Enzyme-free Passage of Human Pluripotent Stem Cells by Controlling Divalent Cations

Enzymes used for passaging human pluripotent stem cells (hPSCs) digest cell surface proteins, resulting in cell damage. Moreover, cell dissociation using divalent cation-free solutions causes apoptosis. Here we report that Mg2+ and Ca2+ control cell-fibronectin and cell-cell binding of hPSCs, respectively, under feeder- and serum-free culture conditions without enzyme. The hPSCs were detached from fibronectin-, vitronectin- or laminin-coated dishes in low concentrations of Mg2+ and remained as large colonies in high concentrations of Ca2+. Using enzyme-free solutions containing Ca2+ without Mg2+, we successfully passaged hPSCs as large cell clumps that showed less damage than cells passaged using a divalent cation-free solution or dispase. Under the same conditions, the undifferentiated and early-differentiated cells could also be harvested as a cell sheet without being split off. Our enzyme-free passage of hPSCs under a serum- and feeder-free culture condition reduces cell damage and facilitates easier and safer cultures of hPSCs.

Post-hoc Tukey's multiple comparison revealed significant differences in remaining-cell ratio between different Mg 21 concentrations (the same Ca 21 data were put together to derive the numbers and bars in (c)) and in cell clump size between different Ca 21 concentrations (the same Ca 21 data were put together to derive the numbers and bars in (e)). Scale bars are 5 mm (b), 1 mm (d).  Fig.  2a-g). These detached cell clumps were then plated into fibronectincoated dishes and reattached as typical hPSC flat colonies on the next day ( Supplementary Fig. 2h-m). In addition, hiPSCs cultured on vitronectin and laminin, which are also used as a coating matrix for culturing hPSCs (Table 1), can be detached from the culture dishes by PBS ca/2 ( Supplementary Fig. 3). These results suggested that enzyme-free solution containing physiological concentration of Ca 21 , but no Mg 21 , could be useful for passaging hPSCs as large cell clumps.
Effects of dissociation and enzymatic digestion. We compared our enzyme-free passage method to both dissociation passaging in a divalent cation-free solution and enzymatic digestion passaging.
Dissociating hPSCs into single cells and replating at low density causes cell damage and death by apoptosis [4][5][6][7][8] . Thus, we hypothesized that detaching and dissociating hPSCs into larger clumps using PBS ca/2 could suppress apoptosis and support higher survival rates than detaching and dissociating these cells into smaller clumps using PBS without Ca 21 and without Mg 21 (PBS 2/2 ). To test this hypothesis, apoptosis was detected in the hiPSCs 253G1 and 201B7 using annexin V-FITC, which stains cell surface phosphatidylserine, four hours after detachment and dissociation in PBS 2/2 and PBS ca/2 followed by floating culture in ESF9a solution. Fluorescence microscopy showed many annexin V-FITC-positive single cells dissociated by PBS 2/2 , whereas annexin V-FITC-negative cells predominated in the large cell clumps dissociated by PBS 2/ca (253G1: Fig. 2a, 201B7: Supplementary Fig. 4a). Quantitative analysis using flow cytometry (FCM) revealed that fewer annexin V-FITC-positive cells were detached and dissociated by PBS ca/2 than by PBS 2/2 , and that addition of a ROCK inhibitor (RI) abolished these differences (253G1: Fig. 2b, 201B7: Supplementary Fig. 4b). RI blocks the dissociation-induced apoptosis of hPSCs 6,7 . To measure cell survival, hPSCs were detached and dissociated in PBS ca/2 or PBS 2/2 , plated  at low density (2 3 10 3 cells/cm 2 ), and then cultured for 3 days. The numbers of hiPSCs passaged in PBS ca/2 were higher than those passaged in PBS 2/2 (253G1: Fig. 2c, 201B7 & Tic: Supplementary Fig.  4c), suggesting that adding physiological concentration of Ca 21 to the dissociation solution increases cell survival rates by decreasing dissociation-induced apoptosis. It is also known that enzymatic digestion damages hPSCs 5,8 . We first used dispase, an enzyme often used to passage hPSCs under serum-and feeder-free conditions (Table 1) 20 . Because we routinely use 0.025-0.6 U/ml dispase (0.05-300 mg/ml), depending on the enzyme activity and on storage conditions 14 , excess dispase (1 U/ ml) was used to evaluate its damaging effect with the expectation that dispase dissociation of cell-cell binding would decrease the size of cell clumps, resulting in apoptosis. However, addition of 1 U/ml dispase in PBS 2/ca did not decrease hPSC clump size (253G1: Fig. 2d, 201B7: Supplementary Fig. 4d). Indeed, large clumps of annexin V-FITC-negative cells were also found when dispase was added to the PBS ca/2 (253G1: Fig. 2e, 201B7: Supplementary Fig. 4e), and quantitative analysis by FCM revealed that the relative percentages of annexin V-FITC-positive apoptotic cells were not changed by addition of dispase (253G1: Fig. 2f, 201B7: Supplementary Fig. 4f). The results were the same when 0.25% trypsin was added to PBS ca/2 , despite trypsin having more potent protease activity than dispase (Supplementary Fig. 5a-c, e-g). These findings together suggested that adding proteolytic enzyme to the PBS ca/2 dose not decrease the cell clump size and thus does not increase dissociation-induced apoptosis. Our results are consistent with a previous report that Ca 21 protects against trypsinization of cell-cell binding 10 . Because Ca 21 did not affect the cell-fibronectin binding during dissociation (Fig. 1bc, Supplementary Fig 1a), we next measured the re-attachment ability of hiPSCs to fibronectin-coated surfaces. To do this, hPSCs were incubated with dispase in PBS ca/2 , and replated in ESF9a medium with RI for counting the next day. The reattachment efficiency decreased with increasing concentrations of dispase, and the mean efficiency values at 1 U/ml dispase were smaller than those for PBS ca/2 alone (253G1: Fig. 2g, 201B7: Supplementary Fig. 4g); a similar result was attained when trypsin was added to PBS ca/2 ( Supplementary Fig. 5dh). These results suggested that addition of enzyme damages cells by suppressing cell-fibronectin rebinding rather than by increasing apoptosis.
These results showed that enzyme-free solution containing a physiological concentration of Ca 21 , without Mg 21 , enables passaging of hPSCs with less cell damage than found using either divalent-free solution or Ca 21 -containing solution with enzyme (dispase or trypsin).
Long-term culture of hPSCs with enzyme-free passaging. Next, we tried long-term culturing of hPSCs under enzyme-, serum-, and feeder-free culture conditions. Two hiPSC lines, 253G1 and 201B7, were successfully cultured for more than 10 passages in ESF9a medium on fibronectin-coated dishes by using the solution with Ca 21 and without Mg 21 (ESF-Fb-EzF condition), with both cell types expressing self-renewal markers ( Supplementary Fig. 6a-c, j-l). Immunocytochemistry of embryoid bodies derived from the two cell lines indicated that pluripotency was maintained ( Supplementary Fig.  6dm). Unexpectedly, karyotype abnormalities were found not only under the ESF-Fb-EzF condition ( Supplementary Fig. 6ir) but also in the sister cultures under the other conditions ( Supplementary Fig.  6hopq), suggesting that the abnormalities were induced before enzyme-, serum-, and feeder-free culture.
To confirm the karyotype normality, we newly performed longterm cultures using hiPSC 201B7 and Tic lines. The 201B7 cell line was pre-validated to ensure a normal karyotype, and then cultured under the ESE-Fb-EzF condition for more than 10 passages. The cells formed normal hiPSC colonies, which were tightly packed, and flat colonies consisting of cells with large nuclei and scant cytoplasm ( Fig. 3a) 1-3 and expressed four self-renewal markers, NANOG, OCT3/4, SSEA4 and TRA 1-60 but not an early differentiation marker, SSEA1 (Fig. 3b-f). The cells differentiated into derivatives of all three primary germ layers in vivo using teratoma formation (Fig. 3gi). The cells showed a normal karyotype (Fig. 3j). Karyotype after long-term culture was also normal in another hiPSC line, Tic ( Supplementary Fig. 7), confirming that karyotype remained stable during the enzyme-, serum-, and feeder-free culture. These results suggested that enzyme-free culture is a useful method for routine culturing of hPSCs.
Cell sheet harvesting. Finally, we tried cell sheet harvesting using our enzyme-free solution. Cell sheet harvesting using special equipment such as a temperature-responsive surface and magnet followed by transplantation is one of the most promising approaches for applying hPSCs to regenerative medicine 21,22 . However, in this study, simple incubation in PBS with Ca 21 followed by gentle pipetting enabled us to harvest the cells as 2-mm-diameter sheets without cells splitting off (Fig. 4a-e) and without specialized equipment. Similar results were obtained for early-differentiated cells induced by bone  morphogenetic protein 4 (BMP4) (Fig. 4f-i) and for hiPSC-derived hepatic progenitors (Fig 4j-n), suggesting that the PBS with Ca 21 could be used to routinely and simply harvest cells as a large sheet without special equipment.

Discussion
The present study showed that cell-fibronectin and cell-cell binding are controlled separately by Mg 21 and Ca 21 , respectively, in hPSC cultures. Using enzyme-free solutions containing Ca 21 without Mg 21 , we successfully passaged hPSCs cultured under serum-and feeder-free conditions as large cell clumps that showed less damage than those passaged in divalent cation-free solution or with dispase or trypsin. The cells were also harvested as a cell sheet without the need for splitting off.  (Fig 2d-g, and Supplementary Fig 4d-g and 5) to reduce the abundance of freefloating DNA fragments derived from damaged cells. Such addition of DNase might reduce cell-cell attachments arising from the free DNA fragments, and thereby also reduce cell clump size. However, even in the presence of DNase, the cell clumps dissociated by PBS Ca/2 without enzyme were significantly larger than those dissociated by PBS 2/2 without enzyme (Supplementary Fig 5ae).
Addition of enzyme increased the sizes of cell clumps in three of the four conditions in the presence of calcium (Fig 2d,  Supplementary Fig 4d and 5ae). A possible reason for this size increase tendency is enzymatic digestion of some cell-fibronectin attachment that enabled cell colonies to detach more easily from the dish. Consequently, large colonies may be harvested intact with less splitting of cell-cell binding by pipetting.
Commonly, hPSCs are passaged with enzyme and in medium containing physiological concentrations of Mg 21 and Ca 21 (Fig. 5 upper right) 1,2 . Single-cell culture methods such as clonal isolation are achieved by dissociating cells in solutions containing low Mg 21 and Ca 21 concentrations (Fig. 5 lower left) 4,6,8 . In the present study we showed that hPSC cell-cell binding can be disrupted with less cell detachment from the dish surface in a solution containing high Mg 21 , but low Ca 21 concentrations (Fig. 5 upper left), and that large cell clumps and sheets can then be harvested by dissociating in low Mg 21 and high Ca 21 solution (Fig. 5 lower right).
The serum-, feeder-, and enzyme-free composition described herein could provide practical culture methods for controlling For long-term culture under enzyme-, serum-and feeder-free condition (ESF-Fb-EzF condition), the cells were passaged with enzyme-free passage solution containing divalent cation-free DMEM-F12 medium (Supplementary table 1) supplemented with the same factors found in ESF9a medium. For subculturing, the cells were rinsed twice with PBS 2/2 and once with the enzyme-free passage solution, before being incubated in the same solution for more than 15 min at 37uC, and then triturated with a 1-ml micropipette tip. The cells were finally harvested by gentle centrifugation (1 min at 10 G) or stood for a few minutes before replating in hESF9a medium with 5 mM RI. Medium was changed daily.
Embryoid bodies formation. In vitro differentiation was induced by the formation of embryoid bodies as described previously 14 . Undifferentiated hiPSCs were cultured by floating in DMEM-F12 medium supplemented with 20% KSR, 0.1 mM 2mercaptoethanol, and MEM non-essential amino acids (Life Technologies). The floating embryoid bodies were then replated onto 1 mg/ml gelatin-coated dishes in DMEM with 10% FBS. The medium was changed every other day with the same floating culture solution.
Karyotype analysis. Metaphase spreads were prepared from cells treated with colcemid (KaryoMAX Colcemid, Gibco 15212-012, final concentration of 40 ng/ml, overnight treatment) or metaphase arresting solution (Genial Genetic Solutions Ltd., Cheshire, UK). We performed a standard G-banding or multicolour fluorescence in situ hybridization (FISH) karyotypic analysis on at least 20 metaphase spreads for each population. For FISH karyotype analysis, 24XCyte Human Multicolor FISH Probe Kit (MetaSystems GmbH Altlussheim, Germany) were used.
Cell detachment and dissociation assay. The dose response size and removal ratio of hPSCs cultured on divalent cations using 24-well plates were measured as follows. The cells cultured in ESF9a on fibronectin-coated dishes were detached and dissociated into cells clump using 0.2-0.5 U/ml dispase, and then were plated in ESF9a medium on 24-well plates coated with 2 mg/cm 2 fibronectin (Wako) at 37uC for more than 1 hour. At 4 or 5 days after plating, the attached cells were stained with 1 mM calcein-AM (Dojindo, Kamimashiki, Kumamoto, Japan), a fluorescent living cell dye, for 20 min at 37uC, and imaged as the control state. Then the cells were rinsed once with PBS 2/2 , rinsed again with PBS containing various concentration of Ca 21 and Mg 21 , incubated in the same PBS for a further 15 min at 37uC, and then triturated 5 times with a 1-ml micropipette tip. For enzymatic digestion experiments, dispase or trypsin was added after 12 min incubation in PBS and left for 3 min. The cells were then triturated 5 times with a 1-ml micropipette tip in the presence of 1 mg/ml DNase I (Roche, Basel, Switzerland), 250 mg/ml trypsin inhibitor (Life Technologies), and 1 mg/ml BSA (sigma), and then 10X volumes of PBS ca/2 were added before spinning down the cells. The detached cells were then transferred to another plate, and the remaining cells were imaged for green fluorescence to estimate detachment efficiency. The detached cell clumps were placed between cover slips and cell clump size was estimated based on fluorescent signal using Image J software (NIH, Bethesda, MD, USA). To estimate the cell clump sizes, randomly picked cell clumps for each condition in a test were analyzed with Excel software (Microsoft, Redmond, WA, USA). The cell clump size was converted from area (mm 2 ) into the number of cells by using the area of single cells, which was estimated to be 240 6 86 mm 2 (mean 6 SD, n 5 107) in a separate experiment.
Teratoma formation. Teratomas were generated in severe combined immunodeficient (SCID) mice from 201B7 hiPSCs grown under ESF-Fb-EzF conditions for more than 10 passages. The cells harvested by dispase were resuspended in DMEM supplemented with RI (10 mM). The cells from a confluent single well in a 6-well plate were injected into the thigh muscle of a SCID (C.B-17/lcrscid/scidJcl) mouse (CLEA Japan, Tokyo, Japan). Nine weeks after injection, tumors were dissected, weighed, and then fixed with 10% formaldehyde Neutral Buffer Solution (Nacalai Tesque, Kyoto, Japan). Paraffin-embedded tissue was sectioned and stained with hematoxylin and eosin (HE). All animal experiments were conducted in accordance with the guidelines for animal experiments of the National Institute of Biomedical Innovation, Osaka, Japan.
Alkaline phosphatase (ALP) staining, immunocytochemistry. The hPSCs were stained with an Alkaline Phosphatase Staining Kit II according to the manual (StemGen, Cambridge, MA, USA). Briefly, the cells were rinsed twice with PBS 1/1 and fixed with Fix Solution from the kit at room temperature for 4 minutes. The fixed cells were rinsed with PBS containing with 0.05% (v/v) Tween20 and incubated in AP Substrate Solution at room temperature for 20 to 30 minutes. Then the cells were rinsed with PBS 1/1 and photographed. Immunocytochemistry was performed as described previously 14,17 . Briefly, hiPSCs were fixed in 4% formaldehyde with 0.5 mM MgCl 2 and 0.5 mM CaCl 2 . Then the cells were permeabilized and blocked with PBS containing 0.1-0.2% Triton X-100, 10 mg/ml BSA, 0.5 mM MgCl 2 , and 0.5 mM CaCl 2 , and then reacted with primary antibodies in the solution. The primary antibody binding was visualized using secondary antibodies. Antibody information is listed in Supplementary Table 2. Nuclei were stained with 1 mM DAPI (Wako). Micrographs were taken using a BZ-8100 fluorescence microscope (Keyence, Osaka, Japan).
Flow cytometry (FCM). FCM analysis was performed as described previously 14,17 . All cells were removed from culture dishes using 0.02% (w/v) EDTA-4Na in PBS 2/2 and then fixed in 4% formaldehyde. The fixed cells were permeabilized and blocked with PBS 2/2 containing 0.1-0.2% Triton X-100 and 10 mg/ml BSA, and then reacted with primary antibodies in the solution. The primary antibody binding was visualized with secondary antibodies. Antibody information is listed in Supplementary Table 2. A cell sorter (JSAN, Bay Bioscience Co., Ltd., Hogo, Japan) was used for data acquisition.
Apoptosis. An annexin V-FITC apoptosis detection Kit was used to detect cell surface phosphatidylserine (BioVision, Milpitas, CA, USA). Cells were floating cultured for four hours in ESF9a solution following detachment and dissociation. For FCM analysis, the cells were re-dissociated by incubating in 0.02% EDTA solution followed by trituration using a 1-mL pipette tip. The living cells were then stained with FITC conjugate annexin V (15100) and 50 mg/ml propidium iodide in binding buffer. The living cells nuclear were stained with 1 mg/ml Hoechst 33342 (Dojindo, Osaka, Japan).
Spot sheet formation. Silicone rubber masks made of polydimethylsiloxane (PDMS, Sylgard 184, 1051 mix; Dow Corning) were perforated with 2-mm-diameter holes using a hole-punch. The bacterial culture dish (non-cell-attachment-treated dishes, Iwaki) with PDMS masking were treated for 60 seconds by air plasma to make hydrophilic spots (YHS-R, SAKIGAKE-Semiconductor Co., Ltd), then coated with 2 mm/cm 2 fibronectin for more than 1 hour at 37uC to make fibronectin spots. After rinsing twice with PBS 2/2 , the PDMS mask was removed and the dish was sterilized under a UV lamp. hPSCs (253G1) cultured under the ESF-Fb-Dsp condition or hiPSC Tic-derived hepatic progenitors were dissociated in calcium-and magnesium-free solution, and then plated in ESF9a solution with 5 mg/ml RI or in CDM medium with 50 ng/ml FGF10 with RI.
Early differentiation was induced by 2 days cultivation in the ESF6 medium with 2 ng/ml recombinant human BMP4 (314-BP, R&D Systems, Inc, Minneapolis, MN, USA) as described previously 17 . Hepatic progenitors were differentiated based on the previously reported protocol 25 . Briefly, the hiPSC Tic line was passaged and grown for 2 days in CDM medium 26 . hiPSCs were then grown for 3 days in CDM/PVA medium 26 with 100 ng/ml activin, 20 ng/ml basic FGF, 10 ng/ml BMP4, and 10 mM LY294003 (9901, Cell Signaling Technology, Beverly, MA, USA), followed by 3 days