Cite this: Lab Chip, 2013, 13, 3728

Pinched-flow hydrodynamic stretching of single-cells3

Received 28th May 2013,Accepted 5th July 2013

DOI: 10.1039/c3lc50649e

www.rsc.org/loc

Jaideep S. Dudani,a Daniel R. Gossett,abd Henry T. K. Tseabd and Dino Di Carlo*abc

Reorganization of cytoskeletal networks, condensation and decondensation of chromatin, and other

whole cell structural changes often accompany changes in cell state and can reflect underlying disease

processes. As such, the observable mechanical properties, or mechanophenotype, which is closely linked to

intracellular architecture, can be a useful label-free biomarker of disease. In order to make use of this

biomarker, a tool to measure cell mechanical properties should accurately characterize clinical specimens

that consist of heterogeneous cell populations or contain small diseased subpopulations. Because of the

heterogeneity and potential for rare populations in clinical samples, single-cell, high-throughput assays are

ideally suited. Hydrodynamic stretching has recently emerged as a powerful method for carrying out

mechanical phenotyping. Importantly, this method operates independently of molecular probes, reducing

cost and sample preparation time, and yields information-rich signatures of cell populations through

significant image analysis automation, promoting more widespread adoption. In this work, we present an

alternative mode of hydrodynamic stretching where inertially-focused cells are squeezed in flow by

perpendicular high-speed pinch flows that are extracted from the single inputted cell suspension. The

pinched-flow stretching method reveals expected differences in cell deformability in two model systems.

Furthermore, hydraulic circuit design is used to tune stretching forces and carry out multiple stretching

modes (pinched-flow and extensional) in the same microfluidic channel with a single fluid input. The

ability to create a self-sheathing flow from a single input solution should have general utility for other

cytometry systems and the pinched-flow design enables an order of magnitude higher throughput

(65 000 cells s21) compared to our previously reported deformability cytometry method, which will be

especially useful for identification of rare cell populations in clinical body fluids in the future.

Introduction

The mechanical properties of cells have emerged as compel-ling biomarkers for cell state and disease.1–3 Such label-freebiomarkers have the potential to reduce cost and variability incell analysis4 by measuring intrinsic properties that do not, forexample, depend on the quantum yield or stoichiometry of alabel or affinity of an antibody. Changes in cytoskeletal andnuclear organization which accompany differentiation, activa-tion, pluripotency, or malignancy result in measurablechanges in cell mechanical properties.5,6 However, the cellularheterogeneity of tissues and biological fluids necessitateseither pre-selection of cells of interest by sample preparationmethods for time-intensive micromanipulation techniqueslike atomic force microscopy or micropipette aspiration7 orhigh-throughput single-cell analysis tools in order to make

confident classifications and diagnoses.8 As pre-selectionopens the door for losses of potentially rare cells of interestand phenotypic and mechanical shifts may occur due to theuse of pre-selection methods based on antibodies, fixatives, orlengthy sample preparation protocols, a high-throughputcontinuous assay with minimal sample handling or prepara-tion is preferred. Several technologies have been developed toassay the mechanical properties of suspended cells in flowincluding transit time analysis,9–14 optical stretching,15,16

electroporative flow cytometry,17 and hydrodynamic stretch-ing.18,19 The throughput of these methods varies across ordersof magnitude and each technique may measure mechanicalproperties of different compartments within the cell, as well asviscous or elastic components due to differences in the appliedstrain, strain rate and the shape of the stress field, and thusmay prove useful in different application niches.

In selecting a mechanical phenotyping technology for aclinical application several factors are important, includingaccounting for cell size differences, minimizing or accountingfor cell-surface interactions, and obtaining sufficient measure-ments in a reasonable time period. Cell size and mechanicalproperties should be collected independently or taken intoaccount in mechanical models to avoid misinterpreting shifts

aDepartment of Bioengineering, University of California Los Angeles, Los Angeles,

California 90095, USA. E-mail: dicarlo@seas.ucla.edu; Fax: +1 310 794-5956;

Tel: +1 310 983-3235bCalifornia NanoSystems Institute, Los Angeles, California 90095, USAcJonsson Comprehensive Cancer Center, Los Angeles, California 90095, USAdCytoVale Inc., South San Francisco, California 94080, USA

3 Electronic supplementary information (ESI) available. See DOI: 10.1039/c3lc50649e

Lab on a Chip

PAPER

3728 | Lab Chip, 2013, 13, 3728–3734 This journal is � The Royal Society of Chemistry 2013

Publ

ishe

d on

05

July

201

3. D

ownl

oade

d by

Uni

vers

idad

e Fe

dera

l do

Para

na o

n 25

/08/

2013

19:

48:0

1.

View Article OnlineView Journal | View Issue

in cell size as shifts in cell mechanical properties. Contact withcell surfaces, for example, in pore-based methods, should beminimized to prevent clogging or controlled for to preventbiasing of results based on differences in cell adhesiveness.And, the throughput should be sufficient to have statisticalconfidence in detection of the cells of interest.

We recently developed a method called DeformabilityCytometry (DC), which uses hydrodynamic forces to alignand apply stresses to cells in flow. Hydrodynamic forces havebeen employed in many microfluidic systems for manipulatingcells and particles, from vortices to control the rotation ofparticles20 or trap cells,21,22 to extensional flow stagnationpoints to hydrodynamically trap cells,23 to inertial focusingand separation.24,25 Uniquely, hydrodynamic forces, such aslift and drag, increase with flow velocity, thus they are usefulin high-speed flows. DC employs inertial lift forces25,26 touniformly position cells in flow and deliver them to the centerof an extensional flow where they are deformed. High-speedmicroscopic imaging and automated image analysis are usedto extract metrics of cell strain and apparent viscosity.18 Usingthis technology, we demonstrated the ability to diagnosemalignancy in pleural fluids19,27 and correlated increaseddeformability with pluripotency in mouse and humanembryonic stem cells.18,28,29

In previous work, we reported a throughput of approxi-mately 2000 cells s21, limited by the time cells reside at thestagnation point of the extensional flow (tens of microse-conds). In this paper, we report an alternative method ofhydrodynamic stretching which avoids this limitation andincreases throughput by over an order of magnitude bysqueezing cells with perpendicular cross-flows. This method,which we refer to as ‘hydropipetting’, may enable newapplications, traditionally in the domain of high-speed flowcytometers, such as the detection of hematopoietic stem cellsand residual leukemic cells.

Results and discussion

Design and operating principles of the hydropipetting method

Hydropipetting is based on the same three core operations asDC: (i) inertial focusing to position cells in flow, (ii)deformation of cells with hydrodynamic forces, and (iii)high-speed microscopic imaging and automated image analy-sis. Importantly, an upgraded frame rate (from 142 857 framess21 to 519 083 frames s21) and lower exposure time (from 1 msto 0.29 ms) have allowed a new hydrodynamic stretchinggeometry to be employed in which the deformation directioncoincides with the direction of motion without significantinterfering motion blur. Inertial focusing is used at twolocations in the microfluidic channel (Fig. 1a). First, cells insolution (the only input into the device) are concentrated tothe center streamlines of the flow using a combination ofinertial lift forces and secondary flows in a series ofasymmetric curving channels at Reynolds number (Re =rUDH/m) = 205, where r is fluid density of 1000 kg m23, U (U= Qin/A) is the average fluid velocity of 4.4 m s21, DH is the

hydraulic diameter calculated as 2WH/(W + H) for a channelwidth of 100 mm and a channel height of 30 mm, and m isdynamic viscosity of 0.001 Pa s, and a Dean number (De =Re(DH/2r)1/2) = 80, where r is the smaller radius of curvature of150 mm (Fig. 1b–c). Cell-free fluid is siphoned off from thesides of the channel that do not contain inertially focused cellsby splitting the microchannel (Fig. 1d, ESI3 Video 1). Thefraction of the flow siphoned is controlled by tuning the fluidicresistances of each branch. The fluidic resistance of the centralchannel containing cells is greater than that of the branches,thus a larger proportion of fluid is siphoned off. Followingsiphoning into the branch channels, a Reynolds number of40.4 (Uc = 1.0 m s21 through the center, and W = 60 mm)through the central channel is sufficient to achieve inertialfocusing (ESI3 Fig. 1a). The two branch fluid streams aresubsequently returned to apply a pinching flow to the centralchannel and hydrodynamic stress to the center-focused cells.Focused cells are then deformed by the rejoining cell-free‘‘sheath’’ fluid (Fig. 1e, ESI3 Video 2). The inertia of thereturning branch flow remains high (Re = 68.8 in each branch,Ub = 1.7 m s21 in each branch, W = 60 mm) by increasing flowvelocity through narrowing of the width of the branchesconducting the sheath fluid.

Cells in the suspension are likely deformed by a combina-tion of several hydrodynamic forces – inertial and viscous inorigin. First, the incident sheath fluid creates two transverseopposite directed pressure gradients across the cells in flow.We estimate the force accompanying this pressure gradient asa pressure drag, FD = 0.5rU2CDAp where U is average transversefluid velocity from the branch flow, CD is the drag coefficientof a sphere, and Ap is the cross-sectional area of a sphere. Thisforce can be calculated for standard operation at 800 mL min21

assuming a density of water of 1000 kg m23, velocity of 3.4 ms21 in the branches, a drag coefficient of 0.47, and a sphere ofdiameter 20 mm, to be approximately of 0.8 mN. This value issmaller than for the previously reported deformability cyto-metry method, such that smaller deformations should beexpected for the given flow conditions.18 Based on thisanalysis, the level of force would be expected to increase withincreasing flow velocity (or flow rate), such that deformabilitymeasures would not be directly comparable at different flowrates without using standard particles to normalize measure-ments to effective physical properties.

Second, viscous force from faster moving sheath fluid mayact on cells in the core fluid as the core fluid is accelerated,stretching cells in the direction of the flow. This force issimilar in nature to the forces present during micropipetteaspiration, however in the opposite frame of reference. For‘hydropipette aspiration’ (HA), instead of the physical micro-pipette remaining stationary and creating viscous drag thatstretches the front half of the cell as it is aspirated, the cellstarts out relatively stationary and is instead ‘aspirated’ by thefast moving fluid sheath walls. Here, the front and back of thecell experience different levels of viscous force, leading to a netstretching force in the cell’s frame of reference. We estimatethe viscous stress as t y c

.m, where c

.is the shear rate and m is

the dynamic viscosity. c.

is estimated as the difference betweenthe mean fluid velocity of the sheathing branch flows (3.4 ms21) and the initial cell velocity (1.01 m s21) divided by cell

This journal is � The Royal Society of Chemistry 2013 Lab Chip, 2013, 13, 3728–3734 | 3729

Lab on a Chip Paper

Publ

ishe

d on

05

July

201

3. D

ownl

oade

d by

Uni

vers

idad

e Fe

dera

l do

Para

na o

n 25

/08/

2013

19:

48:0

1.

View Article Online

diameter. Dynamic viscosity is 0.001 Pa s. The shear stressmultiplied by the surface area of a sphere results in a shearforce of approximately 1.5 nN. As cells travel through thedeformation region, they are forced through a 2D fluid orificeformed at this interface conceptually analogous to a 3Dmicropipette orifice (ESI3 Fig. 2). While smaller cells areexpected to experience a smaller drag and shear force, they onaverage deform to a higher aspect ratio (ESI3 Fig. 3). Analysis ofthree cell lines (Jurkat leukemia cells, MCF7 breast cancercells, and HeLa cervical cancer cells) revealed three separatedeformability profiles showing an inverse relationshipbetween size and deformability (ESI3 Fig. 3a,c). Smaller cellscould yield higher deformability for several potential reasons:small absolute changes in shape of a smaller cell results in ahigher aspect ratio compared to the same changes in a largercell, and larger cells occupy extended z-heights within thechannel including flow regions with reduced velocity andpressure (ESI3 Fig. 2).

Despite being limited by the pressures sustainable in PDMSreplica molded microfluidic devices, we achieved a throughputof 65 000 cells s21 in this device. We estimate the throughputof the system to be ultimately limited by the maximumconcentration (cmax) of cells that can be inertially focused (e.g.,cmax?Q) and the effect of motion blur and reduced number of

frames which becomes significant at above 10 m s21 flowvelocity. The maximum cell concentration can be estimated,starting from an observed length fraction. We found that themaximum length fraction and volume fraction (l, a measure ofthe fractional length of the microchannel occupied by cellswhen focused) to be approximately 0.5. The volume fraction ofcells in suspension (Vf) can be calculated as Vf = lpa2/6WH,where a is cell diameter. Using a mean cell diameter of y20mm, channel height of 30 mm, channel width of 100 mm, themaximum volume fraction is approximately 0.034. Using thesame mean cell diameter, the maximum concentration thatcan be inertially focused is 8.6 6 106 cells mL21. Therefore, forthe flow rate at which we operate (800 mL min21), themaximum throughput feasible is 115 000 cells s21.Increasing the cell velocity further will result in shorterdeformation times, which may not be resolvable with currentlyavailable high speed cameras, and higher shear stresses couldlead to negative effects on cell membrane integrity. Beyondincreased throughput, this method promises improved uni-formity in stretching time versus our extensional flow method(Relative Standard Deviation (RSD) = 5.3% for hydropipettingversus RSD = 37% for deformability cytometry; RSD = tStd Dev/tAve where tStd Dev is the standard deviation in residence time inthe stretching region and tAve is the average residence time).

Fig. 1 Method of hydropipetting. a. Top view of the microchannel. b. Cells are positioned in flow by a combination of drag from the secondary flow, FD, present incurving microchannels and inertial lift forces, FL,s and FL,w, present in high-speed confined flows. c. In a straight channel, cells are more tightly focused to the channelcenterline mitigating offsets due to secondary flows. d. The channel branches with focused cells traveling down the center branch and cell-free fluid siphonedthrough the side branches. Inset is a high-speed observation of the cells being concentrated to the center and the fluid siphoning (scale bar = 50 mm). e. The branchesjoin with the cell-free fluid incident on the focused cell suspension. Inset is high-speed observation (scale bar = 50 mm). f. A combination of hydrodynamic forces acton focused cells, deforming them. g. Cell diameter and deformability, which is defined as the aspect ratio of the maximally-deformed cell, are extracted from high-speed microscopic videos. High-speed observation of several maximally-deformed cells (scale bar = 10 mm)

3730 | Lab Chip, 2013, 13, 3728–3734 This journal is � The Royal Society of Chemistry 2013

Paper Lab on a Chip

Publ

ishe

d on

05

July

201

3. D

ownl

oade

d by

Uni

vers

idad

e Fe

dera

l do

Para

na o

n 25

/08/

2013

19:

48:0

1.

View Article Online

Hydropipetting reveals induced changes in cell structure

Using the system we were able to measure increased strains ina common model of cancer invasiveness, employed tocharacterize other flow-through methods15 and increasedstrains upon disruption of intermediate filaments. HeLacervical cancer cells were incubated with 16 mM 12-O-tetradecanoylphorbol-13-acetate (TPA) for 18 h in culture,which has been reported to increase cell invasiveness,metastatic potential, and deformability.15,30 This transforma-tion resulted in an increased deformability (as defined inFig. 1g) versus the control (Fig. 2a; increase in mediandeformability from 1.38 to 1.49; p , 0.001, Mann–Whitney–Wilcoxon Test). We also evaluated the effect of disruptingvimentin intermediate filaments (IF). Jurkat leukemia cellswere incubated with Calyculin A. Calyculin A is a phosphataseinhibitor, which disrupts IF structure as vimentin organizationis regulated by phosphorylation.31 The Calyculin A treatmentalso resulted in an increase in deformability (Fig. 2b, ESI3Video 3) versus the vehicle control, as expected (increase inmedian deformability from 1.33 to 1.96; p , 0.001, Mann–Whitney–Wilcoxon Test). Cells were injected into the device atQin = 800 mL min21.

Because of the reduced hydrodynamic forces, this techniqueexacts a smaller strain on cells as they pass through the

‘hydrolumen’ as compared to DC. For example, the deform-ability of Jurkat cells using conventional DC is 1.7 (versus 1.33here). While one of the benefits of deformability cytometry isthat large strains are easy to image and resolve, smaller strainsmay help reveal different biophysical properties, presumablymore focused on membrane stiffness and cytoplasmiccytoskeletal networks of cells and lead to new applications. Iflarger strains are required for a particular application, highervelocity branches or a higher ratio of branch-to-core flow canbe implemented.

Combined system for pinched-flow and extensional flowstretching enables more comprehensive mechanicalcharacterization

Hydropipetting can also be combined with DC in the samemicrochannel to reap the benefits of both systems (e.g.,measure cells with different strain rates or magnitudes ofstretching flows). We designed a channel that first deforms thecells using the hydropipetting method and then deforms themin an extensional flow field similar to that of DC after a briefperiod to allow some relaxation of the initial strain (Fig. 3a).Importantly, this microchannel was designed to operate with asingle inlet by use of hydraulic circuit analysis. Specifically, thefluidic resistance of the channels involved in hydropipettingwere designed to match that of the branches which join toform the second half of the DC stretching flow field (ESI3Fig. 1b,c). Similar to hydropipetting alone, cells are concen-trated to the center of the channel and cell-free fluid issiphoned. Cell-free fluid is siphoned off into four channels(Fig. 3b). Two channels rejoin the concentrated cell suspen-sion to form the hydropipetting sheath flows. The other twochannels join each other and then are directed against the cellsuspension forming the extensional flow, like in the DCjunction (Fig. 3c, ESI3 Video 4). The cell first deforms in thedirection of the flow as it passes through the hydrolumen andthen enters the extensional flow to be deformed perpendicu-larly to the direction of flow and exits through one of twooutlets. Unlike DC, cells do not reach the center of theextensional flow due to reduced inertia in this design (limitedagain by the maximum pressures tolerable in PDMS micro-channels). Despite this, the majority of cells of all sizesexperience a greater strain in the extensional flow compared tothe perpendicular squeezing flow (Fig. 3d). This design tostretch the same cell twice with different stresses increases thenumber of perturbations we can apply to each cell andtherefore increases the amount of data on cell mechanicalproperties that can be extracted. Importantly, the initial cellstrain induced by the pinch flow is smaller in magnitude thanthe second extensional-flow-induced strain, and conductingmeasurements in this order is likely to have less of an effect onthe final DC measurement, although the effect of pre-straincannot be entirely ruled out. Hydropipetting also yields lowerstrain rates than DC, which again can enable exploration ofcomplementary application areas to the higher strain ratesseen in DC (Fig. 3e). The average strain rate (SR) is calculatedas SR = (Dmax,t2 2 Dt1)/(t2 2 t1), where Dmax is the maximumdeformability and Dt1 is the deformability measurement of thecell several microseconds before the maximum deformability(Fig. 3g).

Fig. 2 Hydropipetting measures an increase in deformability in invasive cellmodels. a. HeLa cervical cancer cell deformability was measured usinghydropipetting (median deformability of 1.38). Treatment with TPA resulted in ahigher median deformability of 1.49. (p , 0.001, Mann–Whitney–WilcoxonTest) b. Jurkat leukemia cell deformability was also measured using hydro-pipetting (median = 1.33). Treatment with Calyculin A (which disruptsintermediate filaments) also resulted in a higher average deformability of 1.96(p , 0.001, Mann–Whitney–Wilcoxon Test). Note we plot the y-axis spanning alarger range for more deformable Jurkat cells.

This journal is � The Royal Society of Chemistry 2013 Lab Chip, 2013, 13, 3728–3734 | 3731

Lab on a Chip Paper

Publ

ishe

d on

05

July

201

3. D

ownl

oade

d by

Uni

vers

idad

e Fe

dera

l do

Para

na o

n 25

/08/

2013

19:

48:0

1.

View Article Online

Integrated sample preparation and pinched-flow stretching

The hydropipetting approach is compatible with inlinemicrofluidic rinsing (ESI3 Fig. 4a) of smaller blood cells andcell debris upstream. We diluted whole blood 156 inhypotonic red blood cell (RBC) lysis buffer for 10 min andspiked large breast cancer cells (MCF7) to yield a lysed bloodcell suspension. This suspension was introduced into acombined device that included an upstream rapid inertialsolution exchange (RInSE)32 region with a phosphate bufferedsaline (PBS) input as the exchange buffer (ESI3 Fig. 4b, ESIVideo 5). In the RInSE exchange channel, MCF7 cells migrateto the centerline faster than white blood cells and RBC debris,which are siphoned off to waste channels (ESI3 Fig. 4c, f). Thisleads to a large decrease in white blood cells reaching thehydropipetting junction and reduces the noise from analyzingthis population (ESI3 Fig. 4d, g, ESI3 Video 6). The MCF7 cellsmove towards the hydropipetting junction where a perpendi-cular flow of PBS that was introduced through a third inletsqueezes these cells. There was a large decrease in visiblebackground blood cells and debris (.70% rejected) as the cellsapproach the perpendicular squeezing flow where they aredeformed (ESI3 Fig. 4h). As both the sample preparationmethod and the mechanical measurement method areoperating continuously in line at rates of several thousandcells per second, recording can be triggered at any point

during operation. Note that the sample throughput in this casewould be limited by the operation of the upstream continuoussample preparation system (y1000 cells s21).

Materials and methods

Numerical simulation

A finite element method analysis solution to the full Navier–Stokes equation was performed using COMSOL Multiphysics(Comsol, Inc., Burlington, MA, USA) at the hydropipettingjunction. Velocity fields were extracted. All branch dimensionswere 60 mm wide, 200 mm long, and 30 mm high. Inletconditions were normal, inflow velocity of 1.93 m s21 for thecentral flow and 3.67 m s21 from the squeezing branch flows.The outlet boundary condition was set to pressure, no viscousstress at 0 Pa. No slip boundary conditions were used forremaining walls. The simulation was solved for 184 542degrees of freedom. The viscosity and density of the fluid arethat of water.

Device fabrication and operation

Silicon master molds of microchannels were made usingstandard photolithography techniques. Replica molding usingpolydimethylsiloxane (PDMS) (Sylgard 184 Silicone Elastomer

Fig. 3 Hydropipetting can be combined in-line with deformability cytometry. a. Top view of the microchannel for a combined hydropipetting and deformabilitycytometry system. b. The channel branches more than in the single hydropipetting device. c. The first pair of branches is routed to impact on the cell suspensionperpendicularly to perform hydropipetting. The second pair of channels meets the cell suspension by flowing towards the cells, forming an extensional flow toperform deformability cytometry. d. Overlaid images of a single Jurkat cell first being deformed in the squeezing flow, relaxing, and then deforming again in theextensional flow. e. The method provides additional parameters for each cell by deforming the Jurkat cells at different levels of strain. Cell deformability fromhydropipette aspiration (HA) is correlated to the deformability at the DC junction (black lines) for each individual cell. The majority of cells deform more greatly in theextensional flow. f. The strain rate for HA was also typically much lower than for DC. g. Definitions of deformability and strain rates. Strain rate is defined as thechange in deformability per unit time.

3732 | Lab Chip, 2013, 13, 3728–3734 This journal is � The Royal Society of Chemistry 2013

Paper Lab on a Chip

Publ

ishe

d on

05

July

201

3. D

ownl

oade

d by

Uni

vers

idad

e Fe

dera

l do

Para

na o

n 25

/08/

2013

19:

48:0

1.

View Article Online

Kit; Dow Corning Corp., Midland, MI, USA) was used to cast allmicrochannels. PDMS was exposed to air plasma and bondedto glass to create the final device. Cell suspensions of variousconcentrations (max , 4 000 000 cells mL21) were pumpedinto the device using a syringe pump (Harvard Apparatus,Holliston, MA, USA) through polyetheretherketone (PEEK)tubing. Devices were mounted onto a microscope andmonitored using a high-speed camera coupled to the micro-scope (Phantom v711, Vision Research Inc., Wayne, NJ, USA).For both the hydropipetting device and the combined device,the inlet flow rate was 800 mL min21. Cell deformations werecaptured using a 106 objective (NA = 0.45) at a resolution of208 6 32 pixels at 519 083 frames per second, and an exposuretime of 0.29 ms. A halogen lamp operating near its maximumvoltage was used to illuminate the channel. Kohler illumina-tion was used to ensure uniform light intensity.

Image analysis, data extraction, and plotting

A custom analysis script in Matlab v2009a (MathWorks) wasused to automate image analysis, data collection, postprocessing, and plotting. This analysis is similar to methodspreviously used to measure cell size and shape.18,33 Briefly,cells are tracked across the field-of-view where the shapeextraction utilizes a polar to Cartesian mapping of the cell at3-degree intervals. The unperturbed shape of the cell ismeasured 200 mm upstream of the hydropipetting lumen.The unperturbed shape is analyzed by averaging the threesmallest diameters measured upstream. The deformabilitymetric is the aspect ratios calculated for all frames as the celltransits down the channel. The reported deformability is themaximum aspect ratio (the ratio of cell length in the ydirection to the cell length in the x direction). We used thedscatter function from MATLAB file exchange for datavisualization.

HeLa treatment with TPA

12-O-Tetradecanoylphorbol-13-acetate (TPA; Cell SignalingTechnology, Denvers, MA, USA) dissolved in 30 mL dimethylsulfoxide (DMSO) was added to 3 mL of cell culture media to afinal concentration of 16 mM and added to HeLa cell culturefor 18 h. HeLa cells incubated with TPA and control cells weredetached and introduced into the device. For both cases, cellswere detached using 0.25% porcine trypsin for 3 min. Cellswere suspended in media and placed on a rocker forapproximately 15 min prior to use.

Jurkat treatment with Calyculin A

Jurkat leukemia cells were incubated with 50 nM Calyculin A(Cell Signaling Technology, Denvers, MA, USA) at 37 uC and5% CO2 in an incubator for 30 min and then in a water bathset at 37 uC for 30 min. These suspended cells were thenimmediately injected into the device to perform measure-ments. Control cells underwent the same procedure withoutthe addition of Calyculin A.

Combined hydropipetting and rapid inertial solutionexchange

Blood was drawn from consenting donors according to aprotocol approved by the UCLA Institutional Review Board.

100 mL of whole blood was diluted 156 in red blood cell lysisbuffer (ammonium chloride; Roche Diagnostics Corporation,Indianapolis, IN, USA). MCF7 breast cancer cells wereharvested from culture. A small volume of concentrated cellswas introduced to the diluted blood and the suspension wasintroduced at 70 mL min21 into the device. Simultaneously, awash solution of phosphate buffered saline (PBS) wasintroduced at 140 mL min21 and another flow of PBS wasintroduced to form the perpendicular squeezing flows at 150mL min21. The main channel for cell transfer was 1 cm long.After the Rapid Inertial Solution Exchange,32 cell deformationswere monitored using the same parameters described above.The device design is depicted in ESI3 Fig. 4a.

Combined hydropipetting and deformability cytometry

Operations occurred in exactly the same manner as describedabove in Device Operation. The device design in depicted inFig. 3a and ESI3 Fig. 1b. Inlet flow rate was the same as thehydropipetting device, but resistances of the channels weredesigned to enable a combined system (ESI3 Fig. 1c).

Conclusions

In this work, we present an alternative hydrodynamic stretch-ing mechanism, which provides a complementary set ofmeasurements (lower strain and strain rate) and capabilities(higher throughput) that can prove useful in emerging label-free cell analysis applications. Applications that might requirehigher throughput include screens for rare cells in concen-trated cellular samples, such as detection of rogue pluripotentstem cells for regenerative medicine therapies or detection ofcirculating cells of interest from blood or other body fluids. Inaddition, this mode of stretching can be used in conjunctionwith deformability cytometry to acquire additional informa-tion about each cell as well as increase uniformity of DCmeasurements by acting as a hydrodynamic flow focuser, or itcan be combined with upstream rapid solution exchange forincreased automation in diagnostic assays. This device is alsobetter suited to conduct combined fluorescence measure-ments and sorting as all cells travel in the same direction offlow. The stretching observed in the HA geometry also suggestsa possible effect of flow on the measurement of cell size (byscatter) in traditional flow cytometers where high-speed sheathflows also squeeze cells. Notably, we demonstrate that a singleinlet can be used to conduct the entire assay, through thedevelopment of a new self-sheathing approach that shouldfind use in a range of cell analysis assays beyond thosemeasuring mechanical properties. A single-input robustsystem with integrated sample preparation has potential totranslate to the research and clinical environment quicklyonce application areas are further demonstrated.

Acknowledgements

This work was partially supported by a Defense AdvancedResearch Projects Agency Young Faculty Award #N66001-11-1-

This journal is � The Royal Society of Chemistry 2013 Lab Chip, 2013, 13, 3728–3734 | 3733

Lab on a Chip Paper

Publ

ishe

d on

05

July

201

3. D

ownl

oade

d by

Uni

vers

idad

e Fe

dera

l do

Para

na o

n 25

/08/

2013

19:

48:0

1.

View Article Online

4125 (D.D.), a David and Lucile Packard Fellowship forScientists and Engineers (D.D.), a National ScienceFoundation CAREER Award #1150588 (D.D.), a GoldwaterScholarship (J.S.D.), and a grant from the Howard HughesMedical Institute to UCLA through the Precollege andUndergraduate Science Education Program (J.S.D.).

References

1 S. Suresh, J. Spatz, J. P. Mills, A. Micoulet, M. Dao, C.T. Lim, M. Beil and T. Seufferlein, Acta Biomater., 2005, 1,15–30.

2 D. Di Carlo, J. Lab. Autom., 2012, 17, 32–42.3 Y. Gambin and A. A. Deniz, Mol. BioSyst., 2010, 6,

1540–1547.4 D. R. Gossett, W. M. Weaver, A. J. Mach, S. C. Hur, H. T.

K. Tse, W. Lee, H. Amini and D. Di Carlo, Anal. Bioanal.Chem., 2010, 397, 3249–3267.

5 S. E. Cross, Y.-S. Jin, J. Rao and J. K. Gimzewski, Nat.Nanotechnol., 2007, 2, 780–783.

6 J. D. Pajerowski, K. N. Dahl, F. L. Zhong, P. J. Sammak andD. E. Discher, Proc. Natl. Acad. Sci. U. S. A., 2007, 104,15619–15624.

7 A. J. Mach, O. Adeyiga and D. Di Carlo, Lab Chip, 2012, 13,1011–1026.

8 D. Di Carlo and L. P. Lee, Anal. Chem., 2006, 78, 7918–7925.9 M. Abkarian, M. Faivre and H. A. Stone, Proc. Natl. Acad.

Sci. U. S. A., 2006, 103, 538–542.10 M. J. Rosenbluth, W. A. Lam and D. A. Fletcher, Lab Chip,

2008, 8, 1062–1070.11 S. Byun, S. Son, D. Amodei, N. Cermak, J. Shaw, J. H. Kang,

V. C. Hecht, M. M. Winslow, T. Jacks, P. Mallick and S.R. Manalis, Proc. Natl. Acad. Sci. U. S. A., 2013, 110,7580–7585.

12 S. Huang, A. Undisz, M. Diez-Silva, H. Bow, M. Dao andJ. Han, Integr. Biol., 2013, 5, 414.

13 Q. Guo, S. Park and H. Ma, Lab Chip, 2012, 12, 2687–2695.14 A. Adamo, A. Sharei, L. Adamo, B. Lee, S. Mao and K.

F. Jensen, Anal. Chem., 2012, 84, 6438–6443.15 J. Guck, S. Schinkinger, B. Lincoln, F. Wottawah, S. Ebert,

M. Romeyke, D. Lenz, H. M. Erickson, R. Ananthakrishnan,D. Mitchell, J. Kas, S. Ulvick and C. Bilby, Biophys. J., 2005,88, 3689–3698.

16 K. B. Roth, C. D. Eggleton, K. B. Neeves and D. W. M. Marr,Lab Chip, 2013, 13, 1571–1577.

17 N. Bao, Y. Zhan and C. Lu, Anal. Chem., 2008, 80,7714–7719.

18 D. R. Gossett, H. T. K. Tse, S. A. Lee, Y. Ying, A.G. Lindgren, O. O. Yang, J. Rao, A. T. Clark and D. DiCarlo, Proc. Natl. Acad. Sci. U. S. A., 2012, 109, 7630–7635.

19 H. T. K. Tse, D. R. Gossett, Y. S. Moon, M. Sohsman,Y. Ying, K. Mislick, R. P. Adams, J. Rao and D. Di Carlo,2013, submitted.

20 J. P. Shelby and D. T. Chiu, Lab Chip, 2004, 4, 168–170.21 S. C. Hur, A. J. Mach and D. Di Carlo, Biomicrofluidics, 2011,

5, 022206.22 A. J. Mach, J. H. Kim, A. Arshi, S. C. Hur and D. Di Carlo,

Lab Chip, 2011, 11, 2827–2834.23 M. Tanyeri, E. M. Johnson-Chavarria and C. M. Schroeder,

Appl. Phys. Lett., 2010, 96, 224101–224103.24 D. Di Carlo, D. Irimia, R. G. Tompkins and M. Toner, Proc.

Natl. Acad. Sci. U. S. A., 2007, 104, 18892–18897.25 D. Di Carlo, Lab Chip, 2009, 9, 3038–3046.26 D. R. Gossett and D. Di Carlo, Anal. Chem., 2009, 81,

8459–8465.27 H. T. K. Tse, D. R. Gossett, A. Lee, A. Ellison, Y. Ying,

R. Kulkarni, J. Rao and D. Di Carlo, in Proceedings of the15th International Conference on Miniaturized Systems forChemistry and Life Sciences, Seattle, WA, 2011, pp. 3–5.

28 M. Masaeli, D. R. Gossett, H. T. K. Tse, D. Gupta, A. T. Clarkand D. Di Carlo, 2013, manuscript in preparation.

29 M. Masaeli, H. T. Tse, D. R. Gossett, A. T. Clark and D. DiCarlo, in Proceedings of the 16th International Conference onMiniaturized Systems for Chemistry and Life Sciences,Okinawa, Japan, 2012, pp. 974–976.

30 D. R. Gossett, H. T. K. Tse, S. Lee, A. T. Clark and D. DiCarlo, in Proceedings of the 14th International Conference onMiniaturized Systems for Chemistry and Life Sciences,Groningen, The Netherlands, 2010, pp. 1382–1384.

31 M. J. Brown, J. A. Hallam, E. Colucci-Guyon and S. Shaw, J.Immunol., 2001, 166, 6640–6646.

32 D. R. Gossett, H. T. K. Tse, J. S. Dudani, K. Goda, T.A. Woods, S. W. Graves and D. Di Carlo, Small, 2012, 8,2757–2764.

33 H. T. K. Tse, P. Meng, D. R. Gossett, A. Irturk, R. Kastnerand D. Di Carlo, J. Lab. Autom., 2011, 16, 422–430.

3734 | Lab Chip, 2013, 13, 3728–3734 This journal is � The Royal Society of Chemistry 2013

Paper Lab on a Chip

Publ

ishe

d on

05

July

201

3. D

ownl

oade

d by

Uni

vers

idad

e Fe

dera

l do

Para

na o

n 25

/08/

2013

19:

48:0

1.

View Article Online