CD45⁺CD326⁺ Cells are Predictive of Poor Prognosis in Non–Small Cell Lung Cancer Patients

Lung cancer is one of the most common types of malignant tumor in the world (1, 2), and non–small cell lung cancer (NSCLC) accounts for around 80% of lung cancer cases (3, 4). Many studies have shown that cancer cells are heterogeneous; such intratumoral diversities may help tumor cells to survive against immune system surveillance and a variety of treatment methods (5–7). A malignant effusion from patients with cancer is considered to be a poor prognostic sign, and tumor heterogeneity is also an important clinical indicator of patient outcomes. For example, higher tumor cell ratios in ascites from gastric cancer patients can contribute to poorer patient outcomes (8). The epithelial-to-mesenchymal transition (EMT) in cancer is known to be associated with therapeutic resistance, aggressiveness, and metastasis (4–6, 9, 10). However, the existence of doubly-positive for epithelial and nonepithelial surface markers, has not previously been demonstrated. Whether or not such CD45/CD326 doubly-positive cancer cells really exist is one of the most important questions in current oncology. If CD45/CD326 doubly-positive cells (CD45/CD326 DPC) can be identified in cancers, they would be promising targets for future research to elucidate the therapeutic resistance, aggressiveness, and metastatic ability of NSCLC.

In this study, we analyzed tissues from 39 patients with NSCLC. Twenty-one were samples of primary solid tumor tissues and 18 were samples of malignant pleural effusion. We attempted to identify CD45/CD326 doubly-positive cancer cells by FACS, fluorescence microscopy, and EGFR mutation analyses. The possible association between the frequency of CD45/CD326 DPC and poor prognosis was also examined. Our present results will contribute to understanding the effect of CD45/CD326 doubly-positive cancer cells on patient outcomes.

Samples from 39 patients with NSCLC were collected between 2010 and 2017 with written informed consent in accordance with the Declaration of Helsinki. The study was approved by the Ethics Committee of the Tohoku University School of Medicine under the accession number 2016-1-452. Tumors were procured within 2 to 3 hours after excision. Resected samples were kept in DMEM/F12 medium supplemented with 2% FBS (Thermo Fisher Scientific) for single cell dissociation. Solid tumors were minced into small pieces which could pass through a 5 mL pipette tip and digested with collagenase 1A solution (Stem Cell Technologies) in a 50 mL tube at 37°C on a shaker at 300 rpm agitation for 1 to 2 hours. The collagenase 1A was inactivated by addition of an equal amount of Iscove's modified Dulbecco's medium (IMDM; Thermo Fisher Scientific) supplemented with 5% FBS. Red blood cells were lysed in 1× ammonium–chloride–potassium (ACK) buffer (pH 7.2). As summarized in Supplementary Table S1, total of 39 patient samples, surgically resected cancer tissues from 21 patients and malignant pleural effusion samples from another 18 patients were used in this study.

Single-cell suspensions were obtained by 2-step sequential filtration using a 100- and a 40-μm mesh and then counted. After the cell preparation, living cells were collected by FACSAria II (BD Bioscience). Debris was excluded by proper gating, and dead cells were excluded with propidium iodide (PI). The remaining living cells were further sorted into single cells and clumping cells by FACSAria II. The clumping cells were detected in groups of varying sizes by using forward scatter (FSC-W) and (FSC-H) and side scatter (SSC-W) and (SSC-H) sequentially. After these sorting processes, single cells without debris and clumping cells were separately purified for detection of genetic alterations and cell surface marker expression.

In this study, CD326 (Miltenyi Biotec), an epithelial cell adhesion molecule, was used as the cell surface marker for epithelial cells (11). CD45 (Miltenyi Biotec), a leukocyte cell surface glycoprotein, was used as the surface marker for mesenchymal cells (12). The fluorescence-labeled antibodies against these surface markers (CD326 PE 130-113-264 1:50, CD45 FITC 130-113-17; 1:10) were incubated in the dark with samples at 4°C for 30 minutes, washed 3 times with FACS buffer, and filtered through a 35-μm mesh before the FACS analyses. Single cells from cancer samples were sorted according to the positivity of CD45 and CD326 on their stained cell surfaces and categorized into the following groups: (i) CD45^neg/CD326^neg cells, (ii) CD45^neg/CD326^pos epithelial cancer cells, (iii) CD45^pos/CD326^neg mesenchymal cells, and (iv) CD45^pos/CD326^pos cells. After sorting, the cells were collected and observed by fluorescence microscopy.

Immuno-fluorescence microscopy used a N-SIM (Nikon Instech Co.), anti-pan Cytokeratin-antibody (KL1), which is commonly used for detection of epithelial cells (1:500; Cosmobio Ltd.), and anti-CD45-antibody (1:200) (CST 13917S, Cell Signaling Technology) to visually confirm the existence of CD45^pos/CD326^pos cancer cells in solid cancer samples. Two secondary antibodies were used: Alexa Fluor 488, a goat anti-mouse IgG antibody that emits a green color, was tagged on anti-KL1-antibody, (1:1,000, Fisher A-11001), and CY3 conjugated goat anti rabbit IgG antibody that emits a red color was tagged on anti-CD45-antibody (1:1,000, Abcam ab6939). Detection of cancer cells in pleural effusion employed 2 antibodies also used in the FACS analyses; PE-labeled anti-CD326-antibody (red) and FITC-labeled anti-CD45-antibody (green). Because merged immunofluorescent images of red and green colors result in yellow emission light, any CD45^pos/CD326^pos cells in the cancer samples will appear yellow in this protocol. 4′,6-Diamidino-2-phenylindole (DAPI) was used for nucleic acid staining to confirm that the substances bound by these labeled antibodies were actually living cells.

To confirm that the CD45^pos/CD326^pos cells are truly cancer cells that harbor cancer-causing EGFR mutations, we performed the following analyses. Cells from the pleural effusion were categorized into the above-mentioned 4 groups, and their DNAs were purified using NucleoSpin Tissue XS (MACHEREY-NAGEL; GmbH & Co. KG), according to supplier's recommendation. Target EGFR regions were amplified by the GeneAmp PCR system 9700 [Applied Biosystems (13)]. Nucleotide sequences were determined by methods described previously (14) using an ABI PRISM310 Genetic Analyzer (Applied Biosystems). Because the EGFR mutations in our patients were all indel type, we could also simply have confirmed the mutations by polyacrylamide gel electrophoreses. Nucleotide sequences of the primers and PCR conditions are summarized in Supplementary Table S2.

We detected a wide range of proportions of CD45^pos/CD326^pos single cells in our samples by FACS (Fig. 1A). These single cells were strong candidates for EMT, so we investigated them further. They were detected in both solid cancer tissue (Fig. 1B) and malignant pleural effusion (Fig. 1C). A serum specimen extracted from a healthy middle-aged male volunteer was used as the control; no CD45/CD326 DPC were detected in it (Fig. 1A). The average size of the CD45/CD326 DPC population was 5.5% (0.4%–17.9%) in solid tumors and nearly 7 times higher, 34.5% (1.1%–96.7%), in malignant pleural effusions.

Using KL1 and CD326 antibodies for epithelial cell markers and an immuno-fluorescence microscope, we visually confirmed the existence of CD45/CD326 DPC (CD45^pos/CD326^pos cancer cells by FACS analyses) in samples with solid tumor tissue or malignant pleural effusion. Figure 2A illustrates immune-fluorescence microscopic images of an adenocarcinoma sample (T1aN0M0; stage IB). In the merged image, the KL-1^pos/CD45^pos doubly-positive cancer cells are clearly present (yellow arrows). The fluorescence of sorted pleural effusion cells from 4 quadrants (Q1–Q4) are shown in Fig. 2B; doubly-positive cells are present only in Q2, not in the other quadrants. Papanicolaou staining was done after sorting the malignant pleural effusion, and the sorted cells in each subpopulation were microscopically observed for positivity of CD45 and CD326 in Q1 through Q4. Q1 and Q2 contained malignant cells with large and irregular nuclei indicated by arrowheads. Based on these findings, CD45^pos/CD326^pos cells are considered to be CD45/CD326 doubly-positive malignant cells.

Mutations of the EGFR gene were analyzed in all the 18 patients with pleural effusion. Five of them had disease-causing mutations, all of which were indel in either exon 19 or 20. We PCR-amplified target mutant regions in 3 of these 5 patients followed by 10% polyacrylamide gel electrophoresis. As shown in Fig. 3A, every DNA sample from CD45^pos/CD326^pos (doubly-positive) cells showed the same mutation observed in its corresponding patient. Notably, a strong upper mutant band was observed, but only a weak normal-sized band was detected in DNAs from Q1 and Q2 in Case #1; this strongly suggested a homozygous mutation in the cancer cells. Based on these results, the CD45/CD326 DPC were confirmed to be malignant from the genetic point of view. Nucleotide sequencing analyses were also performed; a typical example is shown in Fig. 3B. A 9-base insertion in exon 20 of EGFR, observed in Case #1, was detected only in Q1 and Q2, the CD326^pos fraction. In contrast, the CD326^neg group did not have such mutations. We summarized all the mutations detected in our study in Table S3. In addition, mutations of KRAS and TP53 as the typical driver genes for NSCLC were analyzed as shown in Supplementary Fig. S1.

Next, we examined the CD45^pos/CD326^pos cells impact on patient outcomes in the 18 patients with pleural effusion. The proportion of CD45^pos/CD326^pos cells in their entire CD326^pos cell population were calculated and their relationship to patients' prognoses was plotted (Fig. 4). Two patients had over 90% of CD45^pos/CD326^pos cells, although number of cancer cells we could have analyzed was not many (data not shown). Four of the 18 (22%) patients remain alive during this cohort study. Three of these 4 survivors have the EGFR mutation and are being treated with tyrosine kinase inhibitor (TKI); the remaining survivor is being treated with PD-1. We further investigated whether there were any associations between CD45^pos/CD326^pos cell proportions and patients' prognoses. We divided the patients into 2 groups based upon their average CD45^pos/CD326^pos cell proportion and found that patients with lower proportions had better prognoses (P = 0.0213 by χ² test, P = 0.0686 by Fisher exact test).

In this study, we showed that varying percentages of CD45^pos/CD326^pos cells existed in all of the 21 solid tissue samples and 18 pleural effusion samples obtained from 39 patients with NSCLC. These doubly-positive cells are highly suspected to be EMT. They appeared to be malignant microscopically and had cancer-causing genetic mutations. Based on these results, we propose that CD45/CD326 doubly-positive cancer cells exist in both solid lung tumor tissues and pleural effusion in patients with NSCLC. Our present report is the first one demonstrating the evidence of such a cell type.

The phenomenon of EMT in cancer tissues is recognized as an important element in metastasis (15–19). However, EMT studied have searched at the tissue level or in cultured cells and/or tissues (20, 21). One previous report mentioned cells with detectable dual epithelial and mesenchymal characteristics in ascites from patients with gastric cancer (8). However, the existence and significance of CD45/CD326 DPC in primary tumors requires more scrutiny. This is the first report demonstrating the integrity of such cells. We used the EGFR mutation as the marker for cancer cell evaluation and monitored its presence in each analyzed sample and found that the CD45^pos/CD326^pos cells as well as the CD45^neg/CD326^pos (epithelial-only singly positive) cells have the same mutation as does each corresponding primary tumor. These indicate that the doubly-positive cells originated from cancer cells. Utilizing these doubly positive cells, we uncovered the intratumoral heterogeneity in the tissues from the patients in this study.

Elucidation of mechanisms of metastasis in solid tumors is a highly important area in cancer research. Epithelial malignant cells need to move via blood vessels or lymphatic ducts in spite of massive immune cell attacks; this is one of the most difficult of current topics (22, 23). EMTs have been suggested as enabling metastasis (24, 25). If some proportion of the epithelial cancer cells can transiently deform to mimic mesenchymal cells, they may be more able to move to other organs, to escape from attacks by immune systems, and to finally accomplish the process of metastasis. The existence of circulating tumor cells may be one example of an EMT process that can lead to worse outcomes (26–28), although precise mechanisms are yet to be determined. Such circulating tumor cells may include the CD45/CD326 DPC we observed in this study. Surprisingly, we observed that the percentages of these CD45/CD326 DPC are higher in malignant effusions than in the solid tumors (Fig. 1B and C). Furthermore, as shown in Figure 4, 2 cases with pleural effusion have over 90% CD45/CD326 DPC, and the total number of cancer cells was lower in these cases. It is possible that CD45/CD326 DPC are more able to infiltrate in pleural effusion than other cancer cells, although the precise mechanisms are not yet understood. It is even possible that such CD45/CD326 DPC may also be able to make the mesenchymal-to-epithelial transition (MET) in addition to the EMT, which is certainly suggested as a requirement in the process of metastasis (29).

In this study, we analyzed primary solid tumor tissue and pleural effusion, but metastatic tumor tissues and serum from these patients were not available. Further investigations using cells from metastatic lesions and/or patients' sera are necessary to verify our hypothesis of the contribution of CD45/CD326 DPC to metastasis. However, detection of CD45/CD326 DPC by liquid biopsy is not technically easy at present (30). Additional investigations using noncancerous samples such as pleural effusion from patients with nonmalignant diseases as controls will add valuable information to future investigations.

In conclusion, we have reported strong indications of the existence of CD45/CD326 DPC in a wide variety of concentrations in both the primary tumor and malignant pleural effusion from patients with NSCLC and shown that a high percentage of such CD45/CD326 DPC is an important prognostic factor. CD45/CD326 DPC will be promising targets for future investigations, particularly in research examining cancer metastasis.

S. Kobayashi reports receiving speakers bureau honoraria from Bristol-Myers Squibb, Taiho Pharmaceutical Co., Ltd., and AstraZeneca. No potential conflicts of interest were disclosed by the other authors.

Conception and design: K. Ishizawa, T. Akaishi, A. Horii

Development of methodology: K. Ishizawa, M. Yamanaka, N.-A. Pham

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Ishizawa, Y. Saiki, E. Miyauchi, S. Fukushige, T. Akaishi, A. Asao, R. Saito, Y. Tojo, A. Sakurada, Y. Okada, M.-S. Tsao

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Ishizawa, M. Yamanaka, S. Fukushige, T. Akaishi, T. Mimori, N. Ishii, S. Kobayashi, M. Nagasaki

Writing, review, and/or revision of the manuscript: K. Ishizawa, M. Yamanaka, E. Miyauchi, S. Fukushige, T. Akaishi, T. Ishii, M.-S. Tsao, A. Horii

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Ishizawa, M. Yamanaka, E. Miyauchi, A. Asao, M. Li, Y. Okada S. Kobayashi, M. Ichinose, A. Horii

Study supervision: K. Ishizawa, T. Mimori, R. Yamashita, M. Abe, T. Ishii, N. Ishii, M. Nagasaki, A. Horii

We are grateful to Dr. Barbara Lee Smith Pierce (Adjunct Professor, University of Maryland University College) for her editorial work in the preparation of the manuscript. We gratefully acknowledge Drs. Makoto Kobayashi, Osamu Usami, Taizo Hirano, Naoki Tode, Teruyuki Sato, Tomohiro Ichikawa, Yusaku Sasaki, and Tatsuma Okazaki from Department of Respiratory Medicine, Tohoku University, and Ms. Chiharu Kudo, Drs. Hikari Sato, and Yoshikazu Okubo from Japanese Red Cross Ishinomaki Hospital, for their collection of patients' samples. We are also grateful to Biomedical Research Core, Tohoku University School of Medicine for their technical supports. This work was supported in part by Grants-in-Aid for Scientific Research (JP17K09641, 26460468, and 17K17589) from the Japan Society for the Promotion of Science.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.