Background
Breast cancer is the most prevalent type of cancer, with the highest incidence and leading cause of cancer-related deaths in women worldwide. The most aggressive intrinsic subtype, triple negative breast cancer (TNBC), accounts for 10–20% of all breast cancer diagnoses and has limited treatment options and low 5-year survival rates [
1].
Ribosomes have been considered as passive translation machineries for decades, stable in stoichiometry and with limited control over the translation process. In recent years, ribosomes have been identified as flexible components and drivers of cancer initiation and progression, and have been proposed as drug targets [
2‐
5]. One substance targeting the eukaryotic ribosome, homoharringtonine, has already found its way into clinical practice for chronic myeloid leukemia therapy [
6].
Ribosomal biogenesis is a highly regulated, multistep process, starting with the transcription of ribosomal DNA followed by association of rRNA with the ribosomal proteins that requires the concerted action of over 200 proteins. Ribosomal biogenesis takes place at the nucleoli, which adapt to the ribosomal need. Nucleolar morphology and quantity are altered in a wide range of malignant tumors and constitute a factor in tumor diagnosis and grading in routine histopathological diagnostics [
7,
8].
Alterations in the ribosome imply deviations in the translation of different mRNAs [
2‐
5,
9,
10]. Mutations and deletions in ribosomal proteins or ribosome biogenesis factors were first linked to rare congenital diseases, the ribosomopathies, which are accompanied by a hyperproliferative phase with an increased risk of developing cancer [
11]. Lately, changes in ribosomal composition have been directly related to cancer. In breast cancer, changes in expression of the ribosomal proteins RPS9, RPS14, RPL5, RPL10, RPL11, and RPL39 have been related to tumor initiation and progression [
12‐
16]. RPL15 was identified to promote metastasis and translation of regulators of translation and cell cycles in circulating breast tumor cells [
17]. Moreover, specific ribosomal mRNA patterns were detected in cell lines and in breast cancer tissue specimens that solely and successfully distinguished a healthy from a pathological condition [
18]. Alterations in the ribosome stoichiometry imply deviations in the translation of different mRNAs [
4]. Ribosome specialization might reflect an adaptation to prolonged environmental changes such as continued nutrient depletion or hypoxia [
19].
The transcription factor Krüppel-like factor 7 (KLF7) has increasingly attracted attention in the context of cancer development [
20,
21]. KLF7 is a member of the conserved Klf/Sp1 transcription factor family and is widely expressed in the human body, e.g., in glandular cells of the digestive tract and in lymphoid tissue of the appendix. In most malignant tumors, KLF7 is present at high expression levels (Human Protein Atlas available from
www.proteinatlas.org [
22]). It has been demonstrated that KLF7 is involved in neuronal differentiation during development, and in negative or positive regulation of proliferation in hematopoietic cells, myoblasts, and preadipocytes, depending on the cell type [
23‐
26]. KLF7 is a target of TP53 and regulates Golgi complex integrity in pancreatic cancer cells [
20]. Despite the high abundance of KLF7 in healthy and pathologic tissues, the biological role of KLF7 is still poorly understood.
Here, we investigated the role of KLF7 expression on cellular mechanisms in breast cancer. Using transcriptomics and proteomics approaches in mammary carcinoma cell lines and in patient tissue samples we aimed at identifying and exploring KLF7 regulated processes.
Materials and methods
RNA sequencing (RNA-Seq) analyses
Cells were grown in a cell culture dish to 80% confluency, harvested, and RNA extracted using the E.Z.N.A. Total RNA Kit I (Omega Bio-tek) according to the manufacturer’s instructions. Library prep. and sequencing were performed at the Genomics and Proteomics core facility, DKFZ Heidelberg. RNA quality was verified and was above RIN > 9,5 for all samples. Library preparation was started with 500 ng of input. The raw RNA sequencing files were pre-processed with trimmomatic [
27] to ensure sufficient read quality by removing adapters and reads in low-quality segment regions with a base quality below 20. Subsequently, the reads were 2-pass aligned using the STAR aligner [
28] and the GRCh37 reference genome from Ensembl. Alignment was followed by normalization and differential expression analysis with the R/Bioconductor [
29,
30] package DESeq2 [
31]. Genes were considered significant with an adjusted p-value (FDR corrected, according to Benjamini-Hochberg)
p < 0.05.
Gene set enrichment analysis
Gene set enrichment analysis (GSEA) of signaling pathways was performed as implemented in the R/Bioconductor package GAGE (Generally Applicable Gene-set Enrichment analysis) [
32], with signaling pathways from gene ontology (GO) [
33,
34], ConsensusPathDB [
35,
36], KEGG [
37] and Reactome [
38]. Pathways were considered significant with an adjusted p-value (Benjamini–Hochberg) < 0.05.
Explorative proteomics
MDA-MB-231 cells with KLF7 over expression (KLF7OE) or control plasmid were grown to 80% confluency, washed twice with ice cold PBS, and harvested by scraping. Sample preparation was performed as previously described [
39,
40]. Briefly, cell pellets were lysed with 0.1% Rapigest, 0.1 M Hepes pH 8.0 supplemented with protease inhibitors, and sonicated for 20 cycles. Protein concentration was determined by BCA assay (Thermo Scientific). Protein (100 µg) was reduced with 5 mM DTT for 15 min at 37 °C and alkylated with 15 mM 2-iodoacetamide for 15 min in the dark. Proteins were tryptically digested with sequencing-grade trypsin in a 1:25 ratio for 2 h at 50 °C, followed by incubation at room temperature for 18 h. Subsequently, Rapigest was degraded by acidification. Peptides were cleared using the iST columns with triethylamine to ensure compatibility with TMT labelling (PreOmics, Martinsried, Germany). Samples were labeled using TMT-11-plex and fractionated by high-pH, reverse phase chromatography (Agilent 1100 HPLC). Dried samples were resolubilized in 0.1% formic acid and analyzed using an Orbitrap Q-Exactive Plus (Thermo, Bremen, Germany). Proteins were identified and quantified in three biological replicates per cell line. Data was analyzed using MaxQuant as described [
39,
40].
TCGA (The Cancer Genome Atlas) analysis
The TCGA data was accessed and downloaded with the TCGAbiolinks R package [
41]. For the analyses, the TCGA-BRCA cohort with the Gene Expression Quantification data type as well as corresponding mutation data was used.
After normalizing the data set, the distribution of the KLF7 gene expression was analyzed. Subsequently, the cohort was divided according to the quantiles of the distribution. For the gene set enrichment analysis, we compared the samples in the top 33% quantile to the samples in the bottom 33% quantile.
Additionally, the expression of KLF7 was compared between patients with a functional TP53 mutation to WT patients. Silent mutations are not considered functional and were therefore excluded. The mean expression values of each group were tested for statistical significance (t-test).
Statistical analysis
Statistics for proteomics were calculated using R4.0.3. Cell culture experiments were statistically analyzed using GraphPad Prism 5. Patient data were analysed using SPSS Version 27. Descriptive statistics with median and percentage of total, as well as estimated median survival, were calculated. The p-value for significance was defined < 0.05. For survival analysis, Kaplan Meier method was performed. Correlations between KLF7 expression (cytoplasmic, nuclear) and clinico-pathological features were calculated using Pearson, Spearman’s rho, and Kendall rank correlation.
Cell culture
MDA-MB-231 cells were grown in DMEM (Dulbecco’s modified Eagle’s medium)/F12 supplemented with 10% FCS and 1% penicillin/streptomycin (P/S). MCF7 were cultured in DMEM, 10% FCS, and 1% P/S. All cell lines were authenticated using Multiplex Cell Authentication by Multiplexion (Heidelberg, Germany), as recently described [
42]. The SNP profiles matched known profiles.
Stable KLF7-expressing MDA-MB-231 cells were generated by viral transduction from the Core Facility in the Signalhaus of the Albert-Ludwigs-University of Freiburg. Stable clones were selected, with puromycin and positive cells sorted according to fluorescent intensities. Stable KLF7-expressing MCF7 cells were generated by nucleofection (Nucleofector 2b, Lonza) followed by selection with 0.9 mg/ml G418.
qPCR
Total RNA was isolated using the Total RNA Purification Kit (#PP-210L, JenaBioscience) and cDNA generated using random primer mix (#S1330S, NEB) and Maxima Reverse Transcriptase (#EP0742, ThermoFisher Scientific) according to the manufacturer’s instructions. Real-time qPCR was performed with the PowerUp SYBR Green Master Mix (A25780, Applied Biosystems) on an Applied Biosystems QuantStudio 6-flex real-time PCR System. Primers were as following: KLF7_for AGCTACAACTTGTCCACGA, KLF7_rev ATTCAAGGCATGTCTGCTG, XPO1_for AGCAAAGAATGGCTCAAGAAGT, XPO1_rev TATTCCTTCGCACTGGTTCCT, NXF1_for AAGAGGCGGTTCTGGTATTCG, NXF1_rev TAGGGGTTGTATCGTACTCGG, NXT1_for CTTCCAGCGAGTTCCAAATCA, NXT1_rev CAGATGACAACAAGGACCGTG, NHP2_for CCCCACCTGTGTGATAATGGT, NHP2_rev GCACTCATCGTAAGCCTCCT, NOP10_for CAGTATTACCTCAACGAGCAGG, NOP10_rev GGCTGAGCAGGTCTGTTGTC, XRN1_for TCCAACTGTATCACACCAGGA, XRN1_rev GCTTTGCTTTCTCGGATCTGA, XRN2_for CCTTCGGCTTAATGTTCTTCGT, XRN2_rev TGAAAACCCAGTCATCAATGCT, NVL_for GAATTGTAGCCCAACTCCTAACC, NVL_rev GTCTGGTCGATTAGTAGCTCCA, POP1_for AGAGGTGTAAAGCACCACAGT, POP1_rev GCTGTCGTGAAGTTCCAGG, RMRP_for CGTAGACATTCCCCGCTTCC, RMRP_rev GCGTAACTAGAGGGAGCTGAC. ActB primers for normalization were from Primer design (HK-SY-HU-1200). Fold changes were calculated using the 2−ΔΔCt method.
Western blot
Cells were harvested and proteins extracted in RIPA buffer (150 mM NaCl, 1% NP-40, 0.5% Sodium deoxycholate, 50 mM Tris–HCl pH 8.0, 2 mM EDTA, 0.1% SDS) including 1× protease inhibitor (Complete Protease inhibitor cocktail, Roche) for 30 min, followed by centrifugation for 15 min (13,000 rpm). For western blotting, 5–10 µg of total protein was loaded on 4–15% Mini-PROTEAN TGX Precast Protein Gels (BioRad) and transferred to PVDF membranes at 100 V for 1 h. Membranes were blocked in 5% milk powder/TBST for 1 h. Primary antibodies were anti-KLF7 (ab197690, Abcam, 1:1000), anti-beta-Tubulin (MA5-16308, ThermoFisher, 1:3000). Secondary antibodies were anti-rabbit-HRP and anti-mouse-HRP (1:25000, JacksonImmunoResearch). Signals were detected using the SuperSignal West Femto Maximum Sensitivity Substrate (ThermoFisher). For reprobing, membranes were stripped with mild stripping buffer (Abcam, 0.2 M glycine, 0.1% SDS, 1% Tween, pH 2.2).
Proliferation assay
Proliferation was measured using the WST-8-based Rotitest Vital assay (Carl Roth). Briefly, cells were seeded at 5*103 cells/well density in 96 well plates. Every 24 h, 10 µl Rotitest Vital reagent was added to the cells and absorbance was measured at 450 nm in a Tecan M200 plate reader after 1 h of incubation. Medium was used as negative control. Each time point was measured in technical triplicates.
Cell cycle analysis
For cell cycle analysis, 106 cells were resuspended in 1 ml PBS and 2.5 ml 100% ice-cold ethanol added dropwise under vortexing for fixation. Cells were incubated overnight at 4 °C, washed with PBS, and stained with 50 µg/ml propidium iodide, 0.1% Triton-X100, and 100 mg/ml RNase A for 10 min at 37 °C. Stained cells were measured by flow cytometry (BD Bioscience). 20,000 cells were analyzed per condition. Analysis was performed with Kaluza Analysis 2.1.
Quantification of transcription and translation
Transcription and translation were quantified using Click chemistry and the CuAAC cell reaction buffer Kit (Jena Bioscience). 5*104 cells were seeded per well in a 96 well optical bottom plate and left to settle overnight. To metabolically label nascent RNA, 1 mM 5-ethynyl uridine (5-EU) in medium was added to the cells and incubated for 60 to 120 min. To measure translation, cells were incubated with 50 µM L-homopropargylglycine (L-HPG, Jena Bioscience) for 45 min. Cells were fixed with 3.7% formaldehyde for 15 min, washed with PBS, 3% BSA, and permeabilized with 0.5% Triton-X100/PBS for 20 min. 5-EU or L-HPG were visualized using the CuAAC Cell Reaction Buffer Kit (Jena Bioscience) according to the manufacturer’s instructions. Briefly, after washing with 3% BSA/PBS, cells were incubated with CLICK reaction cocktail containing 20 µM 5-TAMRA-Azide (Jena Bioscience), 33.33% CuSO4, 166.66 mM THPTA, and 100 mM Na-Ascorbate (Jena Bioscience) for 30 min at room temperature. Cells were washed twice with 3% BSA/PBS and twice with PBS and nuclei stained with DAPI for 2 min. Images were acquired on a AxioVision microscope (Carl Zeiss) using identical settings.
Patients and cohort
Core needle biopsies of 77 female breast cancer patients, initially diagnosed between 2004 and 2009 at the Department of Obstetrics and Gynecology, University Medical Center Freiburg were included in the study. Written informed consent was obtained from all patients before study inclusion. Ethics approval was obtained by local authorities of the Ethics Committee of the University Medical Center Freiburg (REF.: 523-19, 6.11.2019). To obviate interferences between neo-adjuvant therapeutic interventions and tumor biology, initial core needle biopsies were included. Subsequently, all patients were surgical treated at the Department of Obstetrics and Gynecology, University Medical Center Freiburg. Due to the small amount of tumor tissue within the biopsy, the immunohistologic subtype was adapted [
43] without including Ki67 staining.
KLF7 immunohistochemistry
Formalin fixed, paraffin embedded (FFPE) core needle biopsies were transferred on a tissue microarray (TMA). Microtome sliced in 2 µm sections were deparaffinized and pretreated in 0.1 M citrate buffer, pH 6.0 in a pressure cooker for 2 min for antigen retrieval. Subsequently, slides were washed in wash buffer (DAKO) followed by incubation with the primary antibody (anti-KLF7, HPA030490, SigmaAldrich) and incubation in H
2O
2 for 5 min, rabbit linker for 60 min, and horseradish peroxidase and secondary antibody for 20 min. Finally, slides were incubated with 3,3′-dianinobenzidine for 10 min. After hematoxylin counterstaining, slides were mounted in xylene. Immunostaining was evaluated with QuPath 0.2.3 [
44]. Tumor regions were annotated with object classifier, and nuclear and cytosolic positive staining quantified using the positive cell detection tool.
AgNOR staining
Silver staining of nucleoli was performed according to Trerè et al. [
45]. Briefly, cells of control and KLF7OE conditions were cultured on µ-slides (IBIDI), fixed with 100% ethanol at − 20 °C for 5 min, post-fixed in Carnoy’s solution (absolute ethanol: glacial acetic acid 3:1), and hydrated through graded alcohols to water. Cells were stained in one volume of pre-warmed 2% gelatin in 1% formic acid and two volumes of 50% silver nitrate solution at 37 °C for 12 min. FFPE TMAs were boiled in 0.1 M citric acid for 25 min in a pressure cooker and rinsed well in water before immersing in staining solution for 25 min. Slides were washed in distilled water, dehydrated, and mounted. Quantification was performed with QuPath 0.2.3 [
44]. Nuclei were annotated using the cell detection command, and nucleoli with the subcellular spot detection. For each condition, 5000 to 20,000 cells were analyzed.
Discussion
KLF7—a ubiquitously expressed protein—is strongly expressed in tumor tissues [
22]. We aimed to elucidate the molecular function of KLF7 in breast cancer and identified a novel and unexpected role of this protein in ribosomal processes and translation. Regulation of ribosomal biogenesis is one of the most energy-consuming cellular processes and is essential for the adaptation and functioning of cells in physiological and pathological conditions [
57]. In comprehensive transcriptomic and proteomic in vitro screens, we identified GO terms indicative of a regulatory role of KLF7 in ribosomal biogenesis. These findings were confirmed by in silico analyses of TCGA breast cancer data showing aberrant cellular processes related to ribosomal biogenesis in high KLF7-expressing breast cancer patients. Consistent with these results, breast cancer tissue and breast cancer cell lines expressing high levels of KLF7 featured disrupted nucleolar morphology and quantity. In vitro, KLF7 overexpression resulted exclusively in increased translation while proliferation and transcription remained unaffected.
To our knowledge, we have found the first connection between a member of the KLF/Sp family of transcription factors and target ribosomal biogenesis processes. This regulation might be important for physiologic and pathophysiologic pathways. So far, the molecular role of KLF7 has mainly been related to proliferation, differentiation, and migration [
20,
21,
23‐
26]. Our results are reminiscent of the wide-ranging importance of the transcription factor MYC, which not only shapes cellular processes such as differentiation, adhesion, or cell cycle through direct transcription and chromatin remodeling but also influences ribosomal biogenesis by regulating the rRNA transcription and ribosomal protein translation [
58,
59]. Our data indicate that KLF7 is similarly implicated in ribosomal biogenesis. The main significant GO terms that overlapped in our RNA-Seq and mass spectrometry data were ribosomal biogenesis, mRNA processing, and translation. Those terms were also identified as the most important altered processes related to KLF7 level in patient data from TCGA.
The exact mechanism of ribosomal regulation by KLF7 remains to be investigated. In MDA-MB-231 cells we demonstrated a downregulation RMRP, the RNA component of mitochondrial RNA processing endoribonuclease. RMRP is a non-coding RNA that binds to multiple proteins to form the RNase MRP complex, one of them is POP1 [
60]. POP1 has equally been found to be downregulated in MDA-MB-231 KLF7OE cells. The RNase RMP complex is involved in pre-rRNA processing essential for ribosomal biogenesis with an essential role of RMRP [
61]. Mutations in RMRP promotor have been correlated with breast cancer [
62] and POP1 has been identified as part of a prognostic signature in breast cancer [
63]. The changes in POP1 and RMRP expression in MDA-MB-231 indicate a regulatory role of KLF7 in this process in triple-negative cells. The luminal MCF7 cells reacted differently. MCF7 cells are—in contrast to MDA-MB-231—estrogen receptor (ER) positive. In MCF7 cells, ERalpha regulates RMRP which is not the case in ER negative MDA-MB-231 cells [
64]. Differences in expression between the MDA-MB-231 and MCF7 cell lines might therefore stem from the molecular subtype of the cell lines which is accompanied with a different tumor biology and patient outcome.
Another potential mechanism is the regulation of ribosomal stoichiometry. The existence of such a specialized cancer ribosome has been detected using methods based on RNA sequencing in human cancer tissue and cell lines [
4,
65]. Proteomic analyses in mouse embryonic stem cells also described a heterogeneous ribosomal composition [
4]. Interestingly, hereditary diseases that are caused by mutations in ribosomal proteins, the ribosomopathies, increase the risk of developing cancer [
66]. KLF7 might therefore play an important role in cancer progression.
Ribosomal biogenesis has been implicated in tumor growth and transformation and has recently been associated with the metastatic capacity of circulating breast cancer tumor cells [
5,
17]. In pathological diagnostics, prominent nucleoli as the sites of ribosomal biogenesis have been used for decades to distinguish malignant/tumorous from benign cells [
50,
51,
67]. Alterations in nucleolar number or shape have been described for several pathologic conditions [
68,
69]. Moreover, ribosomal biogenesis has been suggested as a prognostic marker in breast cancer [
70]. We have identified morphological aberrations in nucleoli in breast cancer tissue and cell cultures. In general, more nucleoli were detected in the triple negative cell line MDA-MB-231 compared to the hormone receptor-positive MCF7 cells. This is in line with a recent analysis in breast cancer tissue and cell lines, demonstrating a higher number of nucleoli in TNBC [
71]. KLF7 expression significantly increased nucleolar numbers.
We further identified KLF7 as a translation regulator. GO terms related to translation in proteomics and transcriptomics data and our in vitro experiments show that KLF7 increases translation independently of transcription. Moreover, in silico analysis of human tissue samples demonstrated that GO terms related to translation are significantly altered depending on KLF7 levels. Dysregulated translation efficiency represents a hallmark of cancer [
72]. Translation is not only related to proliferation but also to the cancer cell plasticity driven by stressors like hypoxia and energy deprivation [
73]. This change in translation is often decoupled from transcription but is controlled at the level of initiation, elongation, termination and protein folding – processes co-determined by the ribosome. As has been reported for Akt and Ras signaling in gliomas, KLF7 upregulates translation independently of increased transcription rate. In Ras/Akt signalling, specific cellular processes such as growth, transcription, and morphology regulation are most likely to be affected by this effect [
74]. Supposably, the effect of KLF7 might likewise apply to a subset of pathways, e.g., cytoplasmic processes such as cell substrate junctions or vesicle trafficking that we have identified by RNA-Seq. How these processes are regulated by KLF7 remains to be investigated. Ribosomal translation might be altered by changes in ribosomal stoichiometry. In bacteria, differing ribosomal composition leads to distinct translational profiles [
75]. Similar effects have been observed in eukaryotes, with important implications for pathologic conditions [
65,
76]. This is in line with the observation that ectopic expression of RPL15 increases overall expression in breast-circulating tumor cells [
17].
We further identified KLF7 as an important contributor to the aggressiveness of breast cancer. Nuclear KLF7 expression levels correlated significantly with the breast cancer subtypes and grading. Higher-graded tumors and more aggressive intrinsic tumor types, namely triple-negative breast cancer, showed high expression levels of KLF7. Triple-negative breast cancer is also more frequently associated with mutated TP53 and a dismal prognosis for the patients [
56]. In TP53-mutated tumors with higher p53 expression level, an increased area covered by nucleoli has been described and AgNOR staining correlated with tumor size and grading [
77]. Our data indicate that TP53 influenced KLF7 expression level. We speculate that in TP53-mutated cancer types, KLF7 levels are increased leading to alterations in ribosomal processes and more aggressive progression.
KLF7 was detected not only in the nuclei in breast cancer tissue but also in the cytoplasm, which corroborates observations of the Human Protein Atlas. Cytoplasmic KLF7 expression correlated with the intrinsic subtype and nuclear staining. A role of KLF7 in Golgi apparatus has been recently demonstrated in pancreatic cancer [
20]]. We speculate that the nuclear and cytoplasmic KLF7 expression might be involved in different cellular processes; the cytoplasmic KLF7 proteins could be involved in affecting the Golgi apparatus, while nuclear KLF7 could contribute to ribosomal regulation.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.