Connexin 43-modified bone marrow stromal cells reverse the imatinib resistance of K562 cells via Ca²⁺-dependent gap junction intercellular communication

Abstract

Background: 

Imatinib mesylate (IM) resistance is an emerging problem for chronic myeloid leukemia (CML). Previous studies found that connexin 43 (Cx43) deficiency in the hematopoietic microenvironment (HM) protects minimal residual disease (MRD), but the mechanism remains unknown.

Methods: 

Immunohistochemistry assays were employed to compare the expression of Cx43 and hypoxia-inducible factor 1α (HIF-1α) in bone marrow (BM) biopsies of CML patients and healthy donors. A coculture system of K562 cells and several Cx43-modified bone marrow stromal cells (BMSCs) was established under IM treatment. Proliferation, cell cycle, apoptosis, and other indicators of K562 cells in different groups were detected to investigate the function and possible mechanism of Cx43. We assessed the Ca²⁺-related pathway by Western blotting. Tumor-bearing models were also established to validate the causal role of Cx43 in reversing IM resistance.

Results: 

Low levels of Cx43 in BMs were observed in CML patients, and Cx43 expression was negatively correlated with HIF-1α. We also observed that K562 cells cocultured with BMSCs transfected with adenovirus-short hairpin RNA of Cx43 (BMSCs-shCx43) had a lower apoptosis rate and that their cell cycle was blocked in G0/G1 phase, while the result was the opposite in the Cx43-overexpression setting. Cx43 mediates gap junction intercellular communication (GJIC) through direct contact, and Ca²⁺ is the key factor mediating the downstream apoptotic pathway. In animal experiments, mice bearing K562, and BMSCs-Cx43 had the smallest tumor volume and spleen, which was consistent with the in vitro experiments.

Conclusions: 

Cx43 deficiency exists in CML patients, promoting the generation of MRD and inducing drug resistance. Enhancing Cx43 expression and GJIC function in the HM may be a novel strategy to reverse drug resistance and promote IM efficacy.

Introduction

Chronic myeloid leukemia (CML) is a myeloproliferative hematological malignancy with a characteristic translocation of the Philadelphia chromosome.^[1,2] Thanks to the discovery of the single oncogene BCR-ABL and the synthesis of the tyrosine protein kinase inhibitor, imatinib mesylate (IM), CML has become much easier to treat.^[2] Drug resistance and drug withdrawal are two emerging unsolved problems. Imatinib resistance can be divided into BCR-ABL-dependent and BCR-ABL-independent forms.^[3] The former is caused by the BCR-ABL mutation, and most patients choose next-generation targeted drugs for therapy.^[4] The latter is usually caused by abnormally activated minimal residual disease (MRD), which becomes difficult to treat.^[5]

The hematopoietic microenvironment (HM) appears to sustain and regulate the proliferation and differentiation of both hematopoietic stem/progenitor cells and leukemia cells.^[6] Bone marrow stromal cells (BMSCs) are the main functional components of the HM. Gap junction intercellular communication (GJIC) of BMSCs participates in the regulation and synchronization of hematopoietic cells.^[7] Previous studies on drug resistance and the generation of MRD mainly emphasized changes in the adhesion and secretion functions of BMSCs.^[7,8] Currently, much evidence suggests that GJIC in HM is also an important factor that affects the proliferation, apoptosis, and drug sensitivity of leukemic cells.^[9,10] Connexin 43 (Cx43), the most common and widespread connexin in humans, is a major structural protein expressed on the cell surface and in the nuclear membrane. It docks with similar molecules and forms gap junctions. Ions, small metabolites, second messengers, and micro ribonucleic acid (RNAs) can be directly delivered between closely adjacent cells.^[11] Our previous studies reported that Cx43-overexpressing human umbilical cord stem cells enhanced the effect of chemotherapeutics and eliminated minimal residual L615 cells in an animal model.^[12] However, the functions and detailed mechanisms of Cx43 in imatinib resistance in CML are poorly understood.

In this study, we investigated the effect of Cx43 in BMSCs on apoptosis, proliferation, and the cell cycle of K562 cells under IM treatment.

Methods

Isolation of BMSCs and treatment

Human BMSCs were isolated from bone marrow (BM) samples separated from healthy donors from the Medical Centre of Hematology, Xinqiao Hospital, as previously described.^[9] All tissue samples used in this study were collected after obtaining written informed consent from patients, and all procedures were conducted in accordance with the study protocol approved by Ethics Committee of Xinqiao Hospital (No. AMUWEC20211606). The study was performed in accordance with the Declaration of Helsinki and its later amendments. The cells were authenticated with a cytometry assay (FCASVerse, BD Biosciences, Franklin Lakes, NJ, USA). BMSCs were cultured in minimum essential medium alpha (MEM-α; Hyclone, Logan, UT, USA) with 10% fetal bovine serum (FBS; Gibco, Waltham, Massachusetts, USA) and 1 ng/mL basic fibriblast growth factor (Sigma−Aldrich, St Louis, MO, USA) at 37°C in a humidified atmosphere with 5% CO₂. BMSCs were treated with 100 μmol/L cobalt dichloride (CoCl₂) (Sigma−Aldrich) for 72 h to prevent the degradation of hypoxia-inducible factor 1α (HIF-1α). The human leukemic cell line K562 (The Cell Bank of Chinese Academy of Science, Beijing, China) was cultured in Roswell Park Memorial Institute 1640 medium (RPMI 1640; HyClone) with 10% FBS.

Immunophenotypic identification and transfection of BMSCs

To identify BMSCs, fluorescently labeled anti-human and anti-mouse cluster of differentiation 73 (CD73), CD90, CD105, CD14, CD20, CD34, and CD45 monoclonal antibodies (BD Biosciences, Franklin Lakes, NJ, USA) were added to each tube, and the samples were mixed and incubated for 15 min in the dark. Next, 1 mL phosphate buffer saline (PBS) was added to the tubes, which were centrifuged at 1000 round per minute for 5 min (r=172 mm). Finally, the BMSCs were resuspended in PBS, detected by a BDVerse flow cytometer (BD Biosciences) and analyzed by FlowJo software (FlowJo, Ashland, OR, USA).

Adenovirus vectors containing different gene sequences were transfected into BMSCs as previously described.^[12] Adenovirus-short hairpin RNA of HIF-1α-green fluorescent protein (Ad-shHIF-1α-GFP), Ad-negative control-GFP (Ad-NC-GFP), Ad-Cx43 overexpression-GFP (Ad-overCx43-GFP), and Ad-shCx43-GFP were purchased from Heyuan Biotechnology (Shanghai, China). Expression of the target genes was measured 48 h after transfection by Western blotting.

Fluorescence recovery after photobleaching (FRAP) assay

Cx43-modified BMSCs were cultured on specialized glass and loaded with 1 μmol/L fluorescein diacetate (Sigma–Aldrich) for 5 to 10 min at 37°C. Laser confocal scanning microscopy (FluoView FV3000, Olympus, Tokyo, Japan) was used to photograph and bleach the target cells with 3−5 neighboring cells. The image was taken using a low laser intensity of acousto-optic tunable filter (AOTF; 0.1%). The laser intensity for photobleaching was increased 400 × (AOTF = 40%, 40.0 μs/pixel) for a total time of 1.126 s. Every image was scanned at intervals of 3.22 s for a total period of 1012 s.

Cell coculturing and transwell assay

BMSCs or Cx43-modified BMSCs were seeded at 1 × 10⁵/well in a 6-well plate. Then, K562 cells were cocultured on top of the BMSC layer for 72 h with or without 1 μmol/L IM (Selleck, Houston, Texas, USA) in RPMI 1640 medium [Table 1].

Table 1 - Five groups of cell co-culturing systems to investigate the effect of Cx43-modified BMSCs in K562 cells under IM treatment.

Groups	Number of K562 cells	BMSCs	IM
K562 group	7.5 × 105	–	–
IM group	7.5 × 105	–	1 μmol/L
overCx43 group	7.5 × 105	BMSCs-overCx43	1 μmol/L
NC group	7.5 × 105	BMSCs-NC	1 μmol/L
shCx43 group	7.5 × 105	BMSCs-shCx43	1 μmol/L

BMSCs: Bone marrow stromal cells; Cx43: Connexin 43; IM: Imatinib mesylate; NC group: K562 cells cocultured with BMSCs-NC; overCx43 group: K562 cells cocultured with BMSCs-overCx43; shCx43 group: K562 cells cocultured with BMSCs-shCx43; NC: Negative control; overCx43: Cx43 overexpression; shCx43: Short hairpin RNA of Cx43; –: Not available.

To validate whether Cx43 mediates cell communication through cell–cell contact or other non-adjacent pathways, a transwell assay was performed to separate K562 cells and Cx43-modified BMSCs. K562 cells were placed in the lower chamber of the Transwell system (0.04 μm, pore filter, Corning, NY, USA), while BMSCs were placed in the upper chamber.

Cell proliferation, cell cycle, and apoptosis

Cell proliferation was assessed with a Cell Counting Kit-8 (CCK8; Dojindo Laboratories, Kumanoto, Japan) assay with at least three duplicate wells. K562 cells from each group were resuspended and incubated in 96-well plates at 37°C for 4 h before measuring the absorbance at 450 nm. For cell cycle analysis, K562 cells were fixed with 70% ethanol overnight at 4°C. The fixed cells were stained with propidium iodide (PI; Beyotime) at 37°C according to the manufacturer's protocol. For apoptosis analysis, K562 cells from each group were stained with annexin V and PI (BD Pharmingen) at room temperature for 15 min followed by flow cytometric testing.

Intracellular Ca²⁺, mitochondrial membrane potential (MMP), and reactive oxygen species (ROS) analysis

Flou-3-pentaacetoxymethyl (Fluo-3/AM; Beyotime) was used for intracellular Ca²⁺ testing. K562 cells from each group were collected and resuspended with 1 μmol/L Fluo-3/AM diluted in 1 mL D-Hank's balanced salt solution (Solarbio, Beijing, China) for 30 min. Then, the cells were incubated for another 30 min and analyzed by a C6 flow cytometer (BD Biosciences).

A 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethyl-imidacarbocyanine (JC-1) kit (Beyotime) was used for MMP measurement. K562 cells at 24 h after coculturing were stained with 0.5 mL JC-1 solution. All samples were incubated for 20 min and detected by a C6 flow cytometer.

2′,7′-Dichlorofluorescein diacetate (DCFH-DA, Beyotime) was used for ROS determination. Twenty-four hours after coculturing, K562 cells were stained with 10 μL of DCFH-DA for 20 min and resuspended in D-Hank's solution, and 2′,7′-Dichlorofluorescein (DCF) fluorescence was also detected by a C6 flow cytometer.

Luminescence assay for adenosine triphosphate (ATP) release

K562 cells (2 × 10⁴) from each group were lysed with lysis buffer (Beyotime) on ice. Then, 100 μL ATP testing diluent (Beyotime) and 10 μL samples or reference were mixed and tested by a GloMax 20/20 Luminometer (Promega−GloMax, Madison, Wisconsin, USA).

Western blotting

The total cellular proteins of K562 cells and BMSCs were extracted by radio-immunoprecipitation assay. Forty micrograms of protein from each cell line was added to each lane, resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and then transferred to polyvinylidene difluoride (PVDF) membranes (Bio-Rad, Hercules, CA, USA). The membranes were blocked in tris-buffer solution and tween (TBST) with 5% non-fat milk for 2 h at room temperature and washed for four times. Then, the membranes were incubated overnight at 4°C using the following primary antibodies: anti-Cx43 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-HIF-1α (CST, Danvers, Massachusetts, USA), anti-Cytochrome C (CST), anti-B cell lymphoma 2 (anti-BCL-2; CST), anti-Cleaved Caspase-9 (CST), anti-Cleaved Caspase-3 (CST), anti-poly (ADP-ribose) polymerase-1 (anti-Cleaved PARP; CST), anti-Ca²⁺/calmodulin-dependent protein kenases II (anti-CaMKII; CST), anti-protein kinase B (anti-Akt; CST), anti-mammalian target of rapamycin (anti-mTOR; CST), anti-BCL2-associated X (anti-BAX; Boster, Wuhan, China), anti-BCR-ABL (Abcam, Cambridge, UK), anti-phosphorylated-CRK like protein (p-CrkL; CST), anti-p-CaMKII (CST), anti-p-Akt (CST), and anti-p-mTOR (Boster). The membranes were washed with TBST and incubated with anti-β-actin (Boster) for 1 h. Finally, Immobilon Western Chemiluminescent HRP Substrate (Millipore, Mass, Massachusetts, USA) was used to determine the target proteins.

Real-time quantitative polymerase chain reaction (RT-qPCR)

RNA was extracted from BMSCs and reversely transcribed into complementary DNA (cDNA) using the Transcriptor First Strand cDNA Synthesis Kit (Roche Diagnostics, Basel, Switzerland) according to the manufacturer's instructions. Quantification of Cx43 and HIF-1α relative to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was performed on the CFX ConnectTM Real Time PCR Detection System using Synergy Brands (SYBR) Green chemistry (Bio-Rad). The forward (F) and reversed (R) primer sequences were as follows:

F-Cx43: 5′-GCCTACTCAACTGCTGGAGG-3′

R-Cx43: 5′-CCAGAAGCGCACATGAGAGA-3′

F-HIF-1α: 5′-GAACGTCGAAAAGAAAAGTCTCG-3′

R-HIF-1α: 5′-CCTTATCAAGATGCGAACTCACA-3′

F-GAPDH: 5′-GTCAAGGCTGAGAACGGGAA-3′

R-GAPDH: 5′-AAATGAGCCCCAGCCTTCTC-3′

In vivo experiments

Nude mice were purchased from the Laboratory Animal Center of Army Medical University (Chongqing, China) and fed in aseptic cages. All murine experiments were performed in accordance with protocols that were approved by the Animal Care and Use Committee of the Centre of Army Medical University (No. AMUWEC20211606). To establish the xenograft tumor model, nude mice aged 4−6 weeks were randomly divided into the K562 group, Cx43 group, and negative control (NC) group. The mice in the three groups received subcutaneous injections of K562 cells + Cx43-modified BMSCs or K562 cells only [Table 2]. Besides, they were intraperitoneally given with 200 μL PBS, 200 μL IM (80 mg/kg, diluted in PBS, Selleck), and 200 μL IM, respectively, every two days from the sixth day after bearing. The volume of the tumor was measured in two directions and calculated as follows: V = 1/2 × length × (width)². All mice were sacrificed by cervical dislocation on the 24th day after cell injection.

Table 2 - Three animal groups to in vivo investigate the effect of Cx43-modified BMSCs in K562 cells under IM treatment.

Groups	K562 cells	BMSCs	IM
K562 group	1 × 107	–	–
Cx43 group	8 × 106	2 × 106 BMSCs-overCx43	200 μL
NC group	8 × 106	2 × 106 BMSCs-NC	200 μL

BMSCs: Bone marrow stromal cells; Cx43: Connexin 43; IM: Imatinib mesylate; NC: Negative control; overCx43: Cx43 overexpression; –: Not available.

Immunohistochemistry and immunofluorescence

Tissues were fixed in 10% neutral buffered formalin overnight at room temperature. The fixed tissues were embedded in paraffin and cut into 4 to 6 μm tissue sections. Then, the slides were stained with hematoxylin–eosin (HE), anti-HIF-1α (Abcam), anti-Cx43 (Santa Cruz), anti-BCR-ABL (Abcam), anti-BAX (Boster), and anti-Ki67 (Bioss, Beijing, China) and terminal dexoynucleotidyl transferase-mediated deoxyurinetriphate (dUTP)-digoxigenin nick end labeling (TUNEL; Beyotime). The slides were imaged under a light microscope, and Cx43 levels in human biopsy specimens were analyzed using ImageProPlus software (Media Cybernetics, Silver Springs, MD, USA).

Cx43-modified BMSCs were cultured in a 6-well plate at 1 × 10⁵/well and fixed with 40 mg/L paraformaldehyde (Sigma) after cell adherence. Then, the cells were stained with anti-Cx43 (Santa Cruz) overnight at 4°C. After washing for three times, the cells were incubated with Anti-mouse Immunoglobin G (IgG; CST) for 30 min. Finally, the cells were stained with 4′,6-diamidine-2′-phenylindole dihydrochloride (DAPI; Boster) for 15 min, and the slides were sealed using 60% buffered glycerol (Sigma−Aldrich) and imaged under an immunofluorescence microscope (Olympus).

Statistical analysis

Statistical analyses were performed using SPSS software version 26.0 (IBM, Armonk, NY, USA). Statistical figures were drawn using GraphPad Prim 5.0 (GraphPad Software Inc., San Diego, CA, USA). Quantitative variables are expressed as the mean ± standard deviation. The unpaired two-tailed Student's t test was applied for comparing any two groups, and the one-way analysis of variance (ANOVA) test, and multiple comparisons test were used for comparison among multiple groups. P < 0.05 was considered statistically significant.

Results

BM of CML patients shows relatively low levels of Cx43 and high levels of HIF-1α

As Cx43 is important for GJIC, we measured Cx43 expression in BM biopsy specimens of ten CML patients at first diagnosis and ten healthy donors by immunohistochemical assay. Patient information is listed in Table 3. The results showed that the mean integral optical intensity (IOD) of Cx43 in CML patients was statistically significantly higher than that in healthy people [Figure 1A]. Among the ten CML patients, four showed negative expression (IOD < 2000 was considered negative), while all BM specimens from healthy donors were positive [Figure 1B]. Cx43 expression was heterogeneous in BM specimens. However, it was decreased in patients who were diagnosed with CML.

Table 3 - IOD (mean) of Cx43 in the biopsy of CML patients and healthy donors.

Patients	Gender	Age (years)	Diagnosis	IOD (mean)	Donor	Gender	Age (years)	IOD (mean)
Patient 1	M	20	CML	623.91	Donor 1	F	53	2185.22
Patient 2	M	22	CML	1710.89	Donor 2	F	44	2166.52
Patient 3	F	28	CML	561.67	Donor 3	F	61	5368.76
Patient 4	F	54	CML	2002.27	Donor 4	F	20	7650.07
Patient 5	F	58	CML	2935.38	Donor 5	M	37	5184.72
Patient 6	M	75	CML	4271.85	Donor 6	M	52	7020.30
Patient 7	F	26	CML	1975.45	Donor 7	M	32	8386.25
Patient 8	M	64	CML	2239.86	Donor 8	M	48	3190.83
Patient 9	F	35	CML	6473.38	Donor 9	F	55	2056.72
Patient 10	M	57	CML	631.65	Donor 10	F	25	2532.66

CML: Chronic myeloid leukemia; Cx43: Connexin 43; F: Female; IOD: Integral optical intensity; M: Male.

Figure 1: Decreased Cx43 and increased HIF-1α were found in biopsies of CML patients. (A) IOD (mean) of Cx43 in biopsies from normal patients (n = 10) and CML patients (n = 10) as analyzed by ImagePro software (^∗ P < 0.05, unpaired two-tailed t test). (B) Number of Cx43-positive and Cx43-negative patients in normal and CML groups. (C) Immunohistochemical assay of bone marrow biopsy (×200). 1,3: healthy donors; 2,4: CML patients. (D) Identification of BMSCs extracted from healthy donors by flow cytometry. (E) Western blotting of Cx43 and HIF-1α levels in BMSCs and BMSCs-CoCl₂. (F, G) HIF-1α and Cx43 mRNA expression in BMSCs and BMSCs-CoCl₂ (^† P < 0.01, unpaired two-tailed t test). (H) Western blotting of Cx43 and HIF-1α in BMSCs, BMSCs-CoCl₂-NC, and BMSCs-CoCl₂-shHIF-1α. APC: Allophycocyanin; BMSCs: Bone marrow stromal cells; BMSCs-CoCl₂: CoCl₂-treated BMSCs; BMSCs-CoCl₂-NC: CoCl₂-treated BMSC-NC; BMSCs-CoCl₂-shHIF-1α: CoCl₂-treated BMSC-shHIF-1α; CML: Chronic myeloid leukemia; Cx43: Connexin 43; FITC: Fluorescein isothiocyanate; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; HIF-1α: Hypoxia-inducible factor 1α; IOD: Integral optical density; ISO: International Organizetion for Standardization; mRNA: microRNA; NC: Negative control; PE: Pgycoerthrin; PerCP-Cy5.5: Peridin-Chlorophyll-Protein Cyanine 5.5; shHIF-1α: Short hairpin RNA of HIF-1α.

Hypoxia is common in many chronic tumor diseases.^[13] Recent studies revealed that physiological hypoxia in BM promotes communication deficiency between hematopoietic stem cells (HSCs) and BMSCs in some hematological malignancies.^[14] HIF-1α is a key transcription factor associated with hypoxia. HIF-1α and Cx43 were measured from biopsy specimens of healthy donors and CML patients [Figure 1C], and we found a negative correlation between HIF-1α level and Cx43 levels in both healthy donors and CML patients.

Hypoxia is negatively correlated with Cx43 expression in BMSCs

As BMSCs are the major component of BM, the physiological hypoxic state of BMSCs was induced with the hypoxia-inducing reagent CoCl₂ (100 μmol/L) to prevent the degradation of HIF-1α. BMSCs were isolated from healthy donors and authenticated by flow cytometry [Figure 1D]. We found that HIF-1α expression was increased in CoCl₂-treated BMSCs (BMSCs-CoCl₂) and that Cx43 expression was decreased at the transcriptional and translational levels, whereas the results in untreated BMSCs were reversed [Figures 1E–1G].

To further validate that HIF-1α is a regulator of Cx43, HIF-1α was knocked down with Ad-shHIF-1α-GFP in BMSCs. As shown in Figure 1H, we found that CoCl₂ treatment resulted in increased HIF-1α expression and decreased Cx43 expression in BMSCs transfected with Ad-NC-GFP (BMSCs-NC), but when HIF-1α was knocked down, the expression of Cx43 was recovered [Figure 1H]. This result indicated that HIF-1α is upstream of Cx43, and reversion of HIF-1α expression and hypoxia led to the recovery of Cx43 expression levels in BMSCs.

BMSCs-overCx43 had the strongest GJIC function

To observe the role of Cx43 in HM in vitro, P₃–P₅ BMSCs were transduced with Ad-overCx43-GFP, Ad-NC-GFP, and Ad-shCx43-GFP, which upregulate, do not change, and downregulate Cx43 expression in BMSCs, respectively. The three BMSCs were defined as BMSCs-overCx43, BMSCs-NC, and BMSCs-shCx43. The expression of Cx43 in the three BMSCs was measured by Western blotting and RT-qPCR [Figures 2A–2C].

Figure 2: Cx43-overexpressing BMSCs showed enhanced GJIC functions. (A) Western blotting of Cx43 in BMSCs-NC, BMSCs-overCx43, and BMSCs-shCx43. (B, C) The mRNA levels of Cx43 in BMSCs-NC, BMSCs-overCx43, and BMSCs-shCx43. (^∗ P < 0.01, ^† P < 0.001, unpaired two-tailed t test). (D) FRAP assay in BMSCs-overCx43, BMSCs-NC, and BMSCs-shCx43 at the time points of pre-photobleaching and 0 s and 1000 s post-photobleaching (×400). (E) Recovery rate-time curve of BMSCs-NC, BMSCs-overCx43, and BMSCs-shCx43. (F) Fitted curve of recovery rates (%) of three Cx43-modified BMSCs and the maximum recovery ratio (%) and the time needed to recover to 20%. BMSCs: Bone marrow stromal cells; BMSCs-NC: BMSCs transfected with adenovirus-negative control-green fluorescence protein; BMSCs-overCx43: BMSCs transfected with adenovirus-Cx43 overexpression-green fluorescence protein; BMSCs-shCx43: BMSCs transfected with adenovirus-short hairpin RNA of Cx43-green fluorescence protein; Cx43: Connexin 43; FRAP: Fluorescence recovery after photobleaching; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; GJIC: Gap junction intercellular communication; mRNA: microRNA; NC: Negative control; overCx43: Cx43 overexpression; shCx43: Short hairpin RNA of Cx43.

GJIC function of the three Cx43-modified BMSCs was tested with the gap-FRAP assay [Figure 2D]. From the data of Figures 2E and 2F, the fluorescence intensity of normal BMSCs recovered to 22% in 300 s and 30% within 500 s, whereas that of BMSCs-over Cx43 recovered to 37% and 48%, and that of BMSCs-shCx43 recovered to 13% and 17% at 300 s and 500 s, respectively. After observation for 1000 s, BMSCs-overCx43 showed the highest recovered rate of fluorescence, with an average recovery percentage of 59%, compared to BMSCs-NC (43%), and BMSC-shCx43 (23%). Moreover, BMSCs-overCx43, BMSCs-NC, and BMSC-shCx43 took approximately 105 s, 140 s, and 759 s to reach 20% recovery, respectively. These results implied that BMSCs overexpressing Cx43 have the strongest GJIC function.

Cx43-modified BMSCs affect K562 apoptosis and the cell cycle but not proliferation under IM treatment

K562 is a hematopoietic progenitor cell line established from a human CML patient in the blast phase, which is classical and is recognized as representative when studying the disease. K562 cells can be killed by IM in a time- and concentration-dependent manner.^[15] Direct contact coculturing systems of Cx43-modified BMSCs and K562 cells were established to explore the effect of Cx43 on the survival state of K562 cells. Different treatments were administered to the five groups: the K562 group (only K562 cells), IM group (K562 cells treated with IM), overCx43 group (k562 cells cocultured with BMSCs-overCx43 and treated with IM), NC group (k562 cells cocultured with BMSCs-NC and treated with IM), and shCx43 group (k562 cells cocultured with BMSCs-shCx43 and treated with IM) [Table 1].

A CCK8 assay was used to detect K562 cell proliferation [Figure 3A]. We found no statistically significant difference in cell proliferation in any of the IM-treated groups at 72 h of coculturing. The cell cycle of K562 cells was detected [Figures 3B–3E], and results showed that the proportion in the G0/G1 phase was statistically significantly increased in the shCx43 group compared with the IM group [Figure 3B], while the opposite result was observed in the overCx43 group [Figures 3C]. Regarding apoptosis [Figures 3F–3I], we found that compared with the IM group, the NC group had a lower apoptosis rate, which confirmed the protective role of BM in leukemic cell viability [Figure 3F]. The apoptosis rate of K562 cells in the shCx43 group was significantly decreased compared with that in the IM group and even the NC group [Figures 3F and 3H]. The overCx43 group showed an apoptosis rate relatively equal to that of the IM group [Figures 3G and 3I]. Taken together, the results indicated that overexpression of Cx43 in BM plays an important role in promoting killing efficacy of IM.

Figure 3: K562 cells in the Cx43-overexpressing HM showed a higher apoptotic rate after IM treatment. (A) K562 cell proliferation in the K562, NC, overCx43, shCx43, and IM groups. (B) Comparison of cell cycle of K562 cells among K562 group, IM group, NC group and shCx43 group (LSD multiple comparisons test). (C) Comparison of cell cycle of K562 cells among K562 group, IM group, NC group and overCx43 group (LSD multiple comparisons test). (D–E) Representative figures of 3B and 3C. (F) Comparison of apoptosis of K562 cells among K562 group, IM group, NC group and shCx43 group (one-way ANOVA). (G) Comparison of apoptosis of K562 cells among K562 group, IM group, NC group and shCx43 group (one-way ANOVA). (H–I) Representative figures of 3F and 3G. ^∗ P < 0.001, ^† P < 0.01, ^‡ P < 0.05. ANOVA: Analysis of variance; BMSCs: Bone marrow stromal cells; Cx43: Connexin 43; HM: Hematopoietic microenvironment; IM: Imatinib mesylate; LSD: Least significant difference; NC: Negative control; n.s.: No significance; OD: Optical density; overCx43: Cx43 overexpression; PI: Propidine iodide; shCx43: Short hairpin RNA of Cx43.

Cx43 influences the efficacy of IM through cell–cell contact

Cx43 not only participates in GJIC, but also mediates communication between nonadjacent cells through tunneling nanotubes, extracellular vesicles, and even external secretion.^[11] To explore whether Cx43 plays its role through GJIC or other non-contact mechanisms, we employed a transwell plate (0.4 μm pores, 6-well) to separate BMSCs and K562 cells. The results showed no statistically significant difference in the apoptosis rate among the overCx43 group, shCx43 group, NC group, and even the IM group [Figure 4A], which suggested that cell–cell contact plays a dominant role in influencing the killing efficacy of IM.

Figure 4: Cell–cell contact played the dominant role in GJIC. (A) Apoptosis of cocultured K562 cells in transwell systems (n.s.: P > 0.05, LSD multiple comparisons test). (B) Cx43 mRNA expression of K562 cells in IM group, NC group, overCx43 group and K562 group (LSD multiple comparisons test). (C) Western blotting of K562 cells in IM group, NC group, overCx43 group and K562 group. (D) Cx43 mRNA expression in BMSCs-NC and BMSCs-overCx43 before and after coculturing with K562 cells (LSD multiple comparisons test). (E) Western blotting of BMSCs-NC before and after coculturing with K562 cells. (F) Immunofluorescence of BMSCs-NC and BMSCs-overCx43 before and after coculturing with K562 cells (200×). ^∗ P < 0.001, ^† P < 0.01. BMSCs: Bone marrow stromal cells; BMSCs-NC: BMSCs transfected with adenovirus-negative control-green fluorescence protein; BMSCs-overCx43: BMSCs transfected with adenovirus-Cx43 overexpression-green fluorescence protein; Cx43: Connexin 43; DAPI: 4′,6-diamidine-2′-phenylindole dihydrochloride; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; GJIC: Gap junction intercellular communication; IM: Imatinib mesylate; LSD: Least significant difference; mRNA: microRNA; NC: Negative control; n.s.: No significance; overCx43: Cx43 overexpression.

After coculturing, Cx43 levels in K562 cells and BMSCs were measured. We found that K562 cells cocultured with BMSCs-overCx43 had enhanced Cx43 expression at both the transcriptional and translational levels [Figures 4B and 4C], while Cx43 was hardly expressed on the surface of untreated K562 cells. Interestingly, we observed that after coculturing, Cx43 expression in BMSCs-overCx43 was higher than that before coculturing in both the PCR and Western blotting experiments [Figures 4D and 4E]. Immunofluorescence assays of BMSCs before and after coculturing were performed and we found the stronger signal in BMSCs after coculturing [Figure 4F]. Moreover, we observed that the increased Cx43 was mainly expressed in BMSC cell and nuclear membranes. The results demonstrated that while HM affected the survival state of leukemic cells, the latter also remodeled the HM at the same time.

Ca²⁺ transfer via the Cx43 channel mediates the synergistic efficacy of IM in the Cx43-overexpressing HM

GJIC permits the rapid exchange of small molecules, including Ca²⁺ and ATP, into the cytoplasm. To explore whether ATP or Ca²⁺ plays a causal role in the drug resistance of CML cells, we detected the concentrations of intracellular ATP and Ca²⁺. No significant difference in ATP concentration was found among any IM-treated groups [Figure 5A], whereas the calcium fluorescence intensity in the overCx43 group was much higher than that in any IM-treated groups [Figure 5B].

Figure 5: Cx43-modified BMSCs influence the efficacy of IM in a Ca²⁺-dependent pathway. (A) Comparison of ATP concentration of K562 cells in the K562 group, NC group, overCx43 group, and IM group (n.s.: P > 0.05, one-way ANOVA). (B) Comparison of Ca²⁺ fluorescence intensity of K562 cells among the four groups (one-way ANOVA). (C) Comparison of MMP of K562 cells among the four groups (one-way ANOVA). (D) Representative figure of 4C. (E) Comparison of ROS fluorescence intensity of K562 cells among the four groups. (F) Western blotting of members of the mitochondrial apoptotic signaling pathway in K562 cells (one-way ANOVA). (G) Western blotting of members of the CaMKII–Akt–mTOR signaling pathway in K562 cells. (H) Western blotting assay of members of the BCR–ABL signaling pathway in K562 cells. ^∗ P < 0.01, ^† P < 0.05. Akt: Protein kinase B; ATP: Adenosine triphosphate; ANOVA: Analysis of variance; BAX: BCL2-associated X; BCL-2: B cell lymphoma 2; BMSCs: Bone marrow stromal cells; BMSCs-NC: BMSCs transfected with adenovirus-negative control-green fluorescence protein; BMSCs-overCx43: BMSCs transfected with adenovirus-Cx43 overexpression-green fluorescence protein; Cx43: Connexin 43; IM: Imatinib mesylate; JC-1: 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethyl-imidacarbocyanine; MMP: Mitochondrial membrane potential; NC: Negative control; n.s.: No significance; p-Akt: Phosphorylated-protein kinase B; PARP: Poly (adenosine diphosphate-ribose) polymerase-1; p-CaMKII: Phosphorylated-Ca²⁺/calmodulin-dependent protein kenases II; p-CrkL: Phosphorylated-CRK like protein; p-mTOR: Phosphorylated-mammalian target of rapamycin; ROS: Reactive oxygen species.

To validate the hypothesis that Ca²⁺ transfer from BMSCs to K562 cells augments the pro-apoptotic function in the Cx43-overexpressing HM, Ca²⁺-related apoptotic indicators were tested. Ca²⁺ is tightly linked with mitochondrial signaling, and mitochondrion-mediated apoptosis is initiated by the loss of MMP.^[16] As MMP depolarization was considered a Ca²⁺-triggered early event of apoptosis, we also compared the MMP depolarization of K562 cells among every group. Consistent with our hypothesis, the MMP of the overCx43 group was significantly increased, while no difference was found between the overCx43 group and the IM group [Figures 5C and 5D]. Mitochondria are a major target of ROS, and accumulated ROS can lead to damage to the mitochondrial membrane and even cell death. The intracellular ROS level of K562 cells was tested by flow cytometry, and we found a statistically significantly increased ROS level in the overCx43 group [Figure 5E].

Ca²⁺ augments the killing effect through both the caspase and CaMKII–Akt–mTOR signaling pathways

Although the observation of Cx43 and Ca²⁺ transfer inducing apoptosis in K562 cells has been validated, the molecular mechanism remains elusive. Therefore, we investigated the classical and novel apoptotic pathways related to Ca²⁺ by Western blotting. The mitochondrial apoptotic pathway and caspase pathway are classical and most common when referring to apoptosis. As shown in Figure 5F, we observed decreased BCL-2, increased BAX, and increased cytochrome C (CytoC) expression in K562 cells cocultured with BMSCs-overCx43 compared with those cocultured with BMSCs-NC. Moreover, the expression levels of cleaved caspase-9, cleaved caspase-3, and cleaved PARP were obviously higher than those in NC group and K562 group.

CaMKII is a general integrator of Ca²⁺ signaling that regulates the development and activity of many cell types. CaMKII is tightly linked with cell growth-related pathways.^[17,18] The CaMKII–Akt–mTOR pathway in K562 cells separated from the four groups (IM group, NC group, overCx43 group, and K562 group) was assessed by Western blotting, and the results showed increased p-CaMKII, decreased p-Akt, and decreased p-mTOR expression in the overCx43 group compared with the NC group [Figure 5G]. These bands indicated that the CaMKII–Akt–mTOR pathway was also involved in the Ca²⁺-mediated GJIC pathway.

We also measured the expression of BCR-ABL and its downstream molecule p-CrkL in K562 cells to test whether Cx43 could directly regulate the BCR–ABL pathway [Figure 5H]. However, no obvious difference was found between the Cx43 group and the NC group.

Cx43-modified BMSCs inhibit the tumorigenicity of K562 cells in vivo

A tumor-bearing model of nude mice was established to evaluate the synergistic effect of Cx43 on IM efficacy in K562 cells in vivo. Mice were randomly divided into three groups (n = 4). A total of 8 × 10⁶ K562 cells plus 2 × 10⁶ BMSCs-overCx43 (Cx43 group), 8 × 10⁶ K562 cells plus BMSCs-NC (control group), and 1 × 10⁷ K562 cells (K562 group) were subcutaneously injected into the right flank of mice in each group. From the sixth day, 200 μL diluent IM was intragastrically injected into the Cx43 group and control group every two days, while 200 μL water was given to the K562 group. The tumor volume was calculated every three days until the 24th day after inoculation. We observed that the tumor volumes of mice in the Cx43 group were consistently smaller with a slower growth rate compared with the control group [Figure 6A]. In addition, the latter showed relatively larger spleens than the former. Mice in the K562 group had the largest tumors and spleens [Figure 6B]. No significant difference was found in the average body weight of the three groups [Figure 6C].

Figure 6: Cx43 overexpression in BMSCs promoted K562 cell apoptosis under IM treatment in vivo. (A) Images of tumors and spleens excised from all mice in different groups on Day 24. (B) Average volumes of tumors in the Cx43 group, control group and K562 group (^∗ P < 0.05, one-way ANOVA). (C) Average body weight of mice in the Cx43 group, control group and K562 group. (D) Immunohistochemical staining of tumors (400×). ANOVA: Analysis of variance; BAX: BCL2-associated X; BCL-2: B cell lymphoma 2; BMSCs: Bone marrow stromal cells; Cx43: Connexin 43; HE: Hematoxylin–eosin; IM: Imatinib mesylate; TUNEL: Terminal dexoynucleotidyl transferase-mediated deoxyurinetriphate (dUTP)-digoxigenin nick end labeling.

Immunohistochemical staining was also performed. From HE staining results, we found a large number of tumor cells in Cx43 group under the microscope, which was much greater than that in the other two groups. In addition, inflammatory infiltration was observed in the Cx43 group [Figure 6D]. Cx43 was strongly positive in tumor tissues of the Cx43 group, while it was hardly stained in the K562 group. Other proliferative (Ki67, BCL-2, and BCR-ABL) and apoptotic (BAX and TUNEL) indicators were also stained, and the results displayed decreased levels of Ki67 and BCL-2 and increased levels of BAX and TUNEL in the Cx43 group compared with the control group, which suggested that the Cx43-overexpressing HM could inhibit tumorigenicity and synergize with the killing effect of IM in vivo. We found that Cx43 influences the killing effect of IM on K562 cells through a Ca²⁺-mediated pathway and further explored the possible mechanism. Our results might provide new ideas for reversing IM resistance.

Discussion

Imatinib resistance is the major obstacle in the treatment of CML. The protective effect of BM on leukemia cells is a vital factor for the occurrence and relapse of leukemia and promotes the development of MRD and drug resistance. BM is a natural shelter in which leukemic cells can avoid chemotherapy. It has been reported that BMSCs protect CML cells from imatinib-induced apoptosis via the chemokine (C-X-C motif) receptor 4/C-X-C motif ligand 12 (CXCL12) axis.^[19] In this study, we observed that Cx43 expression in the CML microenvironment was decreased, and enhancing the level of Cx43 might be a novel strategy to reverse drug resistance and promote with the effect of IM.

A hypoxic state is common in many chronic tumors.^[20] It was reported that hypoxia can induce Cx43 dysregulation in H9c2 cardiomyocytes.^[16] In a rat model, hypoxia was shown to lead to cell apoptosis and decreased Cx43 expression in BMSCs.^[12] As HM serves as the sustenance for hematopoietic cells and leukemia cells, we detected the expression of Cx43 and HIF-1α in normal BMSCs and found a negative correlation between Cx43 and HIF-1α. Reversing the hypoxic state may be a good way to increase Cx43 expression in HM of patients and further improve treatment efficacy.

Cx43 is regarded as a tumor suppressor in many kinds of tumors due to its antitumor function.^[21] Previous studies on the formation of MRD and drug resistance in leukemia have mainly focused on the adjacent cells and secretory cytokines around leukemic cells.^[7,8] In addition, some studies reported that exosomes secreted from mesenchymal stromal cells could also mediate the antileukemia process.^[22,23] In this study, K562 cells were studied, as they are a classical and representative cell line established from human CML patients. We proved that Cx43-mediated apoptosis was gap junction dependent via a transwell assay and found that the lack of Cx43 can also arrest the cell cycle of K562 cells in G0/G1 phase, which is beneficial for escaping chemotherapy. Animal experiments showed that nude mice bearing K562 cells plus BMSCs-overCx43 manifested the smallest tumor volume under IM treatment, which indicated that increased Cx43 in HM can help enhance the killing effect of IM and delay the progression of diseases in vivo. Interestingly, we observed that the expression of Cx43 in K562 cells and BMSCs is reciprocal: Cx43 overexpression in BMSCs can improve the expression of Cx43 in K562 cells, and vice versa. This observation is suggestive and worth further exploration.

It has been reported that gap junctions can facilitate the direct cytoplasmic delivery of small molecules or metabolites and that Cx43 regulates Ca²⁺ entry and ATP release.^[24] ATP and Ca²⁺ are two major components that can be released from Cx43 channels and affect cellular function.^[25,26] Ca²⁺ is known as a proapoptotic signaling molecule.^[6] In the present study, we found increased Ca²⁺ fluorescence intensity in K562 cells cocultured with BMSCs-overCx43, whereas no difference in ATP concentration was detected among the three IM-treated groups, which implies that Ca²⁺, but not ATP, plays a vital role in the regulation of K562 cells through GJIC.

CaMKII is a general integrator of Ca²⁺ signals, which are ubiquitously expressed in multiple systems and cell types.^[27] It has been reported that overexpression of CaMKII leads to disturbance of Ca²⁺ levels and promotes apoptosis of myocardial cells. Liu et al^[18] demonstrated that Ca²⁺–CaMKII signaling as well as ROS could mediate the inhibition of the Akt/mTOR pathways and finally lead to neural cell apoptosis. The downstream mTOR signal is involved in many metabolic processes of tumor cells, and its activated formation is recognized as a signal of cell proliferation. In this study, in addition to the classical mitochondrial-induced apoptotic pathway, the novel mitochondrial-independent CaMKII–Akt–mTOR pathway was assessed and established [Figure 7].

Figure 7: Mechanism of BMSCs-overCx43 on reversing drug resistance and promoting the efficacy of IM. BMSC: Bone marrow stromal cell; HIF: Hypoxia-inducible factor. Akt: Protein kinase B; BAX: BCL2-associated X; BCL-2: B cell lymphoma 2; MSCs: Bone marrow stromal cells; BMSCs-NC: BMSCs transfected with adenovirus-negative control-green fluorescence protein; BMSCs-overCx43: BMSCs transfected with adenovirus-Cx43 overexpression-green fluorescence protein; CaMKII: Ca²⁺/calmodulin-dependent protein kenases II; Cx43: Connexin 43; CytC: Cytochrome C; HIF-1α: Hypoxia-inducible factor 1α; IM: Imatinib mesylate; mTOR: Mammalian target of rapamycin; PARP: Poly (adenosine diphosphate-ribose) polymerase-1.

There are still some limitations in this study. We did not block the core molecule Ca²⁺ in K562 cells because Ca²⁺ is very important to cell survival, and chelating Ca²⁺ with the chelating agent ethylenebis (oxyethylenenitrilo) tetraacetic acid (EDTA) would lead to the death of both K562 cells and BMSCs. This study also gives us some new ideas. As Cx43-overexpressing BMSCs indeed improve the killing effect of IM and inhibit drug resistance to some degree, we can focus on repairing GJIC function in the HM for leukemia patients before chemotherapeutic resistance develops and even before transplantation. There are two ways to repair GJIC function in BM. All-trans retinoic acid (ATRA) has been used in the treatment of acute promyelocytic leukemia (APL) and other hematologic diseases based on the induction of differentiation and inhibition of growth. It has also been reported that ATRA treatment can increase Cx43 expression in leukemic BMSCs at both the mRNA and protein levels and result in intercellular communication.^[28] In addition, coinfusion of human umbilical mesenchymal stem cells after HSCT is another way to repair GJIC function, eliminate MRD, and inhibit relapse and drug resistance, as umbilical mesenchymal cell infusion has been widely used in clinical experiments with a variety of advantages.^[29]

In conclusion, this study demonstrated that hypoxia-induced downregulation of Cx43 is common in CML HM and benefits drug resistance. Cx43 overexpression in BMSCs weakened the protective effect of BMSCs and enhanced IM-induced apoptosis via Ca²⁺-dependent GJIC.

Funding

This work was supported by the National Key R&D Program of China (2022YFA1103300), the National Natural Science Foundation of China (81873424, 81570097), the Natural Science Foundation of Chongqing Innovation Group Science Program (cstc2021jcyj-cxttX0001), Clinical Medical Research Project of Army Medical University (2018XLC1006), and Translational Research Grant of NCRCH (2020ZKZC02).

Conflicts of interest

None.