Supplementary Figure 1 Real-time deformability cytometry: contour detection and theoretical modeling. (a) Image of cell deformed in constriction; contour (red) according to image analysis algorithm. Scale bar, 5 µm. (b) Characteristic deformed bullet shape of an elastic sphere after application of the pressure and shear stress distributions shown in Figure 1d. Color code indicates displacement relative to non-deformed sphere.
Supplementary Figure 2 RT-DC sensitivity to cytoskeletal drugs. In addition to the basic result of a dose response curve for cytochalasin D (cytod) at a flow rate of 0.04 µl/s presented in the main paper (cf. Fig 2a), we carried out additional experiments presented here. Each scatter plot contains more than 2,000 cells measured. (a) RT- DC scatter plots of HL60 control cells (left column) and cells treated with 0.1 µm cytod (middle column) for three different flow rates of 0.04, 0.08 and 0.12 µl/s in 20 µm 20 µm channels. Right column shows 50%-density contour plots of the data shown in the two columns to the left plotted together for easier comparison. Thin lines are iso-elasticity lines (cf. Fig. 1f). As expected, cells treated with cytod showed increased deformation, and a shift to lower elasticity bands, compared to controls at all three flow rates. As an additional basic performance check, both controls and treated cells each showed a consistent increase in deformation at increasing flow rates, which induce larger stresses. For all conditions, spherical shape of the cells before entering the channel was verified explicitly (Supplementary Fig. 7a,b). (b) The softening of HL60 cells with cytod treatment is also in agreement with earlier reports of this effect on the same cell line, and has been explicitly confirmed again here using an optical stretcher in creep compliance measurements. See Ekpenyong et al. for experimental details. Shown are averaged single-cell creep-compliance curves of untreated (N = 89, black curve) and cytod-treated (N = 68, blue curve) HL60 cells. The error bars represent the standard error of the mean. Comparing the compliance of both populations shows a softening after cytod treatment. (c) Dose-response curve showing the increasing ratio of cytod-treated cell deformation relative to control cells as a function of different concentrations and for three different flow rates. All experiments were performed in 20 µm 20 µm channels. Data points show the mean of three independent measurement sets, errors are standard error of mean. The data for the flow rate of 0.04 µl/s are the raw data presented in Fig. 2a. Of note, the relative increase in deformation of treated cells compared to controls is independent of the flow rate. Inset shows the effect of DMSO alone at different concentrations for a flow rate of 0.12 µl/s. The short measurement time per experimental condition of about 1 min enables the relatively simple acquisition of such dose-response curves (here a total of 60 experiments, each on a new chip, 150,000 cells in total), which would be tedious and time-consuming with previous low-throughput methods. However, the basic trend of increasing deformation with increasing concentration of cytod has previously been reported with other techniques and on other cell types. Similar dose-response curves were obtained for several other drugs affecting actin (jasplakinolide, blebbistatin) and microtubules (nocodazole, paclitaxel), which all showed the expected results (data not shown). 1. S. C. Erzurum, M. L. Kus, C. Bohse et al., Am. J. Respir. Cell Mol. Biol. 5 (3), 230 (1991). 2. A. E. Ekpenyong, G. Whyte, K. Chalut et al., PLoS One 7 (9), e45237 (2012).
3. C. Rotsch and M. Radmacher, Biophys. J. 78 (1), 520 (2000). 4. S. Kasas, X. Wang, H. Hirling et al., Cell Motil. Cytoskeleton 62 (2), 124 (2005).
Supplementary Figure 3 Confirmation of cytochalasin D effect on filamentous actin. Confocal fluorescence images of a representative control (left) and cytod-treated (right) cell stained for F-actin (red) and DNA (cyan). Top row shows x-y slices (top view), bottom row x-z reconstructions (side view; white line indicates microscope slide surface). Treated cells have a less pronounced actin cortex and are flatter after centrifugation onto the microscope slide for imaging. Scale bar is 10 µm. The greater deformation of cytod treated cells (cf. Fig. 2a, Supplementary Fig. 1) is thus likely caused by the reduction of actin cortex thickness due to the known F-actin depolymerizing effect of cytod. 5. J. A. Cooper, The Journal of Cell Biology 105 (4), 1473 (1987).
Supplementary Figure 4 Progression through the cell cycle. Change in deformation during progressing through the cell cycle. Each scatter plot contains more than 2,000 cells measured. (a) RT- DC scatter plots of HL60 cells chemically synchronized in G1 (gap phase 1), S (synthesis phase), G2 (gap phase 2) and M (mitosis) phases of the cell cycle. These are the raw data from the main text (cf. Fig. 2b). Thin lines are iso-elasticity lines (cf. Fig. 1f). Cells were measured in a 20 µm 20 µm channel with a flow rate of 0.04 µl/s. From G1 to S phase cells showed both an increase in size and stiffness (as judged by the comparison with the iso-elasticity lines) at constant observed deformation; from S to G2 phase cells mainly increased in size with approximately constant stiffness (comparison with iso-elasticity lines) albeit an increase in observed deformation. However, with the transition from G2 into M phase cells stiffened significantly, based on both comparison with iso-elasticity lines and the resulting deformation, while their size remained identical. Please note that similar G2 results were obtained during a release from double thymidine block after 4 5 h instead of CDK1 inhibition (data not shown). (b) Fluorescence confocal microscopy images of G2 cells (top) and M cells (bottom) stained F-actin (red) and DNA (cyan) show that the increase in cell stiffness during mitosis is at least partly caused by an increase in the extent of the actin cortex. Scale bar is 10 µm. (c) Comparison of RT-DC data of HL-60 cells synchronized in M phase with those incubated with 1 µm cytod shows a shift of the distribution towards iso-elasticity lines of lower stiffness and greater deformation at approximately the same size. Cells were measured in a 20 µm 20 µm channel with a flow rate of 0.04 µl/s. (d) 50%-density contour plots of the cell populations in M phase (blue) and after treatment with cytod (green) as shown in (c) for direct comparison. As described in the main text, this points to a contribution of F-actin in the stiffening of M cells. Such an increase in cell stiffness during mitosis has previously been reported from experiments with optical stretcher and atomic force microscopy. 6. H. K. Matthews, U. Delabre, J. L. Rohn et al., Dev. Cell 23 (2), 371 (2012). 7. P. Kunda, A. E. Pelling, T. Liu et al., Curr. Biol. 18 (2), 91 (2008). 8. M. P. Stewart, J. Helenius, Y. Toyoda et al., Nature 469 (7329), 226 (2011).
Supplementary Figure 5 Mechanical phenotyping of blood precursor cells and their differentiated progeny. Comparison of HL60-derived and bone marrow-derived CD34+ (BM-CD34+) cells. (a) RT-DC scatter plots of HL60 cells differentiated into granulocytes (HL60-gran), monocytes (HL60-mono) and macrophages (HL60-mac) measured at a flow rate of 0.04 µl/s in a 20 μm 20 μm channel. Thin lines are iso-elasticity lines (cf. Fig. 1f). An increase in deformability of HL60 cells differentiated to granulocytes had also been found with other techniques. (b) RT-DC scatter plots of BM-CD34+ cells differentiated in vitro into granulocytes (HSC-gran), monocytes (HSC-mono) and macrophages (HSC-mac) measured at a flow rate of 0.16 µl/s in a 30 µm x 30 µm channel (HSC-macs are larger than 20 µm in diameter, which requires a wider channel). Thin lines are iso-elasticity lines (cf. Fig. 1f). (c) FACS scatter plot of forward- and sidescattering, respectively indicative of size and granularity, of HSC-mac (red), HSC-mono (green) and HSCgran cells (blue).
Supplementary Figure 6 FACS analysis of whole blood. FACS of whole blood and of whole blood after red blood cell (RBC) depletion. (a) Forward vs. side scatter plot of whole blood showing a rather homogeneous distribution with no distinct peaks due to red blood cell scattering. (b) Depletion of red blood cells reveals three distinct subpopulations, which can be identified as lymphocytes (47.6 %, red solid line), monocytes (3.7 %, dashed line) and granulocytes (38.7 %, dotted line) based on gating.
Supplementary Figure 7 Mechanical phenotyping of cell types in whole blood. In whole blood, in addition to white blood cells, also platelets and erythrocytes (red blood cells, RBC) can be found, with the latter being the largest fraction by far (RBC/other blood cells ~ 170:1). This poses the challenge whether any of the other cell types can be seen against the very large background of RBCs. To identify the remaining cells better, we reduced the number of RBCs by sedimentation in a dextran solution (see Online Methods for details) and measured the remaining sample. All samples shown in this figure were measured at a flow rate of 0.04 µl/s in a 20 µm 20 µm channel. (a) RT-DC scatter plot of whole blood with most of the RBCs sedimented out. Here, the three remaining peaks are much more prominent (cf. Fig. 3d and Supplementary Fig. 6f,g). One of the remaining peaks can clearly be identified as platelets, based on their known very small size and separate RT-DC analysis (data not shown). Separation of (b) RBCs, (c) peripheral blood mono-nucleated cells (PBMCs; containing monocytes, lymphocytes, immature leukocytes) and (d) granulocytes (gran) by gradient density centrifugation, and subsequent RT-DC analysis clearly identified the remaining peaks. The correct relative numbers of these cells were confirmed with FACS analysis (see Supplementary Fig. 5b). (e) Isolated granulocytes were stimulated with phorbol-12-myristate-13-acetate (PMA), which resulted in a clear shift in their mechanical fingerprint. (f,g) RT-DC scatter plots of whole blood (diluted 1:50 in PBS containing 0.5 % methylcellulose) from two different donors (in addition to the one in Fig. 3d) showing very little inter-donor variability of the relative positions of the four peaks.
Supplementary Figure 8 Verification of spherical cell shape before the cells enter the narrow channel. The plots summarize RT-DC data of (a) HL60 cells, (b) HL60 cells treated with 0.1 µm cytochalasin D, and (c) CD43 + cells derived from peripheral blood (PB-CD34 + ) cells inside the reservoir before entering the channel where the measurements were carried out. Flow rates were 0.04 µl/s (HL60) and 0.08 µl/s (PB-CD34 + ), respectively. Fitting log-normal distributions to the 1-dimensional projection of the deformation in all populations reveals a mode of 0.01 close to the theoretical expected value of 0 for an ideal sphere. Results for other cell types and conditions appearing in the main text were similar (data not shown).
Supplementary Figure 9 Viability and rate of growth of HL60 cells after RT-DC analysis. Experiments were carried out using a 20 µm x 20 µm channel and a flow rate of 0.12 µl/s. (a) After cell recovery the viability of post RT- DC cells was determined using a standard Annexin V-FITC/PI FACS assay and compared to a control sample (b) (see Online Methods). The FACS scatter plots show that the relative number of necrotic / apoptotic cells in the control sample (3.07 %) is very close to the value after RT-DC measurement (3.89 %). The viability test in (a) and (b) shows results from one measurement, but was repeated three times (viability post RT-DC: 94 ± 1 % and viability control 97 ± 1%). (c) In parallel the rate of growth of post RT-DC HL60 cells was observed for 5 days and compared to a control sample. Viability tests using Trypan Blue (see Online Methods) after 6, 24, 48, 72, 96 and 120 hours reveal a viability of 98% (control) and 97% (post RT-DC).