ISSN 2040-3364

www.rsc.org/nanoscale Volume 4 | Number 7 | 7 April 2012 | Pages 2169–2510

COVER ARTICLELee et al.A quartz nanopillar hemocytometer for high-yield separation and counting of CD4+ T lymphocytes

FEATURE ARTICLEVirgili et al.Confocal ultrafast pump–probe spectroscopy: a new technique to explore nanoscale composites

Series Editors:

Harry Kroto FRS,

Florida State University, USAFlorida State University, USA

Paul O’Brien

Manchester University, UKManchester University, UK

Ralph Nuzzo

University of Illinois at Urbana-University of Illinois at Urbana-

Champaign, USA Champaign, USA

Joao Rocha

University of Aveiro, PortugalUniversity of Aveiro, Portugal

Xiaogang Liu

National University of Singapore, National University of Singapore,

ChinaChina

This book series covers the wide

ranging areas of Nanoscience

and Nanotechnology. In

particular, the series provides

a comprehensive source

of information on research

associated with nanostructured

materials and miniaturised

lab on a chip technologies.

Titles are intended to provide

an accessible reference for

professionals and researchers in

academia and industry and all

are fully referenced to primary

literature.

RSC Nanoscience and

Nanotechnology Series

Package

Includes 10 leading

series titles

Comprehensive source

of information

Cutting-edge developments

in research

SAVE 35% on the individual

purchase price of these leading

series titles – purchase as one

complete package!

NEW Second Edition

Fullerenes

Fernando Langa De La Puente, Universida de Castilla-La Mancha, SpainJean-François Nierengarten, Université de Strasbourg, FranceISBN: 9781849731362 | Price: £144.99

This fully revised new edition is an up-to-date survey of the research areas of carbon nanotubes, inorganic nanotubes and nanowires including synthesis, characterisation, properties and applications. An ideal resource both for graduates needing an introduction to the field of nanomaterials as well as for professionals and researchers in academia and industry.

Nanoparticles in Anti-Microbial Materials

Fiona Regan, James Chapman and Timothy SullivanDublin City University, Ireland ISBN: 9781849731591 | Price: £132.99

This book documents the most up-to-date research on the area of nanoparticles showing anti-microbial activity and discusses their preparation and characterisation. With its user-friendly approach to applications, this book is an excellent reference for practical use in the lab.and technology, where phage-derived nanomaterials can be most beneficial.

Biological Interactions with Surface Charge in Biomaterials

Tofall Syed, Univesity of Limerick, IrelandISBN: 9781849731850 | Price: £121.99

This book is the first to comprehensively address the complex phenomenon of biological interactions with the surface charge of biomaterials.

Nanoporous Gold

Arne Wittstock, Marcus Bäumer, University of Bremen, GermanyJürgen Biener, Lawrence Livermore National Laboratory, USAJonah Erlebacher, Johns Hopkins University, USAISBN: 9781849733748 | Price: £132.99

Provides a broad, multi-disciplinary platform to discuss and learn more about the fascinating properties of nanoporous gold from an inter-disciplinary perspective.

To order any of these titles please email booksales@rsc.org or visit the website!

www.rsc.org/nanoscienceRegistered Charity Number 207890

New for 2012 Manipulation of Nanoscale MaterialsKatsuhiko Ariga, NIMS, JapanISBN 9781849734158 | Price £153.99

Fully referenced to the primary literature, this title presents an excellent source of information, and inspiration, to the reader and should appeal to experienced materials scientists, nanotechnologists and postgraduate students.

Towards Efficient Designing of Safe NanomaterialsJerzy Leszczynski, Jackson State University, USA | Tomasz Puzyn, University of Gdansk, PolandISBN 9781849734530 | Price £153.99

This book is the first to provide a comprehensive review of recent progress and challenges in the risk assessment of nanomaterials by empirical and computational techniques.

The RSC Nanoscience & Nanotechnology Series

Publ

ishe

d on

04

Janu

ary

2012

. Dow

nloa

ded

by Y

ale

Uni

vers

ity L

ibra

ry o

n 01

/07/

2013

18:

35:3

5.

View Article Online / Journal Homepage / Table of Contents for this issue

Dynamic Article LinksC<Nanoscale

Cite this: Nanoscale, 2012, 4, 2500

www.rsc.org/nanoscale PAPER

Publ

ishe

d on

04

Janu

ary

2012

. Dow

nloa

ded

by Y

ale

Uni

vers

ity L

ibra

ry o

n 01

/07/

2013

18:

35:3

5.

View Article Online

A quartz nanopillar hemocytometer for high-yield separation and counting ofCD4+ T lymphocytes†

Dong-Joo Kim,a Jin-Kyeong Seol,a Yu Wu,b Seungmuk Ji,c Gil-Sung Kim,a Jung-Hwan Hyung,a

Seung-Yong Lee,a Hyuneui Lim,c Rong Fan*bd and Sang-Kwon Lee‡*ab

Received 18th September 2011, Accepted 22nd November 2011

DOI: 10.1039/c2nr11338d

We report the development of a novel quartz nanopillar (QNP) array cell separation system capable of

selectively capturing and isolating a single cell population including primary CD4+ T lymphocytes from

the whole pool of splenocytes. Integrated with a photolithographically patterned hemocytometer

structure, the streptavidin (STR)-functionalized-QNP (STR-QNP) arrays allow for direct quantitation

of captured cells using high content imaging. This technology exhibits an excellent separation yield

(efficiency) of �95.3 � 1.1% for the CD4+ T lymphocytes from the mouse splenocyte suspensions and

good linear response for quantitating captured CD4+ T-lymphoblasts, which is comparable to flow

cytometry and outperforms any non-nanostructured surface capture techniques, i.e. cell panning. This

nanopillar hemocytometer represents a simple, yet efficient cell capture and counting technology and

may find immediate applications for diagnosis and immune monitoring in the point-of-care setting.

1. Introduction

Capture, separation and counting of rare cells is one of the most

essential tasks required for studying biological characteristics of

these cells and applying these cells to transplantation, anti-tumor

cell therapy, etc.1,2 The nanometre-sized magnetic bead method

(Dynabeads�, Invitrogen, USA) is widely used for capturing and

isolating specific cells, i.e., T-lymphocytes and circulating tumor

cells (CTCs). It utilizes capture-agent-coated magnetic beads to

immunologically recognize the specific cells in the blood followed

by passing the cell–bead complex through a magnetic field for cell

isolation. However, this approach still needs further improve-

ment in capture efficiency and is limited by the viability and

functionality of the cells isolated.3–5 Other variants such as cell

separation using magnetic nanowires (NWs) have been demon-

strated.6–8 To address the need to overcome the exposure of

separated cells to magnetic field and to develop a cost-effective

material-based system, we recently demonstrated a novel plat-

form to separate CD4+ T lymphocytes among mouse splenocytes

using streptavidin (STR)-functionalized silicon nanowire

aDepartment of Semiconductor Science and Technology, Chonbuk NationalUniversity, Jeonju, 561-756, Korea. E-mail: sk_lee@chonbuk.ac.kr;sangkwon.lee@yale.edu; Tel: +82-63-270-3973bDepartment of Biomedical Engineering, Yale University, New Haven, CT,06511, USA. E-mail: rong.fan@yale.edu; Tel: +1-203-432-9905cDepartment of Nature-Inspired Nanoconvergence Systems, KoreanInstitute of Machinery and Materials (KIMM), Daejeon, 305-343, KoreadYale Comprehensive Cancer Center, New Haven, CT, 06520, USA

† Electronic supplementary information (ESI) available. See DOI:10.1039/c2nr11338d

‡ Currently Visiting Professor at Department of Biomedical Engineering,Yale University. Tel.: +1-203-823-8121.

2500 | Nanoscale, 2012, 4, 2500–2507

(SiNW) arrays with a separation efficiency of greater than 93%.3

These great properties suggest that the silicon NW array is

a promising platform for sensing biomaterials and for molecular

bioelectronics, due to good biocompatibility of the silicon

surface and ease of control over the density, diameter, and length

of SiNWs. Wang et al. demonstrated that three-dimensional

(3D)-nanostructured SiNWs on a substrate coated with antibody

against epithelial cell adhesion molecule (anti-EpCAM) exhibi-

ted high capture efficiency when employed to isolate EpCAM

positive rare tumor cells.9,10 Despite a significant improvement in

the separation efficiency of SiNW-based capturing/isolating

devices, a reliable and semi-automated platform for simulta-

neous capture and counting of primary cells is yet to be realized.

Herein, we propose to develop nanostructured quartz for

quantification of captured cells via cost-effective methods that

will provide valuable information about managing and early

stage detection of cancer and human immunodeficiency virus

(HIV) infected patients.11–16 The quartz nanopillar (QNP) array-

based cell separation system we developed is capable of selec-

tively capturing/isolating a single cell population, and counting

isolated cells with a photolithographically patterned grid struc-

ture on the STR-functionalized-QNP (STR-QNP) arrays on one

chip. In addition, we first demonstrate the ability to visualize and

quantify the captured T-cells on a large-scale STR-QNP array.

Our technology possesses a number of key merits as compared to

those in previous reports, including our previous report on

SiNW-based cell capturing/isolating devices.3,9,10 First, we suc-

ceeded in separating CD4+ T lymphocytes extracted from mouse

spleen using transparent QNP arrays, and present a separation

efficiency greater than �95%. Second, we established a system

capable of separating lymphocytes and simultaneously

This journal is ª The Royal Society of Chemistry 2012

Publ

ishe

d on

04

Janu

ary

2012

. Dow

nloa

ded

by Y

ale

Uni

vers

ity L

ibra

ry o

n 01

/07/

2013

18:

35:3

5.

View Article Online

performing specific cell counting in a chip. Third, we further

extend this platform to large-area cell capture and counting,

making it a practical tool for studying rare cells. Fourth, this

study demonstrates informative analysis of captured cells

including cell size, shape, and distribution through high-content

fluorescence imaging.17,18 Fifth, this research is not limited to

separating CD4+ T lymphocytes; our proposed approach is also

applicable to other lymphocytes such as B, NK, and NKT cells,

as well as circulating tumor cells (CTCs).9,10,19,20 It may also

enable in situ measurement of cytokine or chemokines secreted

from these cells via spotting the cognate antibodies on the same

QNP substrate.21–25

2. Experimental details

2.1 QNP array preparation

Fig. 1a–h show a schematic diagram of the QNP fabrication

process and mechanism of capturing CD4+ T lymphocytes on the

STR-functionalized QNP arrays. A monolayer of polystyrene

(PS) nanoparticles (NPs) was prepared using spin coating on

a quartz substrate. In this method, the NP monolayer acts as an

etching mask. The colloidal PS beads (about 100 nm in diameter)

were purchased from Polysciences Inc. in the USA. The typical

etching process consists of three steps: (1) size reduction of the PS

NPs with CF4 plasma treatment, (2) nanostructuring with CF4 +

H2 plasma etching, and (3) removal of the residual PS NPs using

the piranha cleaning process (96% H2SO4 : 30% H2O2 ¼ 1 : 1).

The length of the QNP arrays is controlled by the etching time.

The diameter and length of synthesized QNP on a quartz

substrate were found to be in the range of 30 to 80 nm and 200 to

1800 nm, respectively, which strongly depended on the plasma

etching times.

Fig. 1 Schematic diagram of the QNP fabrication process and QNP

surface functionalization. (a) Spin-coating process of polystyrene (PS)

nanoparticles (NPs) on a flat quartz substrate. The size of the PS NPs was

about 100 nm. These were purchased from Polyscience Inc. in the USA.

(b)–(e) Plasma etching and cleaning process. The spin-coated PS NPs

were etched and cleaned in three steps, (b) size reduction with CF4 plasma

etching, (c and d) quartz plasma etching with the mixture of CF4 and H2,

and (e) removal of the residual PS NPs on the surface using the standard

Piranha cleaning process. (f) Quartz surface functionalization. The

etched QNP arrays employed were 3-aminopropyltriethoxysilane

(APTES), glutaraldehyde (GA) and streptavidin (STR) in turn. (g) Cell

suspension loading. A cell suspension containing CD4+ T-lymphocytes

pre-mixed with biotinylated CD4-mAb was then loaded onto STR-

functionalized QNP arrays. (h) Blocking and washing process. Unbound

splenocytes were washed out with PBS solution on a 2D rocker.

This journal is ª The Royal Society of Chemistry 2012

2.2 CD4+ T-lymphocyte separation from mouse splenocytes

using STR-QNP arrays

The QNP surface was treated with O2 plasma for 60 s to present

the hydroxyl groups on the surface of the QNP arrays. The 3-

aminopropyltriethoxysilane (APTES, Sigma-Aldrich, USA) was

treated in ethanol for 30 min at room temperature to function-

alize the QNP surface with amine groups. The amine groups of

the QNP arrays were reacted with glutaraldehyde (GA, Sigma-

Aldrich, USA) to confer an aldehyde group that sequentially

reacted with streptavidin (STR, Sigma-Aldrich, USA). For the

preparation of mouse splenocytes, C57BL/6/mice (6–8 weeks)

were purchased from Nara Biotech Co. in Korea. The extracted

spleens were passed through a nylon wool column to deplete B-

lymphocytes, remaining CD4+ T, CD8+ T, NK, and NKT cells in

mouse splenocytes. A certain quantity of cells (�105 cells per mL)

was counted using a conventional hemocytometer (Hausser

Scientific Co., USA). Prior to loading the cell suspension in the

culture medium, the cell population (105 cells per plate) was first

reacted with biotinylated anti-CD4 mAb (clone GK 1.5, eBio-

science Inc. USA) and incubated at 4 �C for 20 min. Non-

specifically unbound cells with biotin-conjugated mAbs were

washed out, while separated CD4+ T cells could bind to QNP

surfaces due to adhesion enabled by the antigen–antibody

interaction. Remnants of unbound CD4+ T-lymphocytes were

then enumerated using a flow cytometer (Becton Dickinson,

USA) after being washed several times in phosphate buffered

saline (PBS) solution on a two-dimensional (2-D) rocker and

analyzed using CellQuest� Pro software (BD Bioscience, USA).

2.3 Large-scale STR-QNP arrays for cell counting devices

Cell-counting reservoirs with nine circular wells (5 mm in diam-

eter) were made by molding a polydimethylsiloxane (PDMS)

elastomer. A mixture of GE RTV 615 PDMS (GE Silicones,

USA) prepolymer part A and part B (10 : 1) was prepared,

homogenized, and applied to a blank Petri dish (55 mm in

diameter). After degassing for 1 h, the PDMS mold was cured at

about 80 �C for 60 min. The solidified PDMSmold was cut off in

the size of 25 � 25 mm2 and the cell-counting reservoir holes

(5 mm in diameter) were drilled with a stainless steel hole punch.

For large-scale application of the STR-functionalized QNP

arrays, CD4+ T-lymphoblast (human T-lymphoblast cell line,

CCRF-CEM) was purchased from American Type Culture

Collection (ATCC) in the USA. Prior to introducing the CD4+ T-

lymphoblast cells into the PDMS wells, the CD4+ T-cells were

premixed with biotinylated anti-human CD4-Ab (clone OKT4,

eBioscience Inc., USA) and then stored at 4 �C for 20 min. A

solution of CD4+T cells conjugatedwith biotinylated anti-human

CD4-Abs (�40 mL in PBS) was pipetted into each eight wells. Cell

counts were manually performed using a conventional hemocy-

tometer (Hausser Scientific Co. USA) within 10% error. A series

of counted and differently diluted cells in RPMI-1640 medium

with a final volume of about 50 mL for each well was introduced

into eight wells (1� 1, 1� 2, 1� 3, 2� 1, 2� 2, 2� 3, 3� 1, and

3 � 2) with cell populations of approximately 500, 1000, 1500,

2000, 2500, 3000, 3500, and 4000 cells per mL, respectively. After

incubation in a 37 �C incubator for 40 min, the PDMS wells were

washed out 1 � PBS with Tween-20 (PBST, KPL Inc., USA) for

Nanoscale, 2012, 4, 2500–2507 | 2501

Publ

ishe

d on

04

Janu

ary

2012

. Dow

nloa

ded

by Y

ale

Uni

vers

ity L

ibra

ry o

n 01

/07/

2013

18:

35:3

5.

View Article Online

at least three times to remove unbound CD4+ T-lymphoblast

cells. Subsequently, all of the immobilized T-lymphoblast cells

were stained with amixture of fluorescein isothiocyanate (FITC)-

labeled CD3-Ab and allophycocyanin (APC)-labeled CD4-Ab

(eBioscience Inc., USA) to identify the CD4+ T-cells with CD4+

and CD3+ expression. To fix the captured cells, 4% para-

formaldehyde (PFA, Santa Cruz Biotechnology Inc., USA) in

PBSwas loaded into the wells for 15min, followed bywashing out

in PBST. The stained STR-QNP arrays were finally given a three-

step cleaning process: PBS, PBS inDI water (1 : 1), andDI water.

Both CD3-FITC and CD4-APC stained cells on the STR-QNP

arrays were then scanned using an Axon Genepix Scanner 4000B

(Molecular Devices, USA) with 635 nm and 532 nm excitation

wavelengths. The finest resolution of this microarray scanner is as

low as about 5 mm. The scanned images of the captured CD4+ T

cells were visualized using Genepix 6.0 software (Molecular

Devices, USA). The visualized CD4+ T-cells images exported

from the Genepix 6.0 were analyzed in order to quantify the

population of the captured CD4+ T cells using CellProfiler�(http://www.cellprofiler.org), cell image analysis software.26

3. Results and discussion

3.1 CD4+ T-lymphocyte separation from mouse whole

splenocytes

We fabricated the QNP arrays using the combination of colloidal

lithography and plasma etching.27 After preparing the QNP

arrays, we employed 3-aminopropyltriethoxysilane (APTES)

and glutaraldehyde (GA) sequentially to immobilize streptavidin

(STR) onto the QNP surfaces. Then, a cell suspension containing

CD4+ T-lymphocytes premixed with biotinylated anti-CD4

mouse antibodies (mAbs) was introduced onto the STR-func-

tionalized QNP arrays prior to the cell separation process (see

details in the Experimental section). As shown in Fig. 2a–c,

Fig. 2 SEM images of CD4+ T lymphocytes bound on the STR-functionaliz

magnification on STR-QNP arrays. (b and c) Side and top-view SEM imag

sectional SEM images of surface-bound CD4+ T lymphocytes on QNP array

regions B and C in (d), respectively. The samples were prepared using a focused

surface are highlighted in red for easy differentiation.

2502 | Nanoscale, 2012, 4, 2500–2507

captured CD4+ T lymphocytes, highlighted in red, firmly adhere

onto the STR-functionalized QNP arrays via a biotin-STR

binding since this conjugation between STR and biotinylated

CD4 mAb has a high binding affinity (Kd ¼ 10�15 M).3 We

noticed that the bound CD4+ T lymphocytes on the surface were

in the range of 3 to 5 mm in diameter as shown in Fig. 2 (upper

panels). To further investigate the binding between separated

cells and the STR-QNP array and to examine the interface

between the cell and the substrate surface, gallium (Ga+) ion

milling of the separated T-lymphocytes was performed using

a focused ion beam (FIB, FEI Co., USA). As shown in the

electron micrographs for the cross-sectional views in Fig. 2

(lower panels), we observed that the captured CD4+ T lympho-

cytes were tightly bound on the STR-QNP arrays with no air

pockets or cracks at the interface.

Fig. 3a presents the fluorescence-activated cell sorting (FACS)

results of CD4+ T lymphocyte separation using an STR-QNP

array and compared to the results with a bare glass wafer (top

and bottom left panels). According to the flow cytometric anal-

ysis shown in Fig. 3a (left panel), the percentages of the CD4+ T

lymphocytes (CD4+/CD3+ double positive) and the non-CD4+ T

lymphocytes (CD4�/CD3� and CD3+) in the whole mouse sple-

nocyte mixture used for the experiment were determined to be

�37.2% and �62.8%, respectively. After introducing this cell

suspension onto the STR-functionalized QNP arrays, we

observed a significant decrease in the percentage of CD4+ T

lymphocytes in the cell suspension to�1.5% (left bottom panel in

Fig. 3a), indicating that the STR-QNP arrays completely sepa-

rated the CD4+ T lymphocyte from the mixed cell population.

The separation yield (efficiency), which is defined by the

percentage of cells captured to cells initially loaded, for different

cell separation platforms—STR-QNP arrays and the STR-

functionalized glass wafer—were calculated using flow cyto-

metric analysis. The average separation efficiency of CD4+ T-

lymphocytes captured on STR-QNP arrays from whole mouse

ed QNP arrays. (a) Tilted SEM image of CD4+ T lymphocytes with low

es of region A in (a) with high magnification, respectively. (d–f) Cross-

s. (d) Low, and (e and f) high magnification image of separated cells of

ion beam (Ga+ ion milling). All of the CD4+ T lymphocytes bound on the

This journal is ª The Royal Society of Chemistry 2012

Fig. 3 Flow cytometric analysis of CD4+ T lymphocytes after cell separation. (a) Flow cytometric analysis of CD4+ T lymphocytes using cell

suspensions after binding to STR-QNP arrays (bottom left panel) and glass wafer (top left panel). Average cell separation efficiency is shown using bar

graphs (right panel). (b) Flow cytometric analysis of CD4+ T lymphocytes before and after cell separation for different diameter-to-length aspect ratios

(ARs) from 1 : 1 to 1 : 10 (top and bottom left panels) with averaged separation efficiency shown in bar graph (top right) and table (bottom right).

Publ

ishe

d on

04

Janu

ary

2012

. Dow

nloa

ded

by Y

ale

Uni

vers

ity L

ibra

ry o

n 01

/07/

2013

18:

35:3

5.

View Article Online

lymphocytes was determined to be�95.3� 1.1% (n¼ 3), which is

the highest value compared to other reports in the literature.3,9,10

The performance is even slightly better than that obtained with

STR-SiNWs.3 For comparison, the bare glass wafer with no

nanostructures gave rise to a cell separation efficiency of�65.8�2.3%. The elevated cell separation efficiency of STR-QNP arrays

compared to previous reports in the literature can be explained as

follows. First, it was evident that nanostructures on the quartz

substrate increased the contact area between cells and the

substrate surface compared to flat glass wafer.3,9 As a result,

QNP arrays can separate a larger number of loaded cells, thereby

increasing cell separation efficiency. Second, the higher separa-

tion efficiency of the STR-QNP arrays compared to our prior

results on STR-SiNW arrays is likely due to the relatively higher

surface area per cell on the QNP arrays than on the STR-SiNW

arrays. The QNP arrays possessed approximately 500 to 800

QNPs per cell as seen from Fig. 2c, while the STR-SiNW

structure only had 3 to 4 SiNWs per cell. Consequently, we

This journal is ª The Royal Society of Chemistry 2012

observed that the QNP array did capture more cells, yielding

higher cell separation efficiency. Additionally, we verified that

the O2 plasma treatment unquestionably assisted STR immobi-

lization on the QNP arrays (see Fig. S1 in the ESI†).

To investigate the effect of the diameter-to-length ARs in

STR-functionalized QNP arrays, four STR-QNP arrays with

different ARs (1 : 1, 1 : 4, 1 : 5, and 1 : 10 diameter-to-length)

were prepared using the same fabrication processes. Fig. 3b

shows the capture efficiency of four different QNP arrays before

and after capturing processes together with a summary of the

results in the bar graph and table. As shown in Fig. 3b (bar graph

and table in the right panel), the capture efficiency of STR-QNP

arrays increased from �84.3 � 4.8% to �95.3 � 1.1% as

increasing ARs from 1 : 1 to 1 : 4. By further increasing the AR

to 1 : 10, the capture efficiency of the STR-QNP arrays then

decreased to �79.5 � 2.2% for the STR-QNP with an AR of

1 : 10. We speculate that the enhancement of capture efficiency

with increasing AR from 1 : 1 to 1 : 4 may be attributed to

Nanoscale, 2012, 4, 2500–2507 | 2503

Publ

ishe

d on

04

Janu

ary

2012

. Dow

nloa

ded

by Y

ale

Uni

vers

ity L

ibra

ry o

n 01

/07/

2013

18:

35:3

5.

View Article Online

increased interaction between nanopillars and the microvillus of

CD4+ T cells.9,11 However, this should not help any more when

the AR is sufficiently large and provides maximum contact area.

As a result, we expect to see a plateau of capture efficiency with

further increasing AR. But we observed a decrease of capture

efficiency with increasing AR to 1 : 10. This could be due to the

reduced nanopillar size caused by plasma thinning of nanosphere

masks and consequently reduced contact area between pillars

and cellular microvillus. Our observations are consistent with

previous studies performed using various types of cells including

platelets and fibrinogen.28–30

Furthermore, to identify that the cells separated using STR-

QNP arrays were CD4+ T lymphocytes and to ensure dead cells,

the captured cells were stained by phycoerythrin (PE)-conjugated

anti-CD4 mAb, 40-6-diamidino-2-phenylindole (DAPI), and

propidium iodide (PI). The cells’ images were then detected using

fluorescence microscopy (Eclipse Ti-U, Japan). DAPI generally

binds to all nuclei of the cells, while PI binds to only dead cells.

Fig. 4 shows optical and fluorescence images of stained CD4+ T-

lymphocytes with PE-labeled CD4 mAb (shown in red of Fig. 4c

and d), DAPI (shown in blue of Fig. 4f and h), and PI (shown in

red of Fig. 4g and h). As shown in Fig. 4c and d, we confirmed

that most cells were CD4+ T lymphocytes from indication of PE-

labeled CD4 mAb positive.

3.2 STR-QNP arrays as a miniaturized cell counting device

(hemocytometer)

For potential use as a miniaturized cell counting device, we

prepared STR-QNP arrays with integrated photolithography-

patterned grids. Fig. 5 shows the images of STR-QNP arrays

with a patterned grid structure (left panels) and the comparison

of cell counts between FACS analysis and the QNP-based

hemocytometer analysis (right panels). We measured the pop-

ulation of CD4+ T lymphocytes captured on the patterned

Fig. 4 Optical and fluorescence image of CD4+ T lymphocytes bound on the

optical images of CD4+ T lymphocytes on STR-QNP arrays. (c and d) Low

surface, where the separated cells were stained by PE-labeled CD4-mAb (red

captured cells are undoubtedly CD4+ T lymphocytes. (e) Optical image of CD

DAPI (blue color) and PI (red color). (h) Overlapped fluorescence image of D

observed from the fluorescence images of DAPI and PI.

2504 | Nanoscale, 2012, 4, 2500–2507

hemocytometer on STR-QNP arrays using optical microscopy

and compared it to the results from the flow cytometry anal-

ysis.7,18 To create the grid on the hemocytometer shown in

Fig. 5a–d, we first used photolithography to define the microscale

grids on the QNP array that define square patterns with a single

square of 200� 200 mm2 over a total area of 3� 3 mm2, followed

by a pattern transfer process using wet-chemical etching with

buffered hydrofluoric acid (BHF). Then, we separated the CD4+

T lymphocytes and averaged the numbers of captured/isolated

CD4+ T lymphocytes in all four-unit areas to achieve statistic

consistency in a manner similar to a conventional hemocytom-

eter (Hausser Scientific Co., USA). The formula for the calcu-

lation of the captured cells was (4 unit areas) � (1/4) � (1/unit

volume) with a unit volume of �1.87 � 10�5 cm3. The total

population of captured/isolated CD4+ T lymphocytes using

STR-QNP arrays was thus determined to be about 8.2 � 105 �5.6 � 104 cells per mL (n ¼ 3). As shown in Fig. 5e, we confirm

that the cell counts obtained from the STR-QNP-based hemo-

cytometer are in close agreement with that determined by the

flow cytometer analysis (CD4+ T lymphocytes �9.4 � 105 �6.9 � 104 cells per mL, n ¼ 3). From the hemocytometer exper-

iments using the STR-functionalized QNP arrays, we found that

our cost-effective and simply structured QNP array system has

promising potential in that it eliminates the use of bulky and

expensive equipments (such as flow cytometers) to quantify the

cell population, and obviates the need for additional sample

processing procedures (such as fluorescence probe staining).

3.3 Large-scale STR-QNP arrays as a cell counting device

To further assess the possibility of extending this platform to

large-scale cell capturing and counting, STR-QNP arrays (25 �25 mm2, see Fig. S3a in the ESI†) were prepared. We selected

human T-lymphoblast cancer cells (CCRF-CEM) as the target

cells for detection and counting. FACS analysis indicated that

surface of STR-QNP arrays wafer. (a and b) Low and high magnification

and high magnification of fluorescence images of separated cells on the

color) for identification of CD4+ T lymphocytes, indicating that all of the

4+ T cells. (f and g) Fluorescence images of CD4+ T lymphocytes stained

API and PI on optical image. No dead cells (red-colored cells in (g)) were

This journal is ª The Royal Society of Chemistry 2012

Fig. 5 STR-QNP array-based miniature cell counter (hemocytometer). (a) Photograph of a STR-QNP cell counter with integrating photolithography-

patterned grid structures. (b and c) Low and high magnification SEM images of a patterned STR-QNP cell counter, respectively. For the grid patterning

on the QNP arrays, plasma etching was performed on the QNP structure after standard photolithography. (d) High magnification optical image of

captured/isolated CD4+ T lymphocytes on STR-immobilized QNP surface, showing surface-bound T-cells that are well distributed on the surface. (e)

Summary of comparison results of counted cell population from FACS analysis and QNP-based cell counter (hemocytometer) (top bar graph and

bottom table). Error bar shows standard deviations (n ¼ 3). For the immobilized cell counting with the QNP-based cell counter, the population of

captured/isolated CD4+ T lymphocytes on four unit areas (200 � 200 mm2) was averaged to reduce errors as we performed using a conventional

hemocytometer (see additional details in the main text).

Publ

ishe

d on

04

Janu

ary

2012

. Dow

nloa

ded

by Y

ale

Uni

vers

ity L

ibra

ry o

n 01

/07/

2013

18:

35:3

5.

View Article Online

these cells comprise both CD4+ CD3+ and CD4+ CD3� pheno-

types while almost all cells express CD4, making it most suitable

for studying CD4 T cell counting. For visualizing and counting

the captured cells on the large-scale area of the STR-QNP arrays,

we used an Axon Genepix Scanner 4000B. T cells were intro-

duced to each polydimethylsiloxane (PDMS) well in the range of

approximately 500 to 4000 cells per well. Fig. 6a and b (see also

Fig. S3a–c in the ESI†) show scanned fluorescence images of

immobilized CD4+ T-lymphoblast cells on large-scale STR-QNP

arrays. Fig. 6c shows the quantitative analysis of these images

using high content cell biology analysis software, CellProfiler�.26

The size distribution in pixels of immobilized CD4+ T-lympho-

blasts provided from the analysis using CellProfiler� is shown in

Fig. 6d. Using immunofluorescence staining with fluorescein

isothiocyanate (FITC)-conjugated antib-CD3 Ab (green) and

allophycocyanin (APC)-labeled CD4-Ab (red), we verified that

captured CD4+ T-lymphoblasts are positive for CD4 and many

express CD3 as shown in Fig. 6a. As shown in Fig. 6a and S4a–c

(ESI†), the population and size of surface-bound CD4+ T-

lymphoblasts on the STR-QNP arrays increase with increasing

the quantity of cells loaded into the well. It was noted that with

increasing cell number two or more CD4+ T-lymphoblast cells

agglomerate to one another as shown in the 2� 3, 3� 1, and 3�2 wells in Fig. 6c. This leads to underestimation of cell counts and

affects cell separation efficiency when the cells of interests are not

so rare (>2000 per well). Fig. 6e shows the numbers of the T-

lymphoblast cells loaded in the wells as compared to the cell

counts measured using the STR-QNP-based hemocytometer. As

shown in Fig. 6e, cell separation efficiency was found to be in the

range of 55 to 75%. These lower nominal capture efficiencies

observed in large-scale STR-functionalized QNP arrays could be

due to several factors. The values computed using FACS analysis

correspond to the percentage of CD4+ T cells removed by

nanopillar arrays, but the imaging cytometry analyses count the

This journal is ª The Royal Society of Chemistry 2012

CD4+ T cells that still remained on the substrate after several

washing steps. With respect to human T cell capture experiments,

although the cell line was identified as a CD4+ phenotype, but

a small fraction of these cells express no or low levels of CD4

antigen. In addition, cell size, shape and adhesion force all vary

among different cell samples and alter the interaction between

cells and nanostructures to some extent, and thus affect the

actual counts of cells remaining on the nanopillar substrate.

Previous reports have also been pointed out that various types of

cells represented different cellular responses when the topo-

graphic features of the surface were varied.9,28–31 With increasing

cell loading ranging from �500 to �2500 cells, the cell count

increased linearly until reaching 4000 cells and then remain

nearly unchanged with further increasing cell loading. To auto-

matically quantitate the captured cell population, the cells were

analyzed according to size distribution shown in the data

generated by the CellProfier� software. Our observation shows

that captured CD4+ T-lymphoblasts were in the size range of

225–1806 mm2 and those greater than 2500 mm2 were believed to

be cell agglomerates. Fig. 6f shows the distribution of captured

cells greater than 10 pixels in each well. This indicates that

loading of more than �2500 cells to a well causes cells to

agglomerate and decreases the cell counting accuracy. On the

basis of these findings, algorithms can be used in the image

analysis program to reduce this effect and extend the linear range

as in Fig. 6e. Moreover, we demonstrated that the use of a high-

content imaging approach allowed for more rapid32–34 and

quantitative analysis of cell markers or signaling proteins using

fluorescence signals from captured cells on the STR-QNP arrays.

In addition, the captured specific cells can be easily visualized

and quantitated in a semi-automated manner over a large-area

providing a practical means for direct counting of rare cells.34

Immobilizing cells on the surface has a number of advantages

including ease of monitoring over time and the possibility to

Nanoscale, 2012, 4, 2500–2507 | 2505

Fig. 6 Microarray-scanned images of immobilized T-lymphoblast (CCRF-CEM) on large-scale STR-QNP arrays. (a) Low magnification and (b)

enlarged images (white box region shown in the first row) of fluorescence-stained T-lymphoblast (CCRF-CEM), which was pre-mixed with biotin-

conjugated anti-human CD4-Ab prior to the cell separation process, for eight wells (1� 1, 1� 2, 1� 3, 2� 1, 2� 2, 2� 3, 3� 1, and 3� 2) with FITC-

labeled anti-humanCD3-Ab (green color), thereby indentifying these cells as CD4+ T cells (double positive CD3+/CD4+). The original images taken from

a microarray scanner were converted into a gray color, as shown in (b), for further analysis. The visualized T-lymphoblast images (shown in (a) and (b))

exported fromGenepix software were analyzed to quantify the population of the captured CD4+ T cells using cell image analysis software, CellProfiler�(ref. 26). (c) CellProfiler� generated outlined images of separated T-lymphoblast cells on STR-QNP arrays for cell counting. The red-marked regions

provided from CellProfiler� in the 2 � 3, 3 � 1, and 3 � 2 wells were excluded from cell counting due to poor image quality. (d) Cell size distribution in

pixels of immobilized T-lymphoblast cells on STR-QNP arrays for each well (5 mm in a pixel), showing that the cell size increased with increase in loaded

cell population. (e) Comparison of cell population of immobilized T-lymphoblast (CCRF-CEM) cells on STR-QNP arrays at each well. The cell

population was modified on the basis of the cell size distribution shown in (a). (f) Cell size and population distribution where the cell size is >10 pixels

(1 pixel is as low as �5 mm) at eight different wells. The cell size of immobilized T-lymphoblasts was analyzed using CellProfiler� software.

Publ

ishe

d on

04

Janu

ary

2012

. Dow

nloa

ded

by Y

ale

Uni

vers

ity L

ibra

ry o

n 01

/07/

2013

18:

35:3

5.

View Article Online

analyze cell morphology, migration and function.33,34 In this

report, we describe the experiments to employ CD4+ T

lymphocytes as target cells. However, its generic platform is not

limited to the study of CD4+ T lymphocytes. First, the conju-

gation of biotin to antibodies against a variety of cell surface

markers allows for the application to capturing and isolating

many other types of cells.35 Second, this platform can be

extended to the study of cellular functions such as the production

of cytokines/chemokines, which are essential for the body’s

immune responses and homeostatic control in both health and

disease.21–25 Many cytokines and chemokines have been identi-

fied as dysregulated in various diseases, and thus such a cell

capture and functional analysis technology would have great

value for clinical diagnosis and immune monitoring. Another

promising application is the detection of CTCs. CTCs are the

2506 | Nanoscale, 2012, 4, 2500–2507

disseminating tumor cells circulating in the peripheral blood of

patients and have been found to be an early biomarker to detect

cancer metastasis. Also the quantification of CTCs in patient

blood offers valuable information for monitoring the progres-

sion or therapeutic responses of cancer patients.19,20

4. Conclusion

In summary, we have succeeded in developing a new technology

platform for capturing and counting specific cells via a minia-

turized quartz nanopillar array hemocytometer. We also pre-

sented the first demonstration of direct visualization and

automated quantitation of captured T-cells over a large-scale

chip, representing a practical tool for low-cost and high-

throughput rare cell numeration. The STR-QNP arrays offer an

This journal is ª The Royal Society of Chemistry 2012

Publ

ishe

d on

04

Janu

ary

2012

. Dow

nloa

ded

by Y

ale

Uni

vers

ity L

ibra

ry o

n 01

/07/

2013

18:

35:3

5.

View Article Online

excellent platform for the separation and isolation of a wide

variety of cells such as regulatory T lymphocytes, circulating

tumor cells or tumor stem cells. With further development, our

system can monitor the function of captured single cells and

detect the secretion of immune effector molecules such as cyto-

kines and chemokines. Thus, the quartz nanopillar array

hemocytometer may serve as a foundational platform to develop

tools for studying the biology of rare cells from mixed pop-

ulations or tissue samples.

Acknowledgements

This study was supported by the Priority Research Centers

Program and by Basic Science Research Program through the

National Research Foundation of Korea (NRF) funded by the

Ministry of Education, Science and Technology (2010-0029706

and 2010-0019694) and also partially supported by a grant from

the KRIBB Research Initiative Program, the U.S. National

Cancer Institute Howard Temin Pathway to Independence

Award (NIH 4R00 CA136759-02, PI: R.F.), and partly by the

Alzheimer’s Association New Investigator Research Grant

(2010). S.K.L. was supported by Visiting Professor Program

(2011) from LG-Yonam foundation for visiting the Biomedical

Engineering Department of Yale University. Y.W. is supported

by the Anderson Postdoctoral Fellowship. S.K.L. and R.F.

thank Dr M. K. Lee, Dr S. T. Kim, Dr Lin Han, and Dr Yao Lu

for the fruitful discussion on our cell separation experiments.

References

1 D. Adeegbe, R. B. Levy and T. R. Malek, Blood, 2010, 115, 1932.2 H. Fujisaki, H. Kakuda, N. Shimasaki, C. Imai, J. Ma, T. Lockey,P. Eldridge, W. H. Leung and D. Campana, Cancer Res., 2009, 69,4010.

3 S. T. Kim, D. J. Kim, T. J. Kim, D. W. Seo, T. H. Kim, S. Y. Lee,K. H. Kim, K. M. Lee and S. K. Lee, Nano Lett., 2010, 10, 2877.

4 M. Cristofanilli, G. T. Budd, M. J. Ellis, A. Stopeck, J. Matera,M. C. Miller, J. M. Reuben, G. V. Doyle, W. J. Allard,L. W. M. M. Terstappen and D. F. Hayes, N. Engl. J. Med., 2004,351, 781.

5 A. H. Talasaz, A. A. Powell, D. E. Huber, J. G. Berbee, K.-H. Roh,W. Yu, W. Xiao, M. M. Davis, R. F. Pease, M. N. Mindrinos,S. S. Jeffrey and R. W. Davis, Proc. Natl. Acad. Sci. U. S. A., 2009,106, 3970.

6 A. Hultgren and M. J. Tanase, J. Appl. Phys., 2003, 93, 7554.7 A. Hultgren, M. Tanase, E. J. Felton, K. Bhadriraju, A. K. Salem,C. S. Chen and D. H. Reich, Biotechnol. Prog., 2005, 21, 509.

8 A. Prina-Mello, A. Diao and M. D. Coey, J. Nanobiotechnol., 2006,4, 9.

This journal is ª The Royal Society of Chemistry 2012

9 S. Wang, H. Wang, J. Jiao, K. J. Chen, G. E. Owens, K. I. Kamei,J. K. Sun, D. J. Sherman, C. P. Behrenbruch, H. Wu andH. R. Tseng, Angew. Chem., Int. Ed., 2009, 48, 8970.

10 S. Wang, K. Liu, J. Liu, Z. T. F. Yu, X. Xu, L. Zhano, T. Lee,E. K. Lee, J. Reiss, Y. K. Lee, L. W. K. Chung, J. Huang,M. Rettig, D. Seligson, K. N. Duraiswamy, C. K. F. Shen andH. R. Tseng, Angew. Chem., Int. Ed., 2011, 50, 3084.

11 H. Zhu, M. Macal, C. N. Jones, M. D. Goerge, S. Dandekar andA. Revzin, Anal. Chim. Acta, 2008, 608, 186.

12 N. N. Watkins, S. Sridhar, X. Cheng, G. D. Chen, M. Toner,W. Rodriguez and R. Bashir, Lab Chip, 2011, 11, 1437.

13 H. Zhu, G. Styhayeva, M. Macal, E. Ramanculov, M. D. George,S. Dandekar and A. Revzin, Lab Chip, 2008, 8, 2197.

14 K. Wang, B. Cometti and D. Pappas, Anal. Chim. Acta, 2007, 601, 1.15 X. Cheng, D. Irimia, M. Dixon, K. Sekine, U. Demirei, L. Zamir,

R. G. Tompkins,W. Rodriguez andM. Toner,Lab Chip, 2007, 7, 170.16 X. Cheng, A. Gupta, C. Chen, R. G. Tompkins, W. Rodriguez and

M. Toner, Lab Chip, 2009, 9, 1357.17 J. Guarner, P. Montoya, C. del Rio and G. Hernandez-Tepichin,

Cytometry, 1997, 30, 178.18 G. G. Sherman, J. S. Galpin, J. M. Patel, B. V. Mendelow and

D. K. Glencross, J. Immunol. Methods, 1999, 222, 209.19 S. Nagrath, L. V. Sequist, S. Maheswaran, D. W. Bell, D. Irimia,

L. Ulkus, M. R. Smith, E. L. Kwak, S. J. Digumarthy,H. Muzikansky, P. Ryan, U. J. Balis, R. G. Tompkins,D. A. Haber and M. Toner, Nature, 2007, 450, 1235.

20 S. I. Kim and H. I. Jung, J. Breast Cancer, 2010, 13, 125.21 K. F. To, P. K. S. Chan, K. F. Chan, W. K. Lee, W. Y. Lam,

K. F. Wong, N. L. S. Tang, D. N. C. Tsang, R. Y. T. Sung,T. A. Buckley, J. S. Tam and A. F. Cheng, J. Med. Virol., 2001, 63,242.

22 D. Unutmaz, V. N. K. Ramani, S. Marmon and D. R. Littman,J. Exp. Med., 1999, 189, 1735.

23 C. K. Wong, C. W. K. Lam, A. K. L. Wu, W. K. Ip, N. L. S. Lee,I. H. S. Chan, L. C. W. Lit, D. S. C. Hui, M. H. M. Chan,S. S. C. Chung and J. J. Y. Sung, Clin. Exp. Immunol., 2004, 136, 95.

24 P. C. Ng, K. Li, T. F. Leung, R. P. O. Wong, G. Li, K. M. Chui,E. Wong, F. W. T. Cheng and T. F. Fok, Clin. Chem., 2006, 52, 1181.

25 R. Y. T. Sung, S. H. L. Hu, C. K. Wong, C. W. K. Lam and J. Yin,Eur. J. Pediatr., 2001, 160, 117.

26 Broad Institute, CellProfiler�, http://www.cellprofiler.org.27 H. Lim, D. H. Jung, D. H. Noh, G. R. Choi and W. D. Kim, Chin.

Sci. Bull., 2009, 54, 3613.28 L. B. Koh, I. Rodriguez and S. Venkatraman, Biomaterials, 2010, 31,

1533.29 W. T. Su, I. M. Chu, J. Y. Yang and C. D. Lin,Micron, 2006, 37, 699.30 J. D. Zawislak, K. W. Kolasinski and B. P. Helmke, Phys. Status

Solidi A, 2009, 206, 1356.31 CLS (cell lines service), http://www.cell-lines-service.de.32 A. Revzin, P. Rajagopaln, A. W. Tilles, F. Berthiaume,

M. C. Yarmush and M. Toner, Langmuir, 2004, 20, 2999.33 C. J. Flaim, S. Chien and S. N. Bhatia, Nat. Methods, 2005, 2, 119.34 A. K. Shalek, J. T. Robinson, E. S. Karp, J. K. Lee, D. R. Ahn,

T. S. Gujral, G. MacBeath, E. G. Yang and H. Park, Proc. Natl.Acad. Sci. U. S. A., 2010, 107, 1870.

35 BD bioscience human CD chart, http://www.bdbiosciences.com.

Nanoscale, 2012, 4, 2500–2507 | 2507