CD274 promotes cell cycle entry of leukemia-initiating cells through JNK/Cyclin D2 signaling

C57BL/6 mice were purchased from the Shanghai SLAC Laboratory Animal Co. Ltd., China. The CD274 knockout mice with a C57BL/6 background were kindly provided by Dr. Lieping Chen from the Johns Hopkins University School of Medicine (Baltimore, MA, USA). All animals were housed under specific pathogen-free conditions at the Laboratory Animal Care-approved facility of Shanghai Jiao Tong University School of Medicine, China. All animal experimental procedures were approved according to the Guide for Central Animal Care and Use of the Committee of Shanghai Jiao Tong University School of Medicine.

MLL-AF9-expressing retroviruses were produced in 293T cells by co-transfection with the MSCV-MLL-AF9-IRES-YFP encoding plasmid and the pCL-ECO packaging plasmid. Then, Lin⁻ fetal liver cells were isolated and infected with MLL-AF9 retroviruses with 4 μg/mL polybrene and centrifuged at 1500 rpm for 2 h at 37 °C as previously described [24]. Cells were cultured overnight in StemSpan SFEM medium (StemSpan, USA) with 20 ng/mL SCF, 20 ng/mL IL-3, and 10 ng/mL IL-6, followed by another round of spin infection. Next, infected cells (2.5 × 10⁵) were transplanted into lethally irradiated (10 Gy) C57BL/6 mice by retro-orbital injection. Further, indicated numbers (0.2–0.4 × 10⁴) of leukemia cells from the primary leukemic mice were injected into the recipient mice for secondary transplantation.

Analyses of leukemic lineages and apoptosis were performed as described earlier [6]. Briefly, for analysis of lineages and LICs, bone marrow cells were stained with anti-mouse Mac-1-PE, anti-mouse Gr-1-APC, anti-mouse CD3e-PE, anti-mouse B220-PE, and anti-mouse c-Kit-APC monoclonal antibodies (eBioscience, USA). For detection of CD274 expression in Mac-1⁺/c-Kit⁺ LICs of murine AML model, anti-mouse CD274-biotin and streptavidin-PE (secondary antibody) were used (eBioscience, USA). Cell cycle status was determined in purified Mac-1⁺/c-Kit⁺ LICs with Pyronin Y and Hoeschst 33342 staining (Sigma, USA) as previously described [25]. Apoptosis analysis was conducted in purified Mac-1⁺/c-Kit⁺ LICs with anti-Annexin V-PE and 7-AAD staining (BD Pharmingen, USA) according to the manufacturer’s instructions.

Mac-1⁺/c-Kit⁺ LICs of the murine AML model were sorted by flow cytometry for further RNA extraction. First-strand cDNA was transcribed using Reverse Transcriptase XL (AMV) (Takara, Japan) and allowed to react with the following primers (10 μM/L) containing 2 × ABI SYBR® Green PCR master mix to measure the expression of the studied genes in AML LICs. Details of the primer sequences used are shown in Additional file 1: Table S1.

A combination of plasmids of PLVX-mouse CD274-Strep II, XZ201-mouse Cyclin D2-HA, and CMV5.1-JNK-Fc were transfected into 293T cells followed by a co-immunoprecipitation (co-IP) process to further analyze their interaction. Alternatively, the CD274 or JNK overexpressed plasmids were transfected into 293T cells to observe the expression of Cyclin D2, p-JNK, and JNK. Whole cell lysates and co-IP samples were electrophoresed on 10% sodium dodecyl sulfate polyacrylamide gels and transferred onto polyvinylidene difluoride membranes (Millipore, USA). After electrophoresis and membrane transfer, the immunoblots were probed with the following primary antibodies: anti-mouse Strep II (Genescript, USA), anti-mouse HA (CST, USA), anti-phospho JNK (Abways, China), anti-JNK (Abways, China), anti-Cyclin D2 (Boster, China), and anti-β-actin (Calbiochem, USA).

Total RNA from 10 mg tissue was isolated with depleting genomic DNA for RNA-seq library construction following the standard TruSeq RNA sample preparation v2 protocol (Illumina). The sequencing libraries were then sequenced using the Illumina HiSeq2500 platform. From the reads averaging 50 bp in length, we generated 16.5 ± 1.3 million reads per sample. Further, we aligned the reads to the mouse reference genome (GRCm38, Ensembl build) using Tophat (version 2.0.12), yielding an average mapping rate of 90.3 ± 2.3%. Gene expression levels, which were represented as fragments per kilobase per million mapped reads (FPKM), were obtained for 27,180 genes/transcripts. The RNA-sequencing data were deposited to the Gene Expression Omnibus (GEO) repository under number GSE85193.

Gene ontology enrichment analysis was carried out by the Bioconductor package “topGO,” and KEGG pathway enrichment analysis was conducted by the Bioconductor package “GSEABase” (http://​www.​r-project.​org/​; http://​www.​bioconductor.​org/​packages/​release/​bioc/​html/​GSEABase.​html). Terms were accepted if they were hit more than one gene, and Fisher’s exact test P value was <0.05.

The indicated number of wild-type (WT) and CD274-null leukemia cells were sorted and plated in methylcellulose (M3534, Stem Cell Technologies) according to the manufacturer’s protocols. The numbers of colonies were calculated after 7–10-day culture. In some cases, the lentiviral vector pLKO.1-GFP was used to express shRNAs designed to target CD274 (sequences listed in Additional file 1: Table S1). WT and CD274-null Mac-1⁺/c-Kit⁺ LICs were infected with shRNA targeting JNK and sorted by flow cytometry, then the cells were cultured both in solution or methylcellulose medium. The cell and colony numbers were calculated at indicated time points.

Statistical analysis was performed using GraphPad and SPSS software program, version 19.0. Statistical differences between groups were determined by Student’s t test. The Kaplan-Meier method with log-rank test was utilized to compare survival data among groups. Results were expressed as means ± SEM. A probability level of P < 0.05 was regarded as statistically significant.

To explore the role of CD274 in leukemogenesis, we first examined the expression of CD274 in phenotypic Mac-1⁺/c-Kit⁺ LICs and different cell population of normal bone marrow hematopoietic cells by quantitative RT-PCR. As illustrated in Fig. 1a, b, CD274 was expressed on LICs at a level that was higher than that in normal hematopoietic stem cells (HSCs), hematopoietic progenitors (LK, Lin⁻ cells), and differentiated cells (Lin⁺ cells). The CD274 level on LICs was further confirmed by flow cytometric analysis (Fig. 1c). Then, we investigated the function of CD274 in the AML model induced by the MLL-AF9 oncogene (tagged with yellow fluorescent protein (YFP)) as previously described [6]. Although no significant difference was observed between the frequencies of YFP⁺ leukemia cells from WT and CD274-null leukemic mice (Fig. 1d, e), the mice transplanted with MLL-AF9-induced CD274-null hematopoietic stem/progenitor cells developed leukemia somewhat more slowly than their WT counterparts (Fig. 1f, p < 0.05). We further performed a secondary transplantation with 4000 WT and CD274-null AML cells from primary recipients and found that the percentage of CD274-null YFP⁺ leukemia cells in the peripheral blood was notably reduced compared to that in the WT ones (42.03 ± 2.15 vs 16.02 ± 1.79, Fig. 1g, h) three weeks after transplantation. More importantly, the recipients transplanted with CD274-null AML cells had significantly extended survival (Fig. 1i, p < 0.05). Besides, our previous study showed that CD274 deletion had no effect on normal HSCs [26]. More strikingly, the recipients transplanted with 2000 CD274-null leukemia cells had significantly prolonged survival compared to that of their WT counterparts (Fig. 1j, p < 0.05).

Then, we further examined the lineages of CD274-null leukemia cells and found that the differentiation status was not altered according to the analysis with the surface markers of Mac-1 and Gr-1 (Mac-1⁺/Gr-1⁺ cells representing a more mature leukemia cell population) (Additional file 2: Figure S1A-B). This finding suggests that the delayed AML development upon CD274 deletion might not have resulted from the enhanced differentiation. In addition, exceedingly few CD3⁺ or B220⁺ (markers for lymphoid lineages) leukemia cells were observed in the recipient mice, indicating the characteristics of a myeloid leukemia model (Additional file 2: Figure S1C). Consistently, we also found that the values of the size and weight of the spleen of the recipients that received primary CD274-null leukemia cells were lower than those of the WT controls (Additional file 2: Figure S1D-F). The histological hematoxylin/eosin staining also revealed that there were much less infiltrated leukemia cells in the spleen of the mice transplanted with CD274-null leukemia cells (Additional file 2: Figure S1G).

Interestingly, the in silico analysis of data extracted from the curated databases (the HemaExplorer, http://​servers.​binf.​ku.​dk/​hemaexplorer/​) or Leukemia Gene Atlas (LGA) (http://​www.​leukemia-gene-atlas.​org/​LGAtlas/​) showed that the level of expression of CD274 in AML cells was much higher than that in normal hematopoietic stem cells, which was inversely correlated to the overall survival of AML patients (Additional file 3: Figure S2A-B) [27]. Taken together, CD274 maintains the proliferation of LICs and may serve as a potential biomarker for AML.

To identify the underlying mechanisms of CD274 functions in the stemness regulation of LICs, we further examined the LIC frequency in the bone marrow of primary and secondary recipients. Although no significant difference in LIC (Mac-1⁺/c-Kit⁺ cells) frequency was found between the WT and CD274-null recipients upon the primary transplantation (Fig. 2a, b), the LIC frequency in the CD274-null recipients was considerably lower than that in the WT ones after the secondary transplantation (Fig. 2c, d). Consistently, the results of the colony-forming unit assay showed that CD274-null AML cells generated much lower colony numbers and total cell numbers than WT controls (Fig. 2e–g), indicating that CD274 depletion led to a notable reduction in the proliferation potential of LICs.

To understand how CD274 controls the growth of LICs, we determined the cell cycle status of Mac-1⁺/c-Kit⁺ LICs by Hoechst 33342 and Pyronin Y staining. Intriguingly, we found that most of the CD274-null LICs were arrested in the G1 phase unlike their WT counterparts (Fig. 2h, i). In addition, there was no significant difference between WT cells and CD274-null LICs in LIC differentiation upon the secondary transplantation (Additional file 4: Figure S3A-B). Furthermore, no distinct difference was found between the apoptosis in WT and CD274-null AML cells from either primary or secondary transplantation (Additional file 4: Figure S3C-F). These data indicate that CD274 controls the cell cycle entry to maintain the pool of LICs in the bone marrow.

To determine the potential targets of LICs controlled by CD274, we performed RNA-sequencing with WT and CD274-null LICs. After aligning the reads to the mouse reference genome, we obtained an average mapping rate of 90.3 ± 2.3%. Gene expression levels, which were represented as FPKM, were evaluated among 27,180 genes/transcripts. Using a cutoff of P value <0.01 and a fold change of >1.5, a total of 457 candidate genes were characterized by comparing CD274-null with WT groups. Gene ontology analysis revealed that the differentially expressed genes were mainly involved in the biological process of many “immune-related activities” (Fig. 3a). Intriguingly, GO analysis indicated CD274 might also play a key role in the process of “protein kinase cascade” and “intracellular signaling cascade” (Fig. 3a). The KEGG analysis further revealed that these genes were enriched in the “MAPK signaling pathway” or “Hematopoietic cell lineage” (Fig. 3b). To investigate the reduced proliferation abilities and cell cycle arrest in the G1 phase in CD274-null LICs, we further analyzed a number of candidate genes related to proliferation and cell cycle activators or inhibitors (Additional file 5: Figure S4). Although the candidate genes related to proliferation were not significantly changed, several cell cycle regulators, including p16, p21, Cyclin D2, and CDK6, were markedly up- or downregulated (Additional file 5: Figure S4). The RNA-sequencing results indicated that CD274 might be involved in the AML development through the cell cycle regulation.

To confirm whether these cell cycle-related genes were potential downstream targets of CD274, we examined their expressions in WT and CD274-null LICs by quantitative RT-PCR and revealed that the levels of both p21 and p16 were increased, whereas those of Cyclin D2 and CDK6 were decreased in CD274-null LICs compared to WT ones (Fig. 3c). Since the expression of Cyclin D2, which is a key regulator in promoting G1-S transition of cell cycle [28], was most significantly downregulated (reduced to 10% of that in WT LICs) among these candidate genes (Fig. 3c), we decided to determine whether Cyclin D2 served as a downstream target for CD274 in leukemogenesis. We further confirmed that Cyclin D2 level consistently decreased in CD274-null LICs by Western blotting analysis (Fig. 3d). Cyclin D2 was then ectopically expressed in the CD274-null AML cells, followed by the transplantation into the recipient mice to test whether it could rescue the loss of function of CD274. Importantly, the recipient mice transplanted with Cyclin D2-overexpressed CD274-null AML cells developed AML much faster than the CD274-null AML counterparts, which was comparable to the case in the WT controls (Fig. 3e, f). Meanwhile, the blockage of G1 to S phase transition in CD274-null LICs was totally reversed upon Cyclin D2 overexpression (Fig. 3g).

These results indicate that as a direct downstream target of CD274, Cyclin D2 is responsible for AML development. In line with our findings, we also found that Cyclin D2 was significantly upregulated when CD274 was overexpressed in 293T cells (Additional file 6: Figure S5A). The in silico analysis of the curated database or LGA data also showed that the level of expression of Cyclin D2 was much higher in AML cells than in normal hematopoietic stem cells and was inversely correlated with the overall survival of AML patients (Additional file 6: Figure S5B-C) [27].

Because the RNA-sequencing results implicated that MAPK signaling pathway might be involved in the CD274 function in leukemogenesis, and a previous study also suggested that JNK signaling may manipulate the activity of Cyclin D2 [29]. We examined the JNK signaling in WT and CD274-null LICs by Western blotting analysis. As shown in Fig. 4a, the phosphorylation level of JNK was significantly decreased in CD274-null LICs. Consistently, phospho-JNK expression was remarkably elevated upon CD274 overexpression in 293T cells (Fig. 4b). Interestingly, the co-immunoprecipitation experiment revealed that both JNK and phospho-JNK were pulled down by CD274 (Fig. 4c). Meanwhile, CD274 was also detected while JNK was immunoprecipitated (Fig. 4d). Moreover, the expression of Cyclin D2 was significantly upregulated upon JNK overexpression in 293T cells (Fig. 4e). We have further knocked down the expression of JNK in both WT and CD274-null LICs with a shRNA, and the knockdown efficiency was confirmed by Western blotting (Fig. 4f). We demonstrated that the knockdown of JNK caused a dramatic decrease in the proliferation of both WT and CD274-null LICs by solution culture in vitro (Fig. 4g). More importantly, the functional analysis of colony formation units also showed that both the colony and total cell numbers from the JNK-knockdown WT LICs were much fewer compared to the control (Fig. 4g–j). As expected, the knockdown of JNK also led to a less reduction in both the colony and total cell numbers from CD274-null LICs (Fig. 4g–j). In summary, we demonstrate that CD274 directly interacts with JNK to enhance its phosphorylation, which further increases the Cyclin D2 levels to promote cell cycle entry and proliferation (Fig. 4l).