. Finally, we further verified by immunohistochemical experiments. As shown in Figure 4E, the metastases in the shHN1L group were significantly less than those in the shControl group in HE staining and immunohistochemical images, consistent with the previous results ( Figure 4E).

| HN1L silencing inhibits the invasion and metastasis of breast cancer by regulating HMGB1
To explore the target of HN1L, GeneChip primeview human analysis was used after silencing HN1L ( Figure S2 and Table S5). Knockdown of HN1L led to a significant accumulation of DDX58, and a decrease of SMAD2, PIM1 and HMGB1 ( Figure 5A). Recently, HMGB1 was reported to be over-expressed in a number of cancers and involved in tumour invasion and metastasis. [23][24][25][26][27] So we hypothesize that HN1L regulates the metastasis of breast cancer cell by HMGB1. To test this hypothesis, HMGB1 was over-expressed in HN1L-silencing cells. The efficiency of HN1L knockdown and HMGB1 over-expression was confirmed by western blot assay ( Figure 5B). HN1L knockdown significantly inhibited the migration ability of breast cancer cell, which was recovered by HMGB1 over-expression ( Figure 5C,D). These findings highlighted a pivotal role of HMGB1 in the process where HN1L-silencing-inhibited invasion and metastasis in breast cancer cell.

| HMGB1 was likely regulated by the HN1L/ HSPA9 interaction
To further investigate the molecular mechanism of HN1L in the invasion and metastasis of breast cancer, the protein interacting with HN1L was examined using LC/MS ( Figure S3). Results suggested HSPA9 is one of the proteins that interact with HN1L, which was confirmed by the CO-IP assay ( Figure 6A,B). Next, the effect of HSPA9 on the metastasis was examined. HSPA9 was knockdown by a specific shRNA oligoes, named shHSPA9. The knockdown efficiency of shHSPA9 was confirmed by western blot ( Figure 6C) and Q-PCR assay ( Figure 6D). Further results showed that knockdown of HSPA9 also led to the significant decrease of HMBG1 ( Figure 6E). At the same time, immunofluorescence experiment showed that HN1L and HSPA9 were co-expressed in breast cancer cells ( Figure 6F).
So we speculated that HMGB1 was likely regulated by the HN1L/ HSPA9 interaction.

| D ISCUSS I ON
Nowdays, targeted therapy is one of the most important parts of breast cancer treatment and has shown good efficacy. 28 For example, the molecular targeted drug trastuzumab has been used clinically and has shown good therapeutic effects. 29,30 Therefore, it is urgent to find new therapeutic target.
A previous report revealed the HN1L is a targetable breast cancer stem cell (BCSC) gene hinting its potential diagnostic and prognostic significance. 12 In addition, the positive correlation of HN1L expression and Ki67 level in a large non-small cell lung cancer samples further suggested the key role of HN1L in the regulation of cell growth. 11 The results of TCGA database data analysis and IHC results also confirmed the over-expression of HN1L in breast cancer. Furthermore, elevated HN1L was significantly associated with Ki67 level and lymph node metastasis in breast cancer patients (Table S4)  Although HN1L expression has been reported to be significantly associated with shorter overall or recurrence-free survival in triple-negative breast cancer (TNBC) patients, 12 the exact molecular mechanism remains unclear. Mechanistically, we examined the protein that interacted with HN1L. Among these proteins, we found a more interesting protein, HSPA9, which was related to proliferation, function maintenance and stress response in cancer cells. The human HSPA9 gene is located on chromosome 5q31.2 and encodes for one of the heat shock protein 70 family members, also known as mortalin/mthsp70/PBP74/GRP75. 34 As a molecular chaperone, HSPA9 interacts with other proteins and functions in regulating cellular stress response, EMT, 35 cell proliferation 36  3 × FLAG-HN1L expressing cells and the control cells were purified using the anti-FLAG magnetic beads and eluted by 150 μg/mL of 3 × FLAG peptide. Eluates were detected with HSPA9 and Flag antibodies for the interaction of HSPA9 with HNIL. WCL means whole cell lysis. (C and D) Knockdown efficiency of shRNA targeting HSPA9. Cells were infected with shRNA for 72 h, proteins and RNA were collected and knockdown efficiency was determined by western blot (C) and Q-PCR (D) (**P < .01). (E) Western blot assay was applied to assess the expression of downstream proteins in the HSPA9-silencing cell line and its control. (F) Immunofluorescence experiment showed that HN1L and HSPA9 were co-expressed in breast cancer cells was reported to be up-regulated in many cancers. 38,39 Besides, the over-expression of HSPA9 promoted invasion and metastasis of breast cancer and associated with histological grade, clinical stage and lymph node metastasis. 35,40 Emerging data have suggested that HMGB1 could promote tumour progression via promoting proliferation and invasiveness of cancer cells. 41,42 HMGB1 was highly expressed in many tumours, such as lung cancer, 41 prostate cancer, 42 gastric cancer, 27 hepatocellular carcinoma, 43 breast cancer, 26 colorectal cancer 44 and ovarian cancer. 45 In addition, HMGB1 was highly expressed in renal cell carcinoma, and the expression level showed a positive correlation with cancer bearing, metastasis and clinical staging and grading. 46,47 As a multifunctional cytokine, HMGB1 plays a key role in tumour formation, metastasis and EMT. 48,49 In our study, Knockdown of HN1L led to a significant accumulation of DDX58, with a decrease of SMAD2, PIM1 and HMGB1. PIM-1 can inhibit the growth of triple-negative breast cancer cells, 50 so the decrease of PIM1 expression caused by HN1L gene silencing may be one of the ways by which HN1L regulates the proliferation of breast cancer cells. Smad2 is the substrate of TGFβ receptor kinase and the key protein of TGFβ/ Smads signalling pathway. 51 It plays an important role in cell proliferation, differentiation and apoptosis. 52 Here, the decrease of SMAD2 expression after HN1L gene silencing may be the result of the effect of HN1L gene silencing on cell proliferation and apoptosis, but the detailed molecular mechanism of HN1L regulating breast cancer cell proliferation needs to be further studied.
It is noteworthy that among the HN1L-associated genes detected by ChIP-seq, HMGB1 has been reported as a key regulator in the EMT process in mesothelial cells, 53,54 suggesting another possible mechanism of HN1L through regulation of HMGB1 expression. In our rescue experiment, over-expression of HMGB1 in HN1L-silenced breast cancer cells saved the migration ability of breast cancer cells, which suggests that HNIL affects the migration ability of breast cancer cells by regulating the expression of HMGB1. Besides, HMGB1 was also reported as a senescence marker. 55,56 Meanwhile, recent studies show that depleted cells of the mitochondrial chaperone HSPA9 induced cell senescence. 56 Considering that HSPA9 is involved in cell senescence and HMGB1 is one of the markers of cell senescence, we hypothesize that HNIL may interact with HSPA9 and participate in breast cancer cell invasion and metastasis mediated by HMGB1.
To test this hypothesis, CO-IP and immunofluorescence assays were first used to find the interaction of HN1L and HSPA9. Results showed that HN1L and HSPA9 were co-expressed and interacted in breast cancer cells which were in line with our hypothesis. Consistent with knockdown of HN1L, HSPA9 knockdown also caused a significant decrease in HMGB1 protein expression. These studies suggest that HN1L/HSPA9-HMGB1 perhaps play an important role in the migration and invasion of breast cancer. However, the detailed mechanism of the interaction between HN1L and HSPA9 affecting the expression of HMGB1 still needs further experimental studies.
In summary, our study proves that HN1L silencing can inhibit the migration and invasion of breast cancer cells in vitro, and lung metastasis in vivo. Although our study offers new insight into the mechanisms of HN1L in breast cancer metastasis, the specific regulatory mechanisms of the interactions among HN1L, HSPA9 and HMGB1 need further study. Taken together, HN1L may provide promising diagnostic and therapeutic options for breast cancer patients in the future.

ACK N OWLED G EM ENTS
We are grateful to Miss. Ran Yang for her help with the immunohistochemistry (IHC) assay. We also thank Dr Tao Hu for critical reading and editing of the manuscript. At last, we especially thank Xingge Liu for helping us in the lentivirus infection experiment.

CO N FLI C T O F I NTE R E S T
No potential competing interests are disclosed.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing is not applicable to this article as no new data were created or analyzed in this study.