Enrichment of circulating tumor cells in tumor-bearing ... - Springer Link

2 downloads 0 Views 2MB Size Report
May 24, 2015 - Abstract Concentration of real tumor cells leaking into blood from cancer was attempted by a deterministic lateral displace- ment (DLD) ...
Biomed Microdevices (2015) 17: 59 DOI 10.1007/s10544-015-9964-7

Enrichment of circulating tumor cells in tumor-bearing mouse blood by a deterministic lateral displacement microfluidic device Hiromasa Okano 1 & Tomoki Konishi 1 & Toshihiro Suzuki 2 & Takahiro Suzuki 1 & Shinya Ariyasu 2 & Shin Aoki 2 & Ryo Abe 2 & Masanori Hayase 1,2

Published online: 24 May 2015 # Springer Science+Business Media New York 2015

Abstract Concentration of real tumor cells leaking into blood from cancer was attempted by a deterministic lateral displacement (DLD) microfluidic device. Spiked cultured cell line tumor cells are often used to verify performance of the circulating tumor cells (CTCs) separation methods. Cultured tumor cells are obviously larger than most of hematocytes and considered not to be appropriate as CTC mimics, while there is uncertainty in identifying real CTCs from clinical samples and there is no practical way to examine CTCs leakage into benign cells during the sorting. In this work, blood samples were prepared from tumor-bearing mice whose tumors were induced by implanting cells with GFP expression to living mice. Therefore, CTCs were identified by their fluorescence emission. We succeeded in the enrichment of tumor cells to 0.05% from the blood, in which CTCs were negligibly detected among three million blood cells, and little loss of CTCs was observed.

Keywords Circulating tumor cell . Deterministic lateral displacement . Microfluidics . Tumor-bearing mouse . Cell sorting

* Masanori Hayase [email protected] 1

Department of Mechanical Engineering, Faculty of Science and Technology, Tokyo University of Science, 2641 Yamazaki, Noda-shi, Chiba-ken 278-8510, Japan

2

Center for Technologies Against Cancer, Research Institute of Science and Technology, Tokyo University of Science, 2669 Yamazaki, Noda-shi, Chiba-ken 278-0022, Japan

1 Introduction Metastasis is one of most serious problems of cancer, and the mechanism is still poorly understood (Chaffer and Weinberg 2011). Circulating tumor cells (CTCs) are tumor cells that leak from a tumor into blood vessels, and transfer to distant locations. Some CTCs may survive and proliferate at the distant sites, and have the potential to establish metastatic tumors. Detection and separation of CTCs from blood is necessary for understanding the process and mechanism of metastasis. Enumeration or capturing of CTCs will be useful for the diagnosis and prediction of metastatic progress, and CTCs may also be useful in monitoring the efficacy of anti-cancer therapy both in vivo and in vitro. The development of simple CTC separation methods are therefore desirable, and many studies about the separating CTCs have recently been published (Jin et al. 2014; Allard et al. 2004; Nagrath et al. 2007; Tan et al. 2009, 2010; Bhagat et al. 2011; Hosokawa et al. 2010; Loutherback et al. 2012; Zheng et al. 2011; Liu et al. 2013a, b; Sheng et al. 2013). Although many efforts have been made for CTC separation, capture and analysis of CTCs are difficult, because of their extremely low concentration in blood, with typically a few CTCs among billions of normal blood cells. To separate CTCs from benign cells in blood, two major approaches can be used: biochemical approaches and biophysical approaches. In biochemical approaches, affinity capture of cell surface antigens (Gao et al. 2013; Nagrath et al. 2007; Ariyasu et al. 2012; Ohnaga et al. 2013; Liu et al. 2013a; Sheng et al. 2013) and fluorescent labeling are generally used to identify CTCs. These examples include the Veridex Cellsearch® system (Raritan, NJ, USA), which has been approved by the U.S. FDA (Food and Drug Administration) for clinical

59 Page 2 of 11

enumeration of CTCs. This system utilizes anti-epithelial cell adhesion molecule antibody (anti-EpCAM antibody), which has been often used in recent studies for the recognition of CTCs. However, further studies suggest that metastatic CTCs are able to undergo a epithelial-tomesenchymal transition (EMT), which results in the reduction of EpCAM expression (Chaffer and Weinberg 2011). Affinity capture approaches might miss the aggressive CTCs with EMT ability (Jin et al. 2014). In the biophysical approaches, differences in the physical properties of CTCs from blood cells are used to identify the CTCs, including size, deformability, and electrical polarizability. Previous studies of physical and morphological properties of CTCs indicate that CTCs are generally larger and harder than leukocytes (Alexander and Spriggs 1960; Vona et al. 2000). It has also been reported that the specific gravity of CTCs is greater than that of leukocytes (Seal 1959). Basically, the physical separation approaches do not modify CTCs, and they are considered to have advantage in cell analysis after the separation. Many studies using microfluidics have recently been published with promising results (Tan et al. 2009, 2010; Bhagat et al. 2011; Hosokawa et al. 2010; Zheng et al. 2011; Loutherback et al. 2012; Liu et al. 2013a, b). Deterministic lateral displacement (DLD) devices are often used in these studies for size sorting (Huang et al. 2004; Davis et al. 2006; Holm et al. 2011; Beech et al. 2012; Inglis et al. 2011), for the recovery of the tumor cells spiked in blood were in high yields (Loutherback et al. 2012; Liu et al. 2013a, b). Ideally, DLD has no mesh blinding and a large amount of specimen is processed. No severe damage to blood occurs due to treatment such as lysis, and returning the benign blood to the body after removing suspicious cells will be possible in principle. While DLD is an attractive technique for CTC separation, most studies use cultured tumor cells, which are generally larger and less deformable than leukocytes. Several studies have shown that aggressive tumor cells might be small and more deformable than benign cells (Jin et al. 2014), suggesting cultured tumor cells are not appropriate for demonstrating CTC separation. Enrichment of real CTCs from clinical samples is desirable, but there is still uncertainty in verifying real CTCs and is no practical way to examine CTCs leakage into benign small cell section during the sorting. To evaluate the capability of the DLD size sorting technique for CTC separation, we prepared the blood samples of tumor-bearing mice, in which CTCs could be identified by their fluorescence emission because tumors were induced by implanting cells with GFP expression, and not only separation or enrichment of CTCs but CTCs leakage into benign section were discussed using a simple microfluidic DLD device.

Biomed Microdevices (2015) 17: 59

2 Experimental 2.1 Microfluidic DLD device Figure 1 shows the structure of the micro fluidic channel used for size sorting in this work, in which a small amount of the specimen is segregated into two parts. Based on the DLD theory (Huang et al. 2004), a periodic array of micro-posts, shifted slightly along the fluid stream, was fabricated using Si substrate by conventional photolithography and the DeepRIE (Deep Reactive Ion Etching) process. Though previous studies employed thick specimen streamlines to obtain high process throughput (Loutherback et al. 2012; Liu et al. 2013a, b), we formed narrow specimen streamline in order to understand the sorting behavior more easily. The narrow stream line was formed by supplying specimen into a buffer flow from an inlet nozzle with small volume ratio. By employing this simple design, though the throughput was not high, higher sorting efficiency was expected, and channel cloggings were detected easily. The height of the micro-posts was 50 μm. After formation of micro-posts structure, the Si chip was thermally oxidized in water vapor at 1100°C for 30 min, and a hydrophilic surface was prepared. A flat glass cover with two inlets and two outlet tubes was put on to the patterned Si chip, and the two plates were tightly clamped on a holder by a metal frame. Figure 2 shows a schematic view of our experimental setup. Solutions were injected into the micro channel by our own syringe pumps. A dispersed cell solution specimen was poured into the specimen inlet, and buffer solution was poured into the buffer inlet. The specimen solution was surrounded by buffer solution, and a thin streamline was formed which entered the micro-posts array zone. In that zone, small cells follow the fluid streamline and travel in Bzigzag^ mode and are segregated into the small cell outlet. While large cells travel across the streamline due to interaction with microposts array (Bbump mode^), and flow at a slight angle along the posts array. Then, large cells are segregated into the large cell outlet. To observe the motion of specimens, the microfluidic device was placed under a microscope. In this study, three Si chip configurations having different critical diameters, Dc, were prepared as shown in Table 1. The critical diameters were calculated according to a formula of Inglis et al. (Inglis et al. 2006), assuming a parabolic flow profile between the posts. 2.2 Cultured tumor cells To check the properties of the DLD device, preliminary testings were performed with cultured tumor cells, SP2/O cells (mouse myeloma cells) in this work, in which enhanced green fluorescent protein (EGFP) was overexpressed. Cell lines expressing the EGFP were generated by electroporation with

Biomed Microdevices (2015) 17: 59

Page 3 of 11 59

Fig. 1 DLD device design. In order to grasp the validity of DLD method for the CTC enrichment, simple design with no novel devices was employed. Microfluidic pattern was fabricated on a Si chip by DeepRIE (MUC-21, Sumitomo Precision Products Co., Ltd., Japan). With this simple design, main specimen flow shows thin stream, and experimental failure such as channel clogging can be easily detected. In the SEM image, it is shown that top section of microposts had different tilted directions. In this design, sorting zone was longer than that was needed for theoretical bumpmode displacement, and the pattern was prepared to avoid stagnation of large cells which reach the edge of the channel

pTAZ2-DHFR-living color expression vector including the EGFP gene. One day after transfection, the cells were seeded on 24-well plates and selected in culture medium (Iscove’s modified Dulbecco’s media (IMDM) supplemented with 10% fetal calf serum (FCS)) containing 0.4 mg/ml Zeocine™. To test the expression of EGFP, selected clones were analyzed using a fluorescent activated cell sorting (FACS) system (Beckman Coulter, Inc. Gallios™, Indianapolis, USA). Stable transfectant cell lines were maintained with RPMI 1640, supplemented with 10% FCS, 100 U/mL penicillin, 100 μg/mL

streptomycin, 10 mM HEPES, pH 7.55, and 50 μM 2mercaptoethanol, in 5% CO2 at 37°C. In the experiments, the cells were recovered and prepared with FACS medium (phosphate buffered saline (PBS) containing 0.5% FCS and 0.1% sodium azide), which had been filtrated by Minisart® syringe filter (0.2μm pore size) before use. SP2/O cells are mouse cells, and as a human cell, Jurkat cells (human T lymphocyte cells) were occasionally used. 2.3 Normal mouse blood Female BALB/c mice were obtained from Sankyo labo service (Tokyo, Japan). These mice were maintained under specific pathogen-free conditions and used at 8 to 10 weeks of age. Prior to the experiments, arterial blood was collected from the tail artery in a tube containing heparin solution (Heparin Sodium 10,000units/10mL for Inj., Mochida Pharmaceutical Co., Ltd., Japan) and physiological saline. As a result of Table 1 Dimension of micro-posts array and calculated critical diameters

Fig. 2 Experimental setup. Specimen and physiological saline were supplied by our own made syringe pumps, and segregated specimens were collected in tubes. The microfluidic device was put under an optical microscope (Olympus BX51, Japan), and the specimen flow was monitored during the experiment. The DLD chip and a glass cover were clamped on a chip holder by a metal frame

Size configuration

Big

Middle

Small

Post diameter / μm Shift / μm Gap / μm Critical diameter, Dc / μm

15 3 23 8.05

15 3 19 7.14

15 3 15 6.0

59 Page 4 of 11

the heparin solution volume, the blood was diluted about 3 times. The blood specimen was preserved at room temperature and was used within three hours of collection. All experiments were performed in accordance with protocols approved by the Animal Care and Use Committee of Tokyo University of Science. 2.4 Tumor bearing mouse The SP2/O-EGFP cells prepared as mentioned above were inoculated into the mice, and tumors were induced. One day before tumor transplantation, the mice were sub-lethally irradiated (5 Gy) and inoculated subcutaneously with 5×106 SP2/ O-EGFP cells. 5 to 6 weeks after the tumor transplantation, sarcoma-like large tumor mass was generated on the flank, and a biopsy of cardiac blood was made.

3 Results and discussion 3.1 Sorting of cultured tumor cells In the DLD theory, a critical diameter Dc is estimated assuming a rigid sphere. Living cells are deformable, however, and a shear stress is applied to the cells as a result of the nonuniform fluid velocity in the micro-posts array zone. Cells may thus become elongated and behave like particles of smaller diameter (Liu et al. 2013a, b). A slowing of the cells due to friction when they make contact with the micro-posts is presumed, and flow deviates from a parabolic profile. Fluid flow in the vicinity of the cell is slower, and a wider stream is then needed to carry the fluid due to the slowing. This makes the critical diameter larger than the calculated Dc assuming a parabolic flow profile between the posts. Both effects lead to underestimation of Dc, and cells with diameter larger than calculated Dc may be segregated into small cell outlet, while some papers reported opposite tendency (R. Quek 2011; S. Ye 2014). DLD design for such deformable materials is not established, and the separation behavior was examined experimentally using chips with three different channel configurations. To confirm the DLD function, sorting of cultured tumor cell was tested first. SP2/O cells were dispersed in cultivation medium (PBS+FCS) at a concentration of 2×105 cells/ml. The dispersed SP2/O cells and physiological saline were injected into the specimen and buffer inlets respectively at flow rates of 7 μl/min and 190 μl/min for 20 min. The partitioned specimens were collected, and after centrifugation, 1% of total volume (about 300 cells) were carefully observed on hemocytometer having 0.1mm depth by optical microscopy. Figure 3 shows some examples of results. With the Bmiddle^ and Bsmall^ configuration chips, no cells were found in the small cell outlet. 16 cells were observed in the small cell outlet with the Bbig^ configuration chip, and the

Biomed Microdevices (2015) 17: 59

number of cells collected in the small cell outlet was about 1/20 of the large outlet cells. Figure 4 shows a microscopic view of SP2/O cells taken from a usual cultivation condition. The minimum cell size was 9–10 μm in diameter among 108 cells. As we presumed, the minimum diameter was larger than the calculated critical diameter, Dc =8.05 [μm], of the Bbig^ configuration. Cell viability was also checked by PI (Propidium Iodide) staining, and more than 99 % cells showed negative fluorescence which means that the cells were alive. No significant impact on cell viability by the sorting was observed, and it was reconfirmed that the DLD is an attractive method for less invasive cell sorting.

3.2 Sorting of mouse blood In the DLD device, most normal blood cells are expected to segregate into the small cell outlet. Blood is a heterogeneous mixture, and the size, shape and deformability of blood cells are highly variable. Several studies have already been made using whole blood, but the separation behavior is not well understood. The channel design was also different, and the sorting of healthy mouse blood was examined experimentally. Whole blood diluted with heparin and physiological saline were injected into the specimen and buffer inlets respectively at flow rates of 7 μl/min and 190 μl/min, for 20 min. Figure 5 shows images of the specimen flow. The main stream of specimen remained narrow until it exited the DLD micro-posts array, and blood cells were mainly partitioned into the small cell outlet. After the separation via the chip, collected specimens were analyzed by FACS as shown by Fig. 6. About 2 ml of the specimen was collected from each outlet. About 109 cells were sorted with the DLD chip, and counting of the whole specimen by FACS was impractical in the measuring time and we could analyze only a part of specimen. To determine the portions of the analyzed specimen volume, known number of highly fluorescent beads, which can be distinguished from cells, were dispersed into the specimens collected after the sorting, and the beads were also counted simultaneously by FACS. Then, the separation ratio, R, was estimated as follows, R¼

CL CL þ CS

where, cL and cS respectively denote the concentration of cells segregated into large and small cell outlets. In all cases, a small number of cells were partitioned into the large cell outlet. According to the mouse supplier, the concentrations of red blood cell (RBC) and white blood cell (WBC) in the mice are about 9.5×109 cells/ml and 3.3×106 cells/ml, respectively. The WBC ratio in whole blood cells is 3.5×10−4. Small cell outlet specimens were reddish, and large cell outlet specimens were clear to the naked eye. No RBC

Biomed Microdevices (2015) 17: 59

Page 5 of 11 59

Fig. 3 Sorting of SP2/O cultured tumor cells. Collected specimens were centrifuged and were concentrated by removing top clear parts. Then, the concentrated specimens were agitated in tubes by pipetting, and 1% of total volume were observed on a hmoecytometer having 0.1mm depth by

an optical microscope. Detected cells are surrounded by circle with broke line. With Bmiddle^ and Bsmall^ configuration chips, no cell was observed in the small cell outlets, while several cells were observed in the small cell outlet with the Bbig^ configuration chip

was found during brief observation with an optical microscope. It was supposed that large or rigid white blood cells were segregated into the large cell outlets, and about 0.4% to 10% of WBC was segregated into the large cell outlet. Using the DLD device, enrichment of CTCs can be expected, but difficulty was also anticipated in separating CTCs by DLD size sorting alone.

number of tumor cells in this experiment. Tumor cell spiked blood and physiological saline were injected into the specimen and buffer inlets respectively at flow rates of 7 μl/min and 190 μl/min for 20 min. Recovery of tumor cells was attempted with two chips with Bmiddle^ and Bsmall^ configuration chips. Because leakage of tumor cells to the small cell outlet was expected, the Bbig^ configuration chip was not tested. After separation on the chip, the collected specimens were analyzed by FACS. Prior to the FACS analyses for the enrichment of spiked cells, the SP2/O-EGFP cells were analyzed alone by FACS for a control data. Figure 7 is the FACS result. The horizontal axis shows the intensity of GFP fluorescence and the vertical axis shows forward scattering (FSC), which reflects cell size. The fluorescence strength was a measure of the EGFP expression. It was assumed that SP2/O-EGFP showed strong fluorescence with large FSC, and red rectangle was determined to identify SP2/O-EGFP cells in the following experiments. Figure 8 shows the results of the tumor cell recovery experiments. Before the sorting, the concentration of tumor cells was 0.0076%, and the tumor cell concentration was successfully enriched in the large cell outlet by the sorting, while there are almost no cells that can be thought as SP2/O-EGFP in the small cell outlet. With the Bmiddle^ configuration chip, the tumor cells were enriched from 0.0076% to 88%, an enhancement factor of more than 10,000. Even after the sorting, a few cells were counted as tumor cells in the small cell outlet, but their fluorescence was not obviously large, and some benign cells may be counted as tumor cells. With the Bsmall^ configuration chip, enrichment of tumor cells was obtained, though the enrichment factor became lower than Bmiddle^ configuration one because a larger number of benign blood cells were segregated into large cell outlet as we expected from the result shown in Table 2. It was

3.3 Enrichment of spiked cultured tumor cells Enrichment of cultured tumor cells dispersed into whole blood was examined by the DLD devices. As a specimen, SP2/O-EGFP cells (2 × 105 cells/ml) and healthy mouse blood with heparin and physiological saline were mixed at a volume ratio 1:1. In order to easily detect leakage of tumor cells to small cell outlet, we spiked relatively large

Fig. 4 Optical microscopic observation of SP2/O cells. Cell diameters were measured with optical microscope images, and histogram of cell diameters was shown. Although the diameters varies largely, the minimum diameter was 9μm, and it was larger than majority of blood cells

59 Page 6 of 11

Biomed Microdevices (2015) 17: 59

Fig. 5 Specimen blood flow in the DLD device. Specimen (mouse blood diluted 3times with heparin and physiological saline) streamline was clearly observed, and majority of the specimen was segregated into the small cell outlet, and occasionally cells were segregated into the large cell outlet

reconfirmed that optimization of critical diameter Dc will be important in practical use. In order to ascertain the device ability, as a human cell, enrichment of Jurkat cells were also attempted in an identical way to SP2/0 cells. Figure 9 shows the result and it was shown that Jurkat cells were enriched to 78.2% in the large cell outlet. Because Jurkat cells were identified by their autofluorescence, the identification was not so precise. Though larger leakage into the small cell outlet was shown in the graph, the FACS graph of the large cell outlet resembles the FACS graph of

Jurkat cells, and we considered that good recovery of Jurkat cell was obtained.

Fig. 6 FACS results of healthy mouse blood. Horizontal axis shows strength of fluorescence around 519 nm excited by blue laser at 488 nm, and vertical axis shows forward scatter which reflects particle size. Specimens were diluted with buffer on the DLD chip, and collected specimens after sorting were concentrated by centrifugation. Then, FACS analysis was performed. In the FACS analysis, control of specimen portion to be processed was difficult, because the number of

cells in large cell outlet was small and cell loss in FACS analysis was not neglected, besides cell sedimentation happened in the tubes during the analysis. Therefore, discussion of separation ratio by simple FACS analysis was not available, but it was interesting that there was no clear threshold in FSC value between large and small cell outlets though FSC was supposed to reflect particle size. It was suggested that the cell deformability affects the DLD sorting and the FSC values

3.4 Sorting of tumor bearing mouse blood Real CTCs might be smaller and deformable than cultured tumor cells. Although enrichment of spiked cultured tumor cells was successfully verified and a high enrichment factor was obtained with the DLD device, most SP2/O-EGFP cells were obviously large compared to usual blood cells, and the

Biomed Microdevices (2015) 17: 59

Page 7 of 11 59

Fig. 7 FACS results of healthy blood and SP2/O-EGFP cells alone. By analyzing SP2/O-EGFP cells prior to each experiment, criterion of fluorescence and FSC to judge tumor cell or benign cell was determined as shown with red line rectangle. Due to autofluorescence of blood cells,

about 1% of healthy blood cells could be miscounted as tumor cells and the error is inevitably contained in following experiments. In the FACS graphs, horizontal and vertical axes show fluorescence and forward scatter, respectively

demonstration using SP2/O-EGFP cells is inadequate for practical application. Other culturable tumor cells are also large in general. Since identification of CTCs is the current challenge, there is no appropriate way to verify CTC separation even if cancer patient blood is available. To model CTCs, a tumor-bearing mouse was prepared. As described in the experimental section, the tumor was induced from inoculated SP2/O-EGFP. It was assumed that the cells in the tumor have EGFP expression, and identification of cells leaking from the tumor (CTC) was enabled by FACS.

In the preliminary experiments, blood collected from the tail frequently clogged the DLD posts, but smooth flow was observed with cardiac blood about 30 min sorting though some residue was observed along blood stream after the sorting. After the generated tumor mass became clearly observable by its appearance, usually 5 weeks after inoculation, 400 μl of cardiac blood was biopsied and collected into 100 μl heparin-containing tubes. No further dilution was applied to the blood specimen. The blood with heparin buffer and physiological saline were injected into the specimen and buffer inlets respectively at flow rates of 7 μl/min and 190 μl/min,

Fig. 8 Sorting of cultured tumor cell spiked blood. SP2/O-EGFP cells were spiked in healthy mouse blood and sorting experiments were performed with Bmiddle^ and Bsmall^ criterion chips. Based on the FACS analysis with SP2/O-EGFP cell alone, the red rectangle criterion was determined as shown in Fig. 7, and percentages of spiked tumor cells

were counted. With Bmiddle^ configuration chip, more than 10,000 times enrichment was obtained. Although some cells were judged as tumor cells in the small cell outlets, fluorescence is not obviously strong. In the FACS graphs, horizontal and vertical axes show fluorescence and forward scatter, respectively

59 Page 8 of 11 Table 2

Biomed Microdevices (2015) 17: 59

Healthy mouse blood separation

and segregated specimens were collected. Before and after the sorting, specimens were analyzed by FACS. Figure 10 shows the FACS results of the tumor-bearing mouse blood. As described above, cells showing strong green fluorescence with large FSC showing were counted as CTCs. 75 μl of the blood was used for each experiment. The number of blood cells in 75 μl of a specimen was estimated to be about 6×108, and the whole specimen could not be analyzed by FACS within a practical time period. 3×106 cells were analyzed, and no CTC was detected from a specimen before the sorting. The number of cells segregated into large cell outlets was countable, and whole specimens in the large cell outlet were analyzed by FACS. Table 3 summarizes the number of detected CTCs. With all size configurations, CTCs were successfully detected in the large cell outlet, and enrichment of CTC was verified. One cell was counted in the small cell outlet with Bmiddle^ configuration chip. Fluorescence was not strong for its FSC value, and it might be an error due to doubling or dust. We also noticed that number of detected CTCs was largest with Bsamll^ configuration chip, and CTC leakage due to

larger critical diameter was suspicious. However, total number of cells into large cell outlet became smaller with larger critical diameter and cell concentration became also smaller. Whole cell count without any loss was impossible with our present system due to cell residues in tubings. We considered that the variation of detected CTC number was mainly due to the imperfect count. To mitigate the imperfect counting, processing larger amount of specimens was desirable. But at this point, channel clogging often happened with bloods of tumor-bearing mice, and series results with three configuration chips (Bbig^, Bmiddle^ and Bsmall^) were barely obtainable. Figure 11 shows a sorting result of a blood from more seriously terminal mouse with the Bmiddle^ configuration chip. In this case, 7 CTCs were detected in 6.6×106 cells before the sorting. Though slight clogging was observed due to the serious mouse condition, sorting was completed. After the sorting, 41 CTCs were detected in large cell outlet from 4.1×105 cells, and about 100 times enrichment of CTC was achieved even with slight clogging. Only one cell was count as CTC in small cell outlet from 6.5×106 cells, but fluorescence was not strong neither. It can be said that enrichment of CTCs was successfully demonstrated in this model, and little CTC leakage into small cell outlet (benign cells) was observed. During the experiments, the number of cells in the large cell outlet was larger than expected. For healthy mouse blood, the percentages of cells segregated into the large cell outlet were measured and are listed in Table 2. From these results,

Fig. 9 Sorting of cultured human tumor cell spiked blood. As a human tumor cell example, Jurkat cells (human T lymphocyte cells) were spiked in healthy mouse blood. Sorting experiment was performed with Bmiddle^ criterion chip with almost identical procedure to Fig. 8

experiment. Similar result to SP2/O-EGFP cell (mouse myeloma cell) was obtained, and enrichment of spiked Jurkat cells was achieved. In the FACS graphs, horizontal and vertical axes show fluorescence and forward scatter, respectively

Size configuration

Big

Middle

Small

Separation ratio, R

1.36×10−6

1.25×10−5

4.04×10−5

Biomed Microdevices (2015) 17: 59

Page 9 of 11 59

Fig. 10 Sorting of tumor bearing mouse blood. Tumor bearing mouse was prepared by inoculating SP2/O-EGFP cells. Sarcoma like large tumor mass was obviously observed, and cardiac blood was biopsied. Specimen blood was diluted about 3 times with heparin and physiological saline solution, and sorting with 3 different size configuration chips were performed. Collected specimens were analyzed by FACS, and tumor cells

were identified by GFP fluorescence as shown by red rectangle. No tumor cell was detected in pre-sorting 1.5 million blood cells, and several tumor cells were detected in the large cell outlet after the sorting. In the FACS graphs, horizontal and vertical axes show fluorescence and forward scatter, respectively

820, 7500 and 24,000 cells are expected to be segregated into the large cell outlet with Bbig^, Bmiddle^, and Bsmall^ configuration chips, respectively. The observed number of cells in the large cell outlet was more than 10 times greater than the expectation as shown in Table 3. FACS analyses of healthy and tumor-bearing mouse blood are shown in Fig. 12. As shown by the red rectangle sectioning, the percentage of cells showing large FSC was elevated in the tumor-bearing mouse blood. We suggest that blood cells, especially WBCs, become larger, or large cells such as monocytes were increased due to cancer. In this experiment the mouse was terminal and its condition was quite bad; the increase of large cells might be significant. In general, WBC increases with injuries, and this large cell increase might be inevitable and make the separation difficult. We had presumed that CTC separation is not achieved by size sorting alone, but enrichment of real CTCs

was demonstrated successfully and no obvious CTC leakage into benign cells was observed. WBCs may be more flexible than CTCs (Alexander and Spriggs 1960; Vona et al. 2000), and if we could combine other properties such as cell deformability with size sorting in the DLD device somehow using fluidic shear force, better separation will be achieved. Biochemical affinity is also attractive property for the next separation stage. It was confirmed that DLD devices are promising for the CTC separation, but channel clogging is problematic to advantage the no mesh blinding theoretical feature of DLD devices. The clogging was significant at the entrance to the micro-posts zone and we did not have obvious residue in the downstream of the micro-posts zone. The factor of clogging must be identified to use the DLD devices practically. In this study, tumors were induced by myeloma cells, and the resulting cancers were sarcoma-like tumor masses. An epithelial tumor may be appropriate as a CTC mimic, but at this point only SP2/O cell lines were available to build the CTC model. Further studies using other cell lines are taking place.

Table 3

Enrichment of tumor cell in tumor bearing mouse blood

Before sorting

Size configuration Large cell outlet Small cell outlet

CTC non-CTC

0 2,950,029

CTC non-CTC CTC non-CTC

Big 12 24,124 0 1,514,037

Middle 15 77,746 1 1,504,804

Small 37 1,142,400 0 1,889,243

4 Conclusions On the basis that the CTCs tend to be large compared to most blood cells, enrichment or separation of CTCs has been attempted by various groups. We followed the idea and a

59 Page 10 of 11

Biomed Microdevices (2015) 17: 59

Fig. 11 An example sorting result of blood from more seriously terminal tumor bearing mouse. The mouse was prepared identically to the experiment in Fig. 10. Cardiac blood was biopsied, and the specimen blood was diluted about 3 times with heparin and physiological saline solution, and sorting was performed. Larger amount of specimen was processed by FACS, and several CTCs were detected from a specimen before the sorting. Enrichment of CTCs was reconfirmed, and one cell was observed in small cell outlet. In the FACS graphs, horizontal and vertical axes show fluorescence and forward scatter, respectively

DLD device was employed. Conventionally, spiked cultured tumor cells were used for verifying the performance. But cultured cell-line cells are obviously larger than most blood cells, and cultured cell line cells are not appropriate specimen for verification. In this study, the sorting performance of the DLD device for living cells was determined. The critical cell diameter was larger than the theoretical critical diameter assuming a rigid sphere. Real cells leaking from tumors into blood were then prepared using a tumor-bearing mouse, and enrichment of the tumor cells leaking in blood was attempted. The tumors

Fig. 12 Comparison of healthy and tumor bearing mouse bloods. Healthy mouse blood and tumor-bearing mouse blood were analyzed by FACS without the DLD sorting. Percentage of large FSC value (larger than 45 as shown by red rectangles) particles were higher with tumor bearing mouse, and increase of large cell due to cancer was suggested. In the FACS graphs, horizontal and vertical axes show fluorescence and forward scatter, respectively

were induced by cells with GFP expression, and the resulting tumor cells were assumed to have GFP expression and were identified by their fluorescence. Tumor cells was enriched to 0.05% from blood in which no tumor cells were found among 3 million cells, while no obvious leakage of CTCs was found. Even after enrichment, the concentration of the tumor cells was still small, and difficulties caused by patient condition could arise. A combination of methods using other properties as well as cell size will be needed for practical CTC separation.

Biomed Microdevices (2015) 17: 59 Acknowledgments Parent SP2/O cells were a generous gift from Prof. Takachika Azuma and Mr. Akikazu Mu- rakami (Tokyo University of Science). This study was supported by the Academic Frontier project for private universities, including a matching fund subsidy from MEXT, 2009–2014.

References R.F. Alexander, A.I. Spriggs, J. Clin. Pathol. 13, 414–424 (1960) W.J. Allard, J. Matera, M.C. Miller, M. Repollet, M.C. Connelly, C. Rao, A.G.J. Tibbe, J.W. Uhr, L.W.M.M. Terstappen, Clin. Cancer Res. 10, 6897–6904 (2004) S. Ariyasu, K. Hanaya, E. Watanabe, T. Suzuki, K. Horie, M. Hayase, R. Abe, S. Aoki, Langmuir 28, 13118–13126 (2012) J.P. Beech, S.H. Holm, K. Adolfsson, J.O. Tegenfeldt, Lab Chip 12, 1048–1051 (2012) A.S. Bhagat, H.W. Hou, L.D. Li, C.T. Lim, J. Han, Lab Chip 11, 1870– 1878 (2011) L. Chaffer, R.A. Weinberg, Science 331, 1559–1564 (2011) J.A. Davis, D.W. Inglis, K.J. Morton, D.A. Lawrence, L.R. Huang, S.Y. Chou, J.C. Sturm, R.H. Austin, Proc. Natl. Acad. Sci. U. S. A. 103, 14779–14784 (2006) Y. Gao, W. Li, D. Pappas, Analyst 138, 4714–4721 (2013) S.H. Holm, J.P. Beech, M.P. Barrett, J.O. Tegenfeldt, Lab Chip 11, 1326– 1332 (2011) M. Hosokawa, T. Hayata, Y. Fukuda, A. Arakaki, T. Yoshino, T. Tanaka, T. Matsunaga, Anal. Chem. 82, 6629–6635 (2010) L.R. Huang, E.C. Cox, R.H. Austin, J.C. Sturm, Science 304, 987–990 (2004) W. Inglis, J.A. Davis, R.H. Austin, J.C. Sturm, Lab Chip 6, 655–658 (2006)

Page 11 of 11 59 W. Inglis, M. Lord, R.E. Nordon, J. Micromech. Microeng. 21, 054024 (2011) C. Jin, S.M. McFaul, S.P. Duffy, X. Deng, P. Tavassoli, P.C. Black, H. Ma, Lab Chip 14, 32–44 (2014) Z. Liu, W. Zhang, F. Huang, H. Feng, W. Shu, X. Xu, Y. Chen, Biosens. Bioelectron. 47, 113–119 (2013a) Z. Liu, F. Huang, J. Du, W. Shu, H. Feng, X. Xu, Y. Chen, AIP Biomicrofluidics 7, 011801 (2013b) K. Loutherback, J. D’Silva, L. Liu, A. Wu, R.H. Austin, J.C. Sturm, AIP Adv. 2, 042107 (2012) S. Nagrath, L.V. Sequist, S. Maheswaran, D.W. Bell, D. Irimia, L. Ulkus, M.R. Smith, E.L. Kwak, S. Digumarthy, A. Muzikansky, P. Ryan, U.J. Balis, R.G. Tompkins, D.A. Haber, M. Toner, Nature 450, 1235–1239 (2007) T. Ohnaga, Y. Shimada, M. Moriyama, H. Kishi, T. Obata, K. Takata, T. Okumura, T. Nagata, A. Muraguchi, K. Tsukada, Biomedical Microdevices, 15, 611–616 (2013) R. Quek, D.V. Le, K.H. Chiam, Phys. Rev. E 83, 056301 (2011) S.H. Seal, Cancer 12, 590–595 (1959) W. Sheng, O.O. Ogunwobi, T. Chen, J. Zhang, T.J. George, C. Liu, Z.H. Fan, Lab Chip 14, 89–98 (2013) S.J. Tan, L. Yobas, G.Y.H. Lee, C.N. Ong, C.T. Lim, Biomed. Microdevices 11, 883–892 (2009) S.J. Tan, R.L. Lakshmi, P. Chen, W.-T. Lim, L. Yobas, C.T. Lim, Biosens. Bioelectron. 26, 1701–1705 (2010) G. Vona, A. Sabile, M. Louha, V. Sitruk, S. Romana, K. Schutze, F. Capron, D. Franco, M. Pazzagli, M. Vekemans, B. Lacour, C. Brechot, P. Paterlini-Brechot, Am. J. Pathol. 156, 57–63 (2000) S. Ye, X. Shao, Z. Yu, W. Yu, J. Fluid Mech. 743, 60–74 (2014) S. Zheng, H.K. Lin, B. Lu, A. Williams, R. Datar, R.J. Cote, Y.-C. Tai, Biomed. Microdevices 13, 203–213 (2011)