In Vitro Cell. Dev. Biol.--Animal 41:185-187, July August 2005 �9 2005 Society for In Vitro Biology 1071-2690/05 $18.00+0.00

A NOVEL CELL ARRAY TECHNIQUE FOR HIGH-THROUGHPUT, CELL-BASED ANALYSIS

A. WATERWORTH, A. HANBY, AND V. SPEIRS ~

Molecular Medicine Unit (A. W.., E S.) and Academic Unit of Pathology (A. W., A. H.), University of Leeds, Leeds, U.K.

(Received 10 May 2005; accepted 7 June 2005)

SUMMARY

Microarray technology has burgeoned over the past few years from nucleic acid-based arrays to tissue microarrays (TMAs). This study aimed to develop a technique to incorporate cell lines into an array and to demonstrate the usefulness of this technique by performing immunohistochemistry for 13-catenin. Cell suspensions were prepared from 23 tumor cell lines. These were fixed in fm~nalin, suspended in agar, and embedded in paraffin to produce a cell block. A "tissue microarrayer" was used to remove triplicate, 0.6 ram-cores from each cell block and to transfer these into a recipient paraffin block at precise coordinates. Immunohistochemistry was used to identify cell lines positive for ~-catenin. Cultured cells were successfully incorporated into the microarray, with preservation of cell architecture and even distribution of cells within each core. A total of 18 of 69 cores (26%) were lost in processing. A total of 16 of 23 cell lines were identified as positive for membrane and cytoplasmic ~-catenin, and 6 of 23 were negative. Only one cell line was unseorable because of complete core loss. We have developed a "cell microarray" technique for analyzing antigen ex- pression by immunohistochemistry in muhiple cell lines in a single experiment. This novel application of microarrays permits high-throughput, cost-efficient analysis, with the potential to rapidly identify markers with potential diagnostic and therapeutic implications in human disease.

Key words: cell lines; microarray; immunohistochemistry.

Microarray technology is offering unprecedented opportunities to identify new diagnostic markers and therapeutic targets in human cancer by greatly increasing the speed and volume of analysis of potentially important molecules. The deoxyribonucleic acid (DNA) microarrays allow measurement of gene expression, mutation, and function and are identifying a large number of candidate genes pos- sibly involved in tumorigenesis (DeRisi et al., 1996; Duggan et al., 1999). Tissue microarrays (TMAs) are a complementary tool and enable high-throughout analysis of protein expression--the ultimate effeetors of any gene function or dysfunction (Kononen et al., 1998; Ginestier et al., 2002). The TMAs sample hundreds of paraffin embedded, whole-tissue blocks and transfer representative cores from these into a single donor block, with the theoretical possibility of representing the entire pathology archive on a single microscope slide. The diameter of cores taken can be varied, hut 0.6 mm is the commonest currently used.

Incorporating cultured cells into a microarray is a relatively new concept. Cell culture is an important research tool. It plays a vital role in assessing the characteristics of tumor cells, can determine how their growth can be manipulated by different culture condi- tions, and also identify potential therapeutic agents. However, cell cultures are labor intensive to maintain, and if many different cell cultures could be incorporated efficiently into a microarray and an- alyzed simultaneously, this would dramatically increase the speed of analysis and reduce costs. An ideal use for such a cell array

1 To whom eolwespondence should be addressed at Molecular Medicine Unit, Clinical Sciences Building, St. James's University Hospital, Leeds LS9 7TF, U.K. E-mail: v.speirs@leeds.ac.uk

would be to determine expression of a particular antigen in a spe- cific cell line or tumor type, optimize antibodies, or find a suitable positive control for further cell culture work.

A frozen cell line array has been proposed by Stephan et al. (2002). Using a custom-made arrayer consisting of 25 blunt pins fixed in a block of Plexiglass, they generated an array of wells in a frozen block of optimal cutting temperature (OCT) medium. The wells were then loaded with cell suspensions of different types of normal and tumor cell lines. A total of 24 different cell types were incorporated into one array block. After loading the different cells, the array was stored at - 7 0 ~ C before sectioning to produce frozen

cell array slides. These slides could also be stored at - 7 0 ~ C until required when the cells were air dried at room temperature and fixed in acetone before proceeding to analysis using inmmnohisto- chemistry. They assessed expression of the human epithelial cell adhesion molecule, Ep-CAM, and found a reliable comparison be- tween the frozen cell array technique and the fluorescence-activated

cell sorting. Oode et al. (2000) describe another technique for manufacturing

a cell array. A glass slide was specially designed with 50 • 2-mm diameter spots. Cuhm'ed cell suspensions were prepared, and ap- proximately 1000 cells were placed on each spot and allowed to air dry. They then used the cell array successfully to measure nuclear DNA content using laser-scanning cytometry for DNA ploidy anal-

ysis. In this study, we describe a simple ahernative technique for pro-

ducing a cell line array that incorporates standard TMA technology using a "tissue microarrayer" (Beecher Instrmnents, Silver Spring,

185

186 WATERWORTH ET AL.

TABLE 1

DETAILS OF THE CELL LINES INCORPORATED INTO THE MICROARRAY"

Cell line Origin

MCF-7 Breast (pleural effusion) MCF-7 Clone 9 Tamoxifen-resistant derivative MCF-7 MCF-7 MMUI Tamoxifen-resistant derivative MCF-7 T47D Breast (pleural effusion) MDA-MD-231 Breast (pleural effusion) MDA-MB-435 Breast (pleural effusio@ ZR75.1 Breast (ascites) SkBr3 Breast (pleural eflhsion) BT20 Breast (primary tumour) HT29 Colorectal HRT18 Colorectal MiaPaCa Pancreatic Panc- 1 Pancreatic ST16 Stomach ST42 Stomach MK-N-28 Stomach KATO III Stomach SK-N-AS Neuroblastoma SKNSH Neuroblastoma SJRH-30 Rhabdomyosarcoma SJSA-1 Osteosarcoma TC32 Neuroepithelioma RT4 Bladder (transitional cell)

~ The origin of MDA-MB-435 has been questioned (Ellison et al., 2003).

MD). Our technique is simple, reliable, and effective, and we dem- onstrate one use of the array for analysis of protein expression using standard immunohistochemistry techniques. [3-Catenin is a mole- cule we have an interest in and is ideal for this study because it can have a role both in the nucleus and in the cell membrane and is also present in the cytoplasmic pool.

The cell line array was developed using a range of different can- cer cell lines. All cell lines were maintained in vendor-recom- mended complete media at 37 ~ C in a humidified atmosphere of 5% CO2 in air. Details are shown in Table I. Cell fidelity was confirmed by short tandem repeat (STR) profiling and validated against published STR profiles (Masiers et al., 2001). For each cell line, 75-em 2 flask of cultured cells at 80% confluence was washed in phosphate-buffered saline (PBS), removed frmn the flask with 0.25% (w/v) trypsin, resuspended in appropriate culture media, and centriiuged at 1000 x g for 5 min to form a cell pellet. The su- pernatant was aspirated, and the cells were fixed in 10% (v/v) for- mol saline for 10 nfin. The fixed cells were reeentrifuged and the supernatant discarded. Nutrient agar, 4%, was melted in a water bath and three drops added to the cell pellet to form a suspension. This cell suspension was quickly centrifuged at 1000 • g for 30 s to remove any air bubbles and condense the block. The block was then left to set at 4 ~ C before adding 10% (v/v) formol saline to dislodge the block and finally embedding in paraffin.

The hematoxylin and eosin (H&E) whole sections were initially obtained from the paraffin-embedded cell blocks to ensure an even distribution of the cells within the agar. A tissue microarrayer (Bee- eher Instruments) was used to take a 0.6-mm core of cells as guided by the H&E section. This core was then transferred into a premade hole in a recipient paraffin block at precise xy coordinates. Three cores were taken from each cell block, and marker cores of chicken

liver were distributed amongst the experimental cores to facilitate orientation when subsequently analyzing the array. When the array was complete, the block was heated to 45 ~ C, just below the melting point for the paraffin, to incorporate the cores into the wax. Sections were then cut from the block using a microtome onto adhesive mi- croscope slides (Superfrost Plus) to prevent core loss during sub- sequent processing. An average of 200 sections were cut from each block. Every 40th section was retained for an H&E stain to ensure that all the cores were present at this depth.

Sections were initially deparaffinized and rehydrated before the heat-mediated antigen retrieval (pressure cooking for 2 inin at full pressure in 10 mM citrate buffer, pH 6.0). The sections were in- cubated with the primary monoelonal antibody for [3-catenin (Clone 14, Transduction Laboratories, Oxford, UK) at 1 in 400 dilution in PBS for 1 h at room temperature. Antibody-antigen binding was detected using biotinylated rabbit anti-mouse inunnnoglobulin (Dako, Ely, UK) followed by peroxidase-eonjugated streptavidin- biotin complex (Dako). Sites of peroxidase activity were demonstrat- ed using diaminobenzidine-H202 solution for .5 min. Cell line TMAs were then scored for the presence or absence of membrane, cyto- plasmic, and nuclear ~-eatenin.

Cuhured cell lines were successfully incorporated into the an-ay with an even distribution of cells within each core and preservation of cell architecture, as shown in Fig. 1. A total of 16 of 23 cell

lines were positive for membrane and cytoplasmic [3-catenin (MCF- 7, MCF-7R, Clone 9, T47-D, ZR-751, BT20, HRT-18, HT29, ST16, ST42, MKN-28, KATOIII, SJSA-1, SJRH-30, RT4, SK-N-SH), and 6 of 23 were negative for membrane, cytoplasmic, and nuclear [3- catenin (NDA-MB-231, SkBra, MiaPaCa, Pane-l , SK-N-AS, TC32). One cell line (MDA-MB-4a5 breast carcinoma) was unseorable be- cause of complete core loss on processing of the slides. A total of 18 of 69 cores (26%) were lost in processing.

We have demonstrated a simple technique for analyzing multiple cell lines in a single experiment. Conventionally, cell culture is laborious and analysis limited to detection of one or two specific features from a given cell line on a single microscope slide. Al- though an investment of time is initially required in the preparation of cell pellets and in the construction of the array, in the long term it has considerable benefits. It is an excellent way of preserving cell lines, and the technique could be extended easily to incorporate precious primary cell cultures. Once made, the cell mieroarray al- lows rapid review of any new molecular markers as they are dis- covered, and it is an excellent screening tool to find suitable pos- itive (and negative) controls for further cell line work. Other appli- cations of the cell microarray could be to analyze cells treated with various chemicals for the analysis of epigenetic silencing, and the mxays would also be suitable to assess ribonucleic acid/DNA by in situ hybridization. Because the cell lines are paraffin embedded before constructing the array, they are very representative of par- affin-embedded clinical material, and therefore, immunohistoefiem- ieal and other protocols can be worked up using this resource with- out wasting valuable clinical material. Although loss of core profiles within the cut sections seems high at 26%, it is in keeping with other published rates of core loss for TMAs of between 15% and 33% (Schraml et al., 1999; Mucci et al., 2000; Richter et al., 2000). Inserting three cores from each cell line block is essential to reduce the incidence of losing a case completely because of such rates of core loss. The cores of cells may be lost in the initial sectioning of the block or in the antigen retrieval and washing steps of the ira-

CELL LINE ARRAY 1 8 7

PIG. 1. Cell cores O.O mm in diameter showing even distribution of cells within each core and preservation of cell architecture. (A) The TC32 cells negative for 13-catenin. (B) The HRT-18 cells positive for membrane and cytoplasmic 13-catenin as shown by the brown di-aminobenzidine (DAB) counterstain. Original magnification x l 0 . Insets •

munohis tochemis t ry technique. Another explanat ion is related to

the length of the core-containing cells that is initially inserted into

the array. Cores should ideally be of the same length with cells

throughout the core. Differences in cell distr ibution or inaccurate

p lacement may translate into an absence of cells on some sections,

as i l lustrated in Fig. 2, emphas iz ing the need for an H&E section

every 40th slide, or so, to check all the cores are present at this

depth.

In conclusion, we have developed a "cell microarray" technique

for analyzing muh ip le cell l ines in a single exper iment allowing

high- throughput , cost-efficient analysis .

FIG. 2. Effect of different core size on mieroan'ay quality. Ideally, cell pellet cores should be of the same length with cell uniformity. If cores are placed at different depths, this will translate into an absence of material on sectioning.

ACKNOWLEDGMENTS

We are grateful to colleagues in the Molecular Medicine Unit and Cancer Resem'ch UK Cancer Medicine Unit at St James's University Hospital, Leeds, for kindly donating cell lines for the array and to Sarah Burdall for excellent technical assistance. Work in the authors' laboratories is supported by York- shire Cancer Research and the Liz Dawn Fund.

REFERENCES

DeRisi, J.; Penland, L.; Brown, E O., et at. Use of a cDNA microarray to analyse gene expression patterns in human cancer. Nat. Genet. 14:457-460; 1996.

Duggan, D. J.; Bittner, M.; Chen, Y.; Meltzel, P.; Trent, J. M. Expression profiling using cDNA microarrays. Nat. Genet. 21:10-14; 1999.

Ellison, G.; Klinowska, T.; Westwood, R. F. R.; Docte~, E.; French, T.; Fox, J. C. Further evidence to support the melanocytic origin of MDA- MB-435. J. Clin. Pathol. 55:294-299; 2003.

Ginestier, C.; Charafe-Jauffret, E.; Bertucci, F., et at. Distinct and comple- mentary infomlation provided by use of tissue and DNA microarrays in the study of breast tmnour markers. Am. J. Pathol. 161:1223-1233; 2002.

Kononen, J.; Bubendoff, L.; Kallioniemi, A., et al. Tissue microarrays tbr high-throughput molecular profiling of turnout specimens. Nat. Med. 4:844~%7; 1998.

Masters, J. R.; Thomson, J. A.; Daly-Burns, B., et al. Short tandem repeat profiling provides an international referenee standard for hunmn cell lines. Proc. Natl. Acad. Sci. USA. 98:8012-8017; 2001.

Mucci, N. R.; Akdas, G.; Manely, S.; Rubin, M. A. Neuroendocrine expres- sion in metastatic prostate cancer: evaluation of high throughput tis- sue microarrays to detect homogenous protein expression. Hum. Pathol. 31:406414; 2000.

Oode, K.; Fui~aya, 31.; Harada, K.; Kawauchi, S.; Yamamoto, K.; Hirano, T.; Sasaki. K. The development of a cell alway and its combination with laser scanning cytometry allows a high-throughput analysis of nuclear DNA content. Am. J. Pathol. 157:723-728; 2000.

Richter, J.; Wagner, U.; Kononen, J., et al. High-throughput tissue micromTay analysis of cyclin E gene amplification and over expression in urinary bladder cancer. Am. J. Pathol. 157:787-794; 2000.

Schraml, E; Kononen, J.; Bubendoff, L., et ah Tissue microarrays for gene amplification surveys in many different tumour types. Clin. Cancer Res. 5:1966-1975; 1999.

Stephan, J. R; Schanz, S.; Wong, A.; Schow, E; Wong, W. L. T. Development of a frozen cell array as a high-throughput approach for cell based analysis. Am. J. Pathol. 161:787-797; 2002.