Overexpression of low-molecular-weight caldesmon is associated with aggressive phenotypes and epithelial-mesenchymal transition in breast cancer cells

Abstract

BACKGROUND

Caldesmon (CAD) has emerged as a promising biomarker of tumor progression and therapeutic response.

AIM

To investigate the expression and functional role of the low-molecular-weight isoform of CAD (l-CAD, about 65 kDa) in breast cancer.

METHODS

We analyzed the expression of l-CAD in breast cancer cell lines enriched for triple-negative breast cancer (TNBC) and in formalin-fixed, paraffin-embedded breast cancer tissues by western blotting and immunohistochemistry, respectively. This was followed by l-CAD silencing and functional analyses. In silico analyses were performed to validate the findings.

RESULTS

L-CAD expression was highly elevated in TNBC cell lines. Silencing of the CALD1 gene in the CAL-51 cell line resulted in the downregulation of mesenchymal markers, upregulation of epithelial markers, and a reduction in cell motility and invasiveness. Transcript-specific reverse transcription-polymerase chain reaction revealed that the low-molecular-weight transcript variant 2 (WI-38 l-CAD II) of CALD1 was the predominant isoform driving tumorigenesis in TNBC cell lines. Immunohistochemical analysis of breast cancer tissues confirmed significant l-CAD expression in the cytoplasm of cancer cells and stromal cells, with significant associations with lymphovascular invasion and human epidermal growth factor receptor 2, estrogen receptor, and progesterone receptor status. Survival analysis demonstrated that high CALD1 expression in TNBC was associated with poor prognosis and reduced disease-free survival.

CONCLUSION

Taken together, these findings highlight that l-CAD overexpression may drive metastasis in TNBC through epithelial-mesenchymal transition and support its relevance as a prognostic biomarker.

Key Words: Low-molecular-weight caldesmon; Breast cancer; Triple-negative; Epithelial-mesenchymal transition; Invasion; Metastasis

Core Tip: Low-molecular-weight caldesmon (l-CAD) plays a key role in triple-negative breast cancer (TNBC) progression by promoting epithelial-mesenchymal transition and enhancing cell motility and invasiveness. This study identifies CALD1 transcript variant 2 as the key isoform driving tumorigenesis in TNBC. Immunohistochemical analysis confirmed l-CAD expression in the cytoplasm of cancer cells and stromal cells, with significant associations with lymphovascular invasion and receptor status. High CALD1 levels in TNBC are associated with poor prognosis and reduced disease-free survival, highlighting l-CAD as a potential driver of metastasis and a promising prognostic biomarker.

INTRODUCTION

Cancer remains a devastating disease, with its incidence continuing to rise worldwide and almost half of the affected patients succumbing to their disease. In 2020, the global cancer burden reached an estimated 19.3 million new cases and 10 million cancer-related deaths[1]. Breast cancer continues to be the most common cancer in women, representing 1 in 4 cancer cases globally. Cancer progression, the principal driver of morbidity and mortality, is largely attributable to the ability of cancer cells to invade, metastasize, and destroy normal tissues. To accomplish this complex process, cancer cells must acquire the ability to migrate and invade through the hostile tumor microenvironment. These phenotypic changes are mediated by the accumulation of multiple genetic and epigenetic mutations and the activation of a plethora of signaling pathways, usually facilitated by a state of genomic instability[2].

Carcinomas, including breast cancers, originate from epithelial cells and characteristically lose their adhesion molecules that maintain cell-cell and cell-basement membrane interactions. Concurrently, they acquire mesenchymal features that allow them to invade and spread to other parts of the body in a process called epithelial-mesenchymal transition (EMT)[3,4]. The EMT program is orchestrated by an intricate network of signaling pathways, prominently involving transforming growth factor beta (TGFβ). During TGFβ1-induced EMT, the expression of cytoskeleton-associated proteins, including the actin-binding protein caldesmon (CAD), is significantly upregulated[5]. More recently, CAD was shown to play a key role in TGFβ-driven EMT in normal murine mammary epithelial cells. Nalluri et al[5] found that induction of EMT by TGFβ1 is mediated by increased CAD expression and phosphorylation, which is associated with increased focal adhesion number and size, as well as increased cellular contractility.

Cell movement, a critical process underlying cancer cell invasion of adjacent tissues, entry into the vasculature, and metastatic spread, depends on repeated cycles of actin filament assembly, cell-substrate adhesion, and acto-myosin-mediated contraction. In smooth muscle cells, regulation of the acto-myosin machinery involves both myosin-associated and actin-associated pathways. The myosin-associated regulatory mechanism primarily relies on Ca²⁺/calmodulin-dependent myosin light-chain kinase and its dephosphorylation by type 1 myosin phosphatase, which is directed to myosin through a specific regulatory subunit[6]. CAD and tropomyosin are crucial components in the actin-linked mechanism. CAD is an inhibitory factor for the actin-myosin interaction, in which the CAD induced inhibition can be released by Ca²⁺/calmodulin. CAD has been proposed to play a major role in cell motility by regulating the contractile system in both smooth muscle and non-muscle cells[7]. Mechanistically, CAD stabilizes actin filaments by binding along the sides of filamentous actin and enhances the binding of tropomyosin to actin[8].

The CALD1 gene encodes CAD, which exists in multiple isoforms. The high-molecular-weight isoform (h-CAD), with a molecular weight ranging from 120 kDa to 150 kDa, is specifically expressed in visceral and vascular smooth muscle cells. This isoform serves as a highly valuable marker for smooth muscle tissues and soft tissue tumors characterized by myofibroblastic differentiation. In contrast, the low-molecular-weight isoforms (l-CAD, 70-80 kDa) are predominantly expressed in non-smooth muscle cells[9-11].

CAD is conserved across nearly all vertebrate cells and regulates actin cytoskeleton assembly and dynamics, which is a significant cellular function. The non-muscle l-CAD is broadly implicated in many aspects of cell motility, including cell migration[12], focal adhesion assembly[13], and podosome dynamics[12,13]. Consequently, CAD has been identified as a promising regulatory factor that may govern key processes in tumor development, cell proliferation, and, as a result, therapeutic responsiveness. Its role in several solid cancers has been analyzed by diverse approaches, including clinical studies, bioinformatic analyses, and in vitro experimental models. However, these studies have yielded conflicting results regarding CAD expression patterns and functional roles across different cancer types. The vast majority of publications suggest an oncogenic role of CAD, particularly l-CAD, in many cancer types, including breast cancer[14], urinary bladder cancer[15,16], oral cavity squamous cell carcinoma[17], colorectal cancer (CRC)[18] including early-onset CRC[19], gastric cancer[20], and lung cancer[21]. Moreover, elevated serum levels of l-CAD have been reported in patients with glioma, suggesting its potential utility as a circulating biomarker for this malignancy[22]. The CALD1 gene undergoes alternative splicing in several cancer tissues, including colon, urinary bladder, and prostate, resulting in tumor-specific transcript variants[23]. Additionally, CAD expression has been linked to tamoxifen resistance in patients with estrogen receptor (ER)-positive recurrent breast cancer[24], as well as 5-fluorouracil resistance in those with locally advanced rectal cancer[25]. In contrast, a smaller number of studies have reported contradictory results. Following earlier reports that CAD is a cell motility suppressor[26,27], tumor suppressor roles for CAD have been shown in vitro in breast cancer and colon cancer cells[28], as well as in prostatic cancer cells[29]. CALD1 was also suggested to be a metastasis suppressor in gastric cancer[30], and its expression was associated with poor prognosis in bladder cancer in an in silico analysis[31].

Given this controversy, we investigated the expression of l-CAD in breast cancer, an aggressive malignancy with a high propensity for metastasis. We observed significant overexpression of l-CAD in triple-negative breast cancer (TNBC) cell lines and subsequently validated its expression in breast cancer tissues by immunohistochemistry. We further observed that l-CAD might play a key role in initiating cancer progression by promoting the EMT program and increasing cancer cell motility. Furthermore, we identified l-CAD transcript variant 2 as the main transcript variant associated with the aggressive phenotype of TNBC cells. Finally, in silico analyses demonstrated that CALD1 overexpression is associated with poor overall survival (OS) and reduced disease-free survival in patients with breast cancer.

MATERIALS AND METHODS

Cell lines and cell culture

A total of eight breast cancer cell lines were used. The HeLa cell line, an aggressive cervical cancer cell line, was used as a positive control for EMT. The breast cancer cell lines included MCF7 and its isogenic derivative 1001 (a tumor necrosis factor-resistant, TP53-mutant clone derived from the parental MCF7 cell line), CAL-51, BT-549, MDA-MB-231, MDA-MB-361, ZR-75-1, and T47-D. The cell lines were cultured in their respective growth media under specific conditions, as previously described[32]. Cell lines, molecular subtypes, and tissue origins of the breast cancer cell lines are described in Table 1.

Table 1 Breast cancer cell lines used in this study and their tumor origin[32,45].

Cell line	Tumor source	Breast cancer subtypes
MCF7	Adenocarcinoma of the mammary gland; cells were obtained from a metastatic site (pleural effusion)	Luminal A (ER+ and/or PR+ HER2-)
MDA-MB-361	Adenocarcinoma of the mammary gland; cells were obtained from a metastatic site (brain)	Luminal (ER+, PR-, and HER2+ amplified)
ZR-75-1	Ductal carcinoma of the mammary gland; cells were obtained from a metastatic site (ascites)	Luminal A (ER+ and/or PR+ HER2-)
CAL-51	Adenocarcinoma isolated from malignant pleural effusion of metastatic breast cancer, normal karyotype with genetic stability	Triple-negative (ER-, PR-, HER2-)
1001	Derived from parental MCF7 (MCF7/R-A1), which are cells exposed to increasing doses of recombinant TNF, transfected with p55 TNF receptor cDNA, mutation in R280K	Luminal A (ER+ and/or PR+ HER2-)
MDA-MB-231	Adenocarcinoma of the mammary gland, cells were obtained from the metastatic site; pleural effusion	Basal subtype receptor status: Triple-negative (ER-, PR- HER2-)
T47D	Ductal carcinoma of the mammary gland, cells were obtained from the metastatic site; pleural effusion	Luminal B (ER+ and/or PR+ HER2+/-)
BT-549	Ductal carcinoma of the mammary gland, cells were obtained from the mammary gland	Basal B subtype receptor status: Triple-negative (ER-, PR-, HER2-)

BC: Breast cancer; ER: Estrogen receptor; PR: Progesterone receptor; TNF: Tumor necrosis factor.

Total protein extraction and western blot analysis

Cell lysates were prepared using Triton lysis buffer (25 mmol/L Tris-HCl, pH 7.4; 150 mmol/L NaCl; 1 mmol/L ethylenediaminetetraacetic acid; 1% Triton X-100) supplemented with 50 mmol/L sodium fluoride, 1 mmol/L sodium orthovanadate, 1 mmol/L phenylmethylsulfonyl fluoride, and 2% protease inhibitor cocktail. Cultured cells were harvested and washed twice with ice-cold 1 × phosphate-buffered saline, and the resulting cell pellets were lysed using pre-chilled Triton lysis buffer. Then, the lysed pellets were incubated on ice for 5 minutes and centrifuged at 14000 rpm for 10 minutes at 4 °C to remove cell debris. Protein concentration was determined using the bicinchoninic acid protein assay (Thermo Fisher Scientific, Waltham, MA, United States). Approximately 30 μg of total protein was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis consisting of an 8.5% resolving gel and a 3% stacking gel. Subsequent procedures were performed as previously described[32]. Proteins were transferred onto a polyvinylidene difluoride membrane (Bio-Rad, Hercules, CA, United States) using the Trans-Blot^® Turbo™ Blotting system (Bio-Rad, Hercules, CA, United States) according to the manufacturer’s standard protocol. Membranes were blocked for 1 hour at room temperature using 5% non-fat dry milk (5 g of powder) (Sigma-Aldrich, St Louis, MO, United States) prepared in 1 × Tris-buffered saline with Tween-20 (TBST; TBS + 1% Tween), followed by incubation with the primary antibody against l-CAD (1:500 dilution; Thermo Fisher Scientific, Waltham, MA, United States) overnight at 4 °C with gentle agitation. A mouse monoclonal anti-b-actin antibody (Sigma-Aldrich, St Louis, MO, United States) was used as a loading protein for normalization. EMT markers were detected using an EMT antibody sample kit (Cell Signaling Technology, Danvers, MA, United States), including rabbit monoclonal antibodies against N-cadherin (clone D4R1H), ZO-1 (clone D7D12), Snail (clone C15D3), and E-cadherin (clone 24E10). All primary antibodies against EMT markers were used at a 1:1000 dilution and incubated overnight at 4 °C with gentle agitation.

Following primary antibody incubation, membranes were washed four times each for 10 minutes with 1 × TBST and incubated with horseradish peroxidase-conjugated anti-rabbit IgG secondary antibody. After three additional washes with 1 × TBST, membranes were incubated with enhanced chemiluminescence substrate (Pierce™ ECL Western Blotting Substrate; Thermo Fisher Scientific, Waltham, MA, United States) for 1 minutes. Signals were detected by exposure to X-ray film (Kodak Cl-XPosure™ Film, Catalog No. 34090; Thermo Fisher Scientific, Waltham, MA, United States), followed by film development. Band intensities were quantified using ImageJ software and normalized to β-actin. Statistical analysis of densitometric data was performed using GraphPad Prism version 10.

Immunohistochemistry

An anti-l-CAD monoclonal antibody (CAD low-molecular-weight antibody, Monoclonal Antibody 5-11777; Invitrogen, Thermo Fisher Scientific, Waltham, MA, United States) was used at a dilution of 1:500. Four-micrometer sections from formalin-fixed, paraffin-embedded (FFPE) tissue blocks were de-waxed and rehydrated in distilled water. Antigen retrieval was performed by heat-induced epitope retrieval in 1 mmol/L ethylenediaminetetraacetic acid buffer (pH 8.0) using a microwave oven at 750 W for 5 minutes, followed by 450 W for 5 minutes. After cooling, the slides were washed in TBST (pH 7.2).

Immunohistochemical staining was performed manually using the Dako EnVision + System, Peroxidase, according to the manufacturer’s instructions (Dako, Glostrup, Denmark). The staining area was manually marked using a Dako pen (Agilent Technologies, Santa Clara, CA, United States) before staining. After blocking endogenous peroxidase activity, and before incubation with the primary antibody, the sections were incubated with 10% normal (non-immune) goat serum (Dako, Glostrup, Denmark) for 30 minutes. Sections were then incubated with the primary antibody at 4 °C. Paired tumor and adjacent normal mucosa were present on the same slide, with normal epithelial tissues and stroma serving as internal reference controls for staining quality. Immunohistochemical procedures and scoring were performed as previously described by AlNuaimi et al[33], and immunoreactive scores were assigned accordingly. Patient clinicopathological characteristics are summarized in Table 2. The median age of the patients was 45 (range: 31-68) years.

Table 2 Breast cancer samples and their pathological characteristics, n = 51.

Variables	n (%)
Age in years
< 40	9 (18.0)
> 40	42 (82.0)
Tumor stage
1	None
2	15 (29.4)
3	32 (62.7)
4	4 (7.8)
Molecular subtype
Luminal A (E2+/PR+) HER2-	36 (70.6)
Luminal B (E2+/PR+) HER2+	4 (8.0)
HER2-enriched	6 (12.0)
Triple-negative	5 (10.0)
Ki-67 index
≤ 50%	21 (41.0)
> 50%	30 (59.0)
Lymphovascular invasion
No	29 (57.0)
Yes	22 (43.0)
In situ component
No	28 (55.0)
Yes	23 (45.0)

ER: Estrogen receptor; PR: Progesterone receptor.

Gene silencing using small interfering RNA

Small interfering RNA (siRNA) targeting full-length CALD1 protein (Catalog No. 4392421; Invitrogen, Thermo Fisher Scientific, Waltham, MA, United States) was used for gene silencing. CAL-51 and BT-549 cell lines were transfected with a CALD1 siRNA or a negative control siRNA (Silencer™ Select Negative Control, Catalog No. 4390844; Invitrogen, Thermo Fisher Scientific, Waltham, MA, United States) under the same conditions using Lipofectamine^® RNA iMAX (Catalog No. PIN56532; Invitrogen, Thermo Fisher Scientific, Waltham, MA, United States), following the manufacturer’s protocol (available from https://assets.thermofisher.com/TFSAssets/LSG/manuals/Lipofectamine_RNAiMAX_Reag_protocol.pdf).

Briefly, (0.25-1) × 10⁶ cells of CAL-51 and BT-549 were seeded in 6-well plates to achieve 60%-80% confluence at transfection. A total of 9 μL of Lipofectamine^® RNAiMAX reagent was diluted in 150 μL of Opti-MEM^® medium, and 3 μL of siRNA (30 pmol; 10 μM) was separately diluted in 150 μL of Opti-MEM^®. The diluted siRNA and Lipofectamine^® RNAiMAX reagent were combined at a 1:1 ratio and incubated for 5 minutes at room temperature to allow complex formation. Subsequently, 250 μL of the siRNA-lipid complex was added to each well. The final amounts per well were 25 pmol siRNA and 7.5 μL of Lipofectamine^® RNAiMAX.

Cells were incubated for 48 hours at 37 °C, after which the transfected cells were harvested for downstream analyses, including invasion assays and western blotting to assess EMT marker levels after protein extraction.

Invasion assay

For the invasion assay, the QCM™ High Sensitivity Non-Cross-Linked Collagen Invasion Assay kit (Millipore, Burlington, MA, United States) was used according to the manufacturer’s protocol and as previously described[32]. Briefly, the assay was performed using a modified Boyden chamber with 8-μm pore-size filter inserts coated with Matrigel and placed in 24-well plates. Approximately 5 × 10⁵ cells were resuspended in serum-free media (DMEM RNBH2833), and 250 μL of the cell suspension was added to the upper chamber of each insert. The lower chamber was filled with 500 μL of complete medium containing 15% fetal bovine serum as a chemoattractant.

After 48-hour incubation at 37 °C, cells remaining in the top chamber were removed, and invaded cells were stained by adding 400 μL of cell stain to each insert for 15 minutes. After several washes with water, the inserts were air-dried, examined under an inverted microscope at × 5 magnification (1-mm field), and representative images were captured. Inserts were then transferred to 200 μL of extraction buffer and allowed to incubate for 15 minutes at room temperature. The extracted dye was quantified using a microplate reader at a wavelength of 492 nm.

Reverse transcription-polymerase chain reaction

Total RNA was extracted using a Qiagen RNA extraction kit (Hilden, Germany) according to the manufacturer’s instructions. Reverse transcription was performed using the QuantiTect Reverse Transcription kit (Qiagen). Reverse transcription-polymerase chain reaction (RT-PCR) was performed using a 5 × HOT FIREPol^® Blend Master Mix (Solos BioDyne, Tartu, Estonia) under the following conditions: Initial denaturation at 95 °C for 5 minutes, followed by 35 cycles of denaturation at 94 °C for 30 seconds, annealing at 56 °C for 50 seconds, and extension at 72 °C for 90 seconds, with a final extension at 72 °C for 60 minutes.

Primers were purchased from e-Oligos and synthesized by Genotech (Daejeon, South Korea), and their sequences were as follows: Reverse Pm, 5′-GTTTAAGTTTGTGGGTCATGAATTCTCC-3′; forward Pn1, 5′-ATGCTGGGTGGATCCGGATC-3′; and forward Pn2, 5′-ATGGATGATTTTGAGCGTCG-3′. These primers were first used by Lee et al[16].

Polymerase chain reaction (PCR) amplicons were resolved by gel electrophoresis on 8% agarose gels with 0.5 μg/mL of ethidium bromide. A GeneDireX, Inc DNA Ladder (100-3000 bp) was used as a molecular weight marker, and gel images were captured using a Gel Doc imaging system (Bio-Rad, Hercules, CA, United States).

Sequencing analysis and sequence alignment

Sequencing analysis was performed by MCLAB (San Francisco, CA, United States). A 10-μL aliquot of each PCR product was sent to Molecular Cloning Laboratories for Sanger sequencing analysis using an ABI 3730XL Genetic Analyzer. Sequence alignment and identity verification were conducted using the BLAST nucleotide alignment tool (nblast) available through the National Center for Biotechnology Information at https://blast.ncbi.nlm.nih.gov/Blast.cgi?PAGE_TYPE=BlastSearch.

In silico analysis of CAD in breast cancer

CAD mRNA expression in breast cancer samples was evaluated using the online Kaplan-Meier Plotter database (www.kmplot.com), which integrates gene expression data and survival information of patients with breast cancer (https://kmplot.com/analysis/index.php?p=service#) (accessed on June 2, 2024). The Kaplan-Meier Plotter was used to assess OS, with a specific focus on TNBC. Analysis parameters included negative immunohistochemical status for ER, progesterone receptor (PR), and HER2, as well as positive lymph node metastasis, resulting in a cohort of 189 patients with TNBC. All available probe sets for CALD1 were selected to generate Kaplan-Meier survival curves. Hazard ratios with 95% confidence intervals and log-rank P values were calculated, with P < 0.05 considered statistically significant. The number of patients at risk was displayed below each survival plot.

Additionally, CALD1 gene expression in breast cancer was analyzed using The Cancer Genome Atlas (TCGA) dataset (n = 1068). OS and disease-free survival were evaluated and plotted using the Gene Expression Profiling Interactive Analysis platform at http://gepia.cancer-pku.cn/index.html (accessed on May 24, 2024).

Statistical analysis

Statistical analyses were performed using Fisher’s exact probability test, the χ² test, or Student’s t-test, as appropriate, to evaluate differences between groups. Correlation and regression analyses were performed to assess relationships between variables. All statistical analyses were performed using MS Excel, the VassarStats web-based statistical program (http://faculty.vassar.edu/Lowry/VassarStats.html), and GraphPad Prism version 10. All reported P values were two-tailed, and P < 0.05 was considered statistically significant.

Densitometric quantification of western blot bands was performed using ImageJ software, with protein expression normalized to β-actin. Data are presented as mean ± SD from at least three independent experiments. Statistical significance is denoted as ^aP < 0.05, ^bP < 0.01, and ^cP < 0.001.

Pearson’s correlation coefficients were calculated using the CORREL function in MS Excel (Data Analysis ToolPak) to assess the strength and direction of linear associations between quantitative variables. A correlation coefficient of + 1 indicated a perfect positive correlation, which means that as variable X increased, variable Y increased proportionally. Conversely, a correlation coefficient of -1 indicated a perfect negative correlation, which means that as variable X increased, variable Y decreased proportionally. Kaplan-Meier survival analysis and log-rank tests for OS were performed using the Gene Expression Profiling Interactive Analysis platform (http://gepia.cancer-pku.cn/index.html) (accessed on May 24, 2024).

RESULTS

L-CAD expression in breast cancer cell lines

We first assessed l-CAD protein expression in various molecular subtypes of breast cancer cell lines by western blot analysis using a specific antibody against l-CAD. The HeLa cell line, an aggressive cervical cancer line, was used as a positive control to represent an invasive cancer phenotype. We found that l-CAD expression was high in five out of eight (63%) breast cancer cell lines examined. Notably, the highest l-CAD expression was observed in TNBC breast cancer cell lines, specifically CAL-51, BT-549, and MDA-MB-231 (Figure 1A). Among these, the CAL-51 cell line exhibited the highest l-CAD expression. The HeLa cell line also showed high expression, comparable to the range observed across the other breast cancer cell lines (Figure 1B).

Figure 1 Differential expression of the low-molecular-weight isoform of caldesmon in breast cancer cell lines. Western blot analysis of the low-molecular-weight isoform of caldesmon (l-CAD) expression in breast cancer cell lines is shown. A: Representative western blot results showing the expression of l-CAD in breast cancer cell line MCF7 and its isogenic derivative 1001, ZR-75-1, CAL-51, MDA-MB-231, BT-549, MDA-MB-361, and T47D cell lines. HeLa cells were used as a positive control. β-actin was used as a loading control; B: Densitometric quantification of l-CAD band intensity normalized to b-actin, presented as a bar graph. l-CAD: Low-molecular-weight isoform of caldesmon.

Silencing of l-CAD using CALD1 siRNA in TNBC cell lines

Given the high l-CAD expression observed in the TNBC cell lines CAL-51 and BT-549 by western blot analysis, we next evaluated the efficiency of CALD1 silencing in these cells (Figure 2). We found a significant reduction in l-CAD protein levels in siRNA-transfected CAL-51 cells compared with untransfected controls (P < 0.0005). In contrast, knockdown of l-CAD in transfected BT-549 cells was less pronounced, with a statistically significant but smaller reduction (P < 0.0151) (Figure 2B). Quantitative analysis revealed an approximately 81% decrease in l-CAD expression in transfected CAL-51 cells, compared with a 38% reduction in transfected BT-549 cells (Figure 2C). Thus, the silencing efficiency of l-CAD was more significant in transfected CAL-51 cells than in transfected BT-549 cells. Therefore, we selected the CAL-51 cell line for subsequent functional analysis after CALD1 silencing.

Figure 2 Silencing of CALD1 in CAL-51 and BT-549 breast cancer cell lines. Small interfering RNA (siRNA)-mediated silencing of CALD1 resulted in a more pronounced reduction (81%) of low-molecular-weight isoform of caldesmon (l-CAD) expression in CAL-51 cells compared with BT-549 cells. A: Representative western blots showing l-CAD expression following transfection with CALD1-specific siRNA in CAL-51 and BT-549 cells. Cells transfected with negative control siRNA and Lipofectamine^® served as controls; B: Relative l-CAD protein expression quantified as the ratio of l-CAD to β-actin; C: Percentage silencing efficiency of l-CAD in CAL-51 and BT-549 cells. ^aP < 0.0151; ^bP < 0.0005. l-CAD: Low-molecular-weight isoform of caldesmon; siRNA: Small interfering RNA; N.C: Negative control small interfering RNA and Lipofectamine^® served as controls.

EMT after CALD1 silencing

Following CALD1 silencing, we examined the expression of EMT markers by western blot analysis (Figure 3). Our results showed an upregulation of the epithelial markers E-cadherin and ZO-1 in siRNA-transfected CAL-51 cells, while these markers were minimally or not detected in untransfected CAL-51 cells (Figure 3A). In contrast, the mesenchymal markers N-cadherin and Snail were strongly expressed in untransfected CAL-51 cells, in cells transfected with negative control siRNA, and in the aggressive HeLa cell line; however, their expression was markedly reduced or absent in CALD1-silenced CAL-51 cells (Figure 3B). Densitometric analysis demonstrated a significant increase in E-cadherin (P < 0.0085) and ZO-1 (P < 0.0026) expression in transfected CAL-51 cells compared with untransfected CAL-51 cells (Figure 3C). Conversely, N-cadherin (P < 0.0013) and Snail (P < 0.0118) expressions were significantly downregulated following CALD1 knockdown (Figure 3D). Taken together, our findings showed a significant loss of the mesenchymal markers N-cadherin and Snail and a significant gain of the epithelial markers E-cadherin and ZO-1 as a result of CALD1 gene silencing.

Figure 3 Loss of mesenchymal markers and gain of epithelial markers after CALD1 silencing in CAL-51 cells. A: Western blot analysis showing increased expression of the epithelial markers E-cadherin and ZO-1 in CALD1-silenced CAL-51 cells (small interfering RNA-CALD1) compared with negative control-transfected cells; B: Western blot analysis of mesenchymal markers N-cadherin and Snail in breast cancer cell lines. HeLa cells were used as a positive control for epithelial - mesenchymal transition epithelial-mesenchymal transition, and MCF-7 cells were included as a less aggressive breast cancer cell line; C and D: Densitometric quantification of epithelial and mesenchymal marker expression normalized to β-actin. ^aP < 0.05; ^bP < 0.01. N.C: Controls; EMT: Epithelial-mesenchymal transition.

CALD1 silencing reduced collagen invasiveness

We then assessed the invasive capacity of CALD1-silenced CAL-51 cells using a collagen invasion assay (Figure 4). Microscopic examination of the untransfected CAL-51 cells revealed that the cells displayed a high number of invasive spindle-shaped cells with aggressive features (increased nuclear size, clumped chromatin, increased cytoplasmic extensions) consistent with a mesenchymal phenotype (Figure 4A). Similar invasive characteristics were observed in cells transfected with negative control siRNA (Figure 4C). In contrast, CALD1-silenced CAL-51 cells showed a marked decrease in the number of cells as well as the invasiveness characteristics (Figure 4B). Photometric analysis confirmed that there was a significant reduction in the number of invading cells after transfection compared with untransfected parental cells (P < 0.0080) (Figure 4D).

Figure 4 Functional consequences of CALD1 silencing on invasion in CAL-51 breast cancer cells. CALD1 silencing significantly reduced collagen invasiveness in the metastatic CAL-51 cell line, as assessed by an invasion assay. A: Representative images showing high invasive capacity of untransfected CAL-51 cells, characterized by a large number of invading cells; B: Marked reduction in invasive cells following CALD1 silencing in CAL-51 cells (small interfering RNA-CALD1); C: Cells transfected with negative control small interfering RNA exhibited invasive characteristics similar to untransfected CAL-51 cells (× 100 magnification); D: Quantitative analysis of invasion measured by absorbance at 492 nm. ^aP < 0.0080. N.C: Negative control small interfering RNA; siRNA: Small interfering RNA.

L-CAD expression in breast cancer FFPE tissues by immunohistochemistry

To assess the clinical significance of l-CAD as a potential biomarker in breast carcinoma, we examined its expression in FFPE tissue specimens obtained from a cohort of 51 breast cancer patients using immunohistochemistry (Figure 5A-F). Associations between CAD expression levels and clinicopathological features of invasive breast carcinoma are shown in Figure 6. There was a positive correlation between l-CAD expression and HER2-positive breast cancer, lymphovascular invasion, the presence of an in situ component, Ki-67 index, and mitotic index. In contrast, a negative correlation was observed between l-CAD expression and tumor stage, ER status, and PR status.

Figure 5 Immunohistochemical expression of the low-molecular-weight isoform of caldesmon in formalin-fixed, paraffin-embedded tissues from resected breast cancers. A: Luminal A breast cancer sample showing strong cytoplasmic low-molecular-weight isoform of caldesmon (l-CAD) positivity; B: Sample from luminal A breast cancer showing negative l-CAD staining; C: HER-2 sample showing strong positive l-CAD expression; D: HER-2 sample showing moderate l-CAD expression; E and F: Triple-negative breast cancer sample showing positive to strong positive l-CAD expression.

Figure 6 Correlation between low-molecular-weight isoform of caldesmon immunohistochemical intensity and clinicopathological features of invasive breast carcinoma. Graph shows the correlation between the low-molecular-weight isoform of caldesmon (l-CAD) immunohistochemical scores in formalin-fixed, paraffin-embedded breast cancer and clinicopathological parameters. Positive correlation coefficients (> 0) indicate associations between l-CAD staining intensity and HER2-enriched breast cancer, lymphovascular invasion, and the presence of an in situ component. Negative correlation coefficients (< 0) indicate inverse associations between l-CAD expression and estrogen and progesterone receptor status.

Table 3 shows the immunohistochemical scores based on the number of positive cells and staining intensity of l-CAD. L-CAD expression was positive in 28 out of 51 patients (55%) and negative in 23 out of 51 patients (45%). Positive expression of l-CAD was significantly associated with adverse clinicopathological characteristics, including estrogen and progesterone (P = 0.0056), HER2 expression (P = 0.0003), and lymphovascular invasion (P = 0.0259). A trend toward association was observed with the presence of an in situ component (P = 0.0565). There was no significant association between the expression of l-CAD and other histological parameters.

Table 3 Low-molecular-weight isoform of caldesmon expression in breast cancer cases and its association with clinicopathological characteristics, n = 51, n (%).

Parameter	Category	Total	L-CAD-negative	L-CAD-positive	P value
Age in years	< 40	9 (18)	4	5	-
	≥ 40	42 (82)	19	23
Histological stage	Early: 1-2	15 (29)	6	9	-
	Late: 3-4	36 (71)	17	19
Ki-67 index	< 50%	21 (41)	11	10	-
	≥ 50%	30 (59)	12	18
Lymphatic invasion	Absent = 0	29 (57)	17	12	0.025865
	Present = 1	22 (43)	6	16
In situ component	Absent	28 (55)	16	12	0.056477
	Present	23 (45)	7	16
HER2 status	HER2+	10 (20)	2	8	0.000277
	HER2-	41 (80)	19	22
E2/PR	E2+/PR+	31 (61)	16	15	0.005555
	E2+/PR-	3 (6)	2	1
	E2-/PR+	5 (10)	2	3
	E2-/PR-	12 (24)	2	10

Immunohistochemistry was performed on 51 formalin-fixed, paraffin-embedded breast cancer specimens using a semi-quantitative scoring system. P values were calculated using MS Excel and the Vassar Stats Web-based statistical program. All P values were two-tailed, and P values < 0.05 were considered statistically significant. ER: Estrogen receptor; l-CAD: Low-molecular-weight isoform of caldesmon; PR: Progesterone receptor.

Identification of the l-CAD transcript variant associated with breast cancer tumorigenesis by RT-PCR and sequencing analysis

Next, RT-PCR and sequence analysis were used to identify the l-CAD variant expressed in CAL-51, BT-549, and MDA-MB-231 cells. RT-PCR was performed using two forward primers, Pn1 and Pn2, in combination with the reverse primer PM[17]. The Pn1 primer detects CALD1 transcript variants 4 and 5, while the Pn2 primer detects transcript variants 1, 2, and 3. The results of RT-PCR of cDNA from all three breast cancer cell lines did not show any amplification with the Pn1/PM primer set (expected size: 700-800 bp). In contrast, positive bands were detected in all breast cancer cell lines using the Pn2/PM primer set. To identify the exact transcript variant amplified by Pn2/PM primers, the amplified products were sequenced and analyzed using BLAST sequence alignment. Sequence analysis revealed 97%, 99%, and 98% identity in CAL-51, BT-549, and MDA-MB-231 cells, respectively, with Homo sapiens CALD1 transcript variant 2 (mRNA Sequence ID: NM_004342.7). These results indicate that the transcript variant of l-CAD responsible for tumorigenesis in the selected TNBC cell lines is indeed transcript variant 2 (Supplementary Figure 1).

In silico analysis of CALD1 expression and clinical associations

High CALD1 expression and overall survival in invasive breast carcinoma: To validate our findings in a separate cohort and assess their generalizability, we analyzed CALD1 gene expression in invasive breast carcinoma using the TCGA dataset (n = 1070). Kaplan-Meier survival analysis showed that patients with high CALD1 expression showed a slightly better OS (Figure 7A). In contrast, high CALD1 expression was associated with poorer disease-free survival after 170 months of follow-up (Figure 7B).

Figure 7 In silico analysis of CALD1 expression in breast cancer using The Cancer Genome Atlas dataset. Analysis of CALD1 gene expression was performed using The Cancer Genome Atlas breast cancer cohort (n = 1070) via the Gene Expression Profiling Interactive Analysis platform at http://gepia.cancer-pku.cn/index.html (accessed on May 24, 2024). A: Kaplan-Meier survival curve shows the overall survival analysis of the cases of The Cancer Genome Atlas according to CALD1 gene expression status. The solid line represents the survival curve, and the dotted line represents the 69% confidence interval (hazard ratio = 0.94; P = 0.69, n = 1070); B: Disease-free survival analysis according to CALD1 expression status. Solid lines represent disease-free survival, and dotted lines represent the 95% confidence interval (hazard ratio = 0.97; P = 0.85, n = 1070). Median expression was used as the cutoff for group stratification. HR: Hazard ratio; TPM: Transcript per million; CAD: Caldesmon.

High expression of CALD1 in TNBC is associated with poor prognosis: The prognostic value of CALD1 in TNBC was assessed using the Kaplan-Meier plotter tool. Selection criteria were applied to mirror our experimental findings, including negative ER, PR, and HER2 status, along with positive lymph node involvement. To comprehensively investigate the correlation between CALD1 expression and OS, all available probe sets for CALD1 were included to generate Kaplan-Meier survival curves. Across multiple Affymetrix probe sets (205525_at, 214880_x_at, 215198_s_at, 215199_at, 201615_x_at, 201616_s_at, 201617_x_at, 212077_at), elevated CALD1 expression was consistently and significantly associated with poorer OS in patients with TNBC (P < 0.05; Figure 8).

Figure 8 Prognostic significance of CALD1 expression in triple-negative breast cancer. Kaplan-Meier Plotter analysis showing the association between CALD1 expression and overall survival in patients with triple-negative breast cancer selected based on molecular subtypes (estrogen receptor-negative, progesterone receptor-negative, HER2-negative) and lymph node-positive status. Survival analyses were generated using all available probe sets per gene across eight Affymetrix probe IDs. High expression of CALD1 is represented by the red curve, whereas low expression of CALD1 is represented by the black curve. Each panel displays the corresponding hazard ratio and P value. P < 0.05 indicates statistical significance. HR: Hazard ratio.

Comparison of l-CAD expression with clinical parameters using the TCGA dataset: To validate the associations observed in our cohort between l-CAD expression and patient clinical characteristics, we analyzed CALD1 gene expression data from the TCGA invasive breast carcinoma dataset. Correlations between CALD1 expression and HER2 (ERBB2), ER (ESR1), PR (PGR), and histological stage were examined (Figure 9A-D). We found a high degree of correlation between CALD1 expression and ERBB2 (HER2) (P = 0.00024, R = -0.14), ESR1 (ER) (P = 3.4 × 10^-22, R = -0.26), and PGR (PR) (P < 0.001, R = 0.34). These findings are consistent with our immunological data demonstrating significant associations between l-CAD expression and HER2 (P = 0.0003) as well as ER/PR status (P = 0.0056) (Table 3).

Figure 9 Correlation between CALD1 expression and clinicopathological features of breast cancer in The Cancer Genome Atlas cohort. Correlation analyses between CALD1 gene expression and clinical parameters were performed using The Cancer Genome Atlas breast cancer dataset via the Gene Expression Profiling Interactive Analysis platform (http://gepia.cancer-pku.cn; accessed on June 4, 2024). A: High degree of correlation between CALD1 and ERBB2/HER2 expression (P = 0.00024, R = -0.14); B: High degree of correlation between CALD1 and ESR1/E2 expression (P = 3.4 × 10^-22, R = -0.26); C: High degree of correlation between CALD1 and PGR expression (P = 0, R = 0.34); D: Association between CALD1 and pathological stage in breast cancer. TPM: Transcript per million.

In addition, we compared the association between l-CAD expression and tumor stage in our cohort with TCGA data. Consistent with our findings, TCGA analysis showed no significant correlation between CALD1 expression and pathological stage (Figure 9D), further validating our results.

DISCUSSION

In this study, we investigated the expression of l-CAD (about 65 KDa) using breast cancer cell lines, patient-derived breast cancer tissues, and in silico datasets. Collectively, our results show that l-CAD is highly expressed in TNBC and its overexpression is associated with poor survival and well-defined molecular features. We identified EMT as the mechanism through which l-CAD exerts its carcinogenic effects and further determined that CALD1 transcript variant 2 is the predominant isoform associated with breast cancer.

TNBCs are associated with broadly poor clinical outcomes, with affected patients experiencing early relapse and high mortality rates[34,35]. Consistent with this aggressive behavior, our western blot analyses demonstrated high l-CAD expression in TNBC cell lines, including CAL-51, BT-549, MDA-MB-231, and MDA-MB-361. Importantly, we identified CALD1 transcript variant 2 as the key functional mediator for pro-tumorigenic activity in all these TNBC cell lines. Revealing the specific transcript variant in our study was crucial for understanding disease pathogenesis, as previous studies have either focused on particular l-CAD transcripts[16] or failed to specify the transcript variant studied[30].

In order to understand the functional role of l-CAD in TNBC, we silenced CALD1 in CAL-51 and BT-549 cells to study its role in cellular functions, including invasion and EMT. These two cell lines showed different degrees of l-CAD suppression after gene silencing. This variability may reflect the intrinsic heterogeneity of breast cancer, which poses important clinical challenges[34,36]. Intratumor heterogeneity in breast cancer involves the presence of diverse cell populations within a single tumor, differing genetically, epigenetically, and phenotypically. This diversity results from ongoing genetic mutations, epigenetic changes, and environmental factors during tumor development. It plays a crucial role in treatment resistance, disease progression, and metastasis. Different tumor regions may harbor unique mutational profiles, leading to subclonal populations, while individual tumor cells can differ markedly in growth rate, invasiveness, and therapeutic response[36].

We further observed a significant downregulation of the mesenchymal markers N-cadherin and Snail following CALD1 silencing in CAL-51 cells, accompanied by increased expression of the epithelial markers E-cadherin and ZO-1. The transition of epithelial cancer cells toward a mesenchymal phenotype is widely considered a pivotal event in the metastatic process. Given that metastasis is responsible for nearly 90% of cancer-related mortality, EMT stands out as one of the most concerning and well-recognized mechanisms driving cancer progression.

Stimulation by growth factors such as TGFβ or epithelial growth factor leads to the initiation of EMT by activating signaling pathways including Wnt and Notch, resulting in downstream induction of transcription factors such as Snail, ZEB, Smad, and Twist[36-38]. In subsequent steps, epithelial cadherin (E-cadherin) is degraded, which leads to loss of cell-cell adhesion and disruption of its interaction with β-catenin. Moreover, cells gain mesenchymal markers such as N-cadherin and vimentin, endowing them with the capacity for invasion and metastasis. These mechanisms contribute to cell migration and invasion, which are hallmarks of EMT[39,40].

Notably, our findings indicate that all of these hallmark EMT-associated changes are closely linked to l-CAD expression in TNBC cells. Studies at both the biochemical and structural levels indicate that when CAD is phosphorylated at Erk-targeted sites, it undergoes a conformational shift that leads to its partial release from actin filaments[41,42]. Our results showed that l-CAD plays a role in cancer cell motility and invasiveness, and consequently, metastasis, in breast cancer, particularly TNBC. The processes of cell adhesion and migration, which play crucial roles in both normal physiological functions and disease states such as cancer, are regulated through intricate rearrangements of the cytoskeletal network. In support of our findings, previous studies have shown that CAD acts as a regulatory component of the actomyosin contractile system, mediating Ca²⁺-dependent regulation in both smooth muscle cells and non-smooth muscle cells[7,11]. In non-smooth muscle cells, CAD participates in the formation and preservation of stress fibers - structures that facilitate the establishment of adhesion sites at the tumor cell periphery and mediate coordinated cell migration[11,12]. In 2005, Mirzapoiazova et al[40] reported that CAD is crucial for the reorganization of actin microfilaments and the motility of endothelial cells, an early event in angiogenesis, which in turn facilitates tumor invasion. Subsequently, in 2013, Chang et al[17] found increased CAD expression in metastatic oral squamous cell carcinoma, which was associated with poor prognosis. CAD expression was also detected in newly formed blood vessels and the surrounding tumor stroma, suggesting a role in the progression and metastasis of oral squamous cell carcinoma. In 2014, Zhang et al[21] demonstrated that the expression of inducible nitric oxide synthase, CAD, and osteopontin was closely correlated to the metastatic potential of lung cancer.

Additional evidence supporting our data came from a study by Kim et al[18], which demonstrated a strong positive association between upregulated expression of l-CAD and colon cancer malignancies. Aberrant expression of l-CAD can promote metastatic traits and modify the responsiveness of CRC cells to combined chemoradiotherapy. Additionally, the level of l-CAD expression may serve as a predictive marker for the effectiveness of neoadjuvant chemotherapy in upper gastrointestinal cancers[18].

CALD1 expression is likely associated with a decrease in ER signaling and the development of a more aggressive tumor phenotype characterized by EMT-like traits. Our RT-PCR results and sequence analyses showed that breast cancer cells express l-CAD encoded by WI-38 L-CAD II (transcript variant 2), suggesting that this isoform is responsible for tumorigenic activity. Consistent with our findings, Thorsen et al[23] reported that extended CALD1 isoforms incorporating elongated exons 5 and 6 were absent or markedly reduced in bladder, colon, and metastatic prostate cancers. The most probable splice variant present in these tumors was transcript variant 2 from the WI-38 L-CADII gene. Using exon array and RT-PCR analyses[23], the study further demonstrated the presence of a bladder cancer-specific splice variant of the CALD1 gene that was significantly differentially expressed between bladder cancer and normal bladder mucosa. In contrast, several studies on bladder tumors of smooth muscle origin either did not specify the CALD1 isoform examined or focused exclusively on the h-CAD isoform in stromal smooth muscle-derived tumors[15,16,43]. It is well known that alternative splicing can generate gene isoforms with opposing effects in cancer. A classic example is the situation of the B-cell lymphoma-extra (BCL-X) gene, in which the long isoform (BCL-XL) exerts antiapoptotic effects, whereas the short isoform, BCL-XS, promotes apoptosis[44]. Overall, the existence of tumor-specific CALD1 splice variants may account for the contrasting functions of the h-CAD and l-CAD isoforms and could provide a rationale for the oncogenic activity of l-CAD in various cancers.

Several published studies have reported conflicting findings regarding the role of CAD in cancer progression. For example, studies have indicated that CAD can inhibit the invasive behavior of cancer cells by modulating invadopodium assembly in colon and breast cancer models[26], or by affecting phosphorylation mediated by type I cGMP-dependent protein kinase in breast cancer cells[27]. In the case of gastric cancer, CAD expression was found to be reduced in lymph node metastases relative to the primary tumor. On the other hand, overexpressing CAD in a gastric cancer cell line derived from metastatic lymph nodes led to a decrease in both migration and invasion[30]. These divergent observations likely reflect the variable functions of CALD1 isoforms, which depend on the specific splice variants present in different tumor settings. However, more research is needed to comprehensively define how CAD isoforms influence cancer progression.

Immunohistochemical analysis revealed cytoplasmic expression of l-CAD in 55% of tumor cells and in the majority of stromal cells. Correlation analysis demonstrated a significant positive correlation between l-CAD expression and lymphatic invasion and HER2 status, as well as a trend toward association with the presence of an in situ component. In contrast, l-CAD expression showed significant negative correlations with ER and PR status. This finding is consistent with our western blot analyses of TNBC cell lines (ER^-/PR^-/HER2^-), which showed high l-CAD expression.

In parallel, in silico analysis of CALD1 gene expression across all breast cancer molecular subtypes showed high CALD1 expression and an association with irregular disease-free survival. However, a key limitation of the in silico analyses is the inability to distinguish between specific CALD1 transcript variants or protein isoforms. Hence, these datasets provide only general guidelines and lack the specificity of our results generated specifically for l-CAD.

To thoroughly explore the relationship between CALD1 expression and its prognostic significance in TNBC, we assessed OS using Kaplan-Meier plotter with specific TNBC criteria (ER^-, PR^-, HER2^-, and lymph node+). This analysis revealed a significant association between CALD1 expression and poor OS, indicating that CALD1 may serve as a negative prognostic marker in TNBC. Notably, this adverse prognostic effect persisted despite potential survival gains from advanced therapeutic interventions, suggesting that CALD1-driven biology may not be adequately targeted by current treatment strategies.

Stage-specific analysis of CALD1 expression demonstrated a trend toward higher expression in primary tumor stages before lymph node metastasis, although these differences were not statistically significant. This finding is consistent with a previous report based on stage-specific expression of l-CAD in CRC[33], where l-CAD expression was linked to the pre-metastatic stage and was still present to a lesser degree during lymph node metastasis. These findings suggest that l-CAD may play a particularly important role in early local invasion rather than exclusively in late metastatic dissemination.

In normal colonic epithelial cells, l-CAD showed only low-level, diffuse cytoplasmic staining, with no specific accumulation at either the luminal surface or the crypt bases. By contrast, malignant cells from CRC samples exhibited l-CAD expression in 71.4% of cases. h-CAD, however, was not detected in either normal or malignant epithelial cells, and its expression was confined to the smooth muscle of the colon, blood vessels, and, to a lesser extent, tumor-associated stromal tissue. Supporting the current findings, l-CAD was consistently observed in stromal cells associated with tumors in all CRC cases, pointing to a potentially broader role for l-CAD in shaping the tumor microenvironment[33].

CONCLUSION

Our findings indicate that l-CAD plays a potential oncogenic role in breast cancer initiation and progression, contributing to poorer patient outcomes. l-CAD is highly expressed in TNBC, the most aggressive breast cancer subtype, highlighting its relevance as a promising candidate for further investigation in larger patient cohorts. Moreover, l-CAD appears to promote the activation of EMT-related transcription factors, enhancing cell migration and metastatic potential and correlating with reduced OS. Finally, the strong cytoplasmic localization of l-CAD in tumor cells supports its potential utility as a diagnostic and prognostic biomarker in breast cancer.