Biomarkers for simplifying HTS 3D cell culture platforms for drug discovery: the case for cytokines feature Drug Discovery Today �Volume 16, Numbers 7/8 �April 2011 PERSPECTIVE P e rs p e ct iv e � F E A T U R E Biomarkers for simplifying HTS 3D cell culture platforms for drug discovery: the case for cytokines Yinzhi Laia, Amish Asthanaa and William S. Kisaalita, williamk@engr.uga.edu In this review, we discuss the microenvironmental cues that modulate the status of cells to yield physiologically more relevant three-dimensional (3D) cell-based high throughput drug screening (HTS) platforms for drug discovery. Evidence is provided to support the view that simplifying 3D cell culture platforms for HTS applications calls for identifying and validating ubiquitous three-dimensionality biomarkers. Published results from avascular tumorigenesis and early stages of inflammatory wound healing, where cells transition from a two-dimensional (2D) to 3D microenvironment, conclusively report regulation by cytokines, providing the physiological basis for focusing on cytokines as potential three-dimensionality biomarkers. We discuss additional support for cytokines that comes from numerous 2D and 3D comparative transcriptomic and proteomic studies, which generally report upregulation of cytokines in 3D compared with 2D culture counterparts. Introduction It is now a well-accepted view that traditional flat-surface cell culture dishes do not adequately represent the natural context of the cell and their use in cell-based assays might, in some cases, result in less accurate drugs’ effects predictions [1–3]. Fig. 1a shows the possible cues that might affect the status of a cell, grouped along che- mical, physical, and spatial and/or temporal ‘dimensions’ [4,5]. Recently, numerous three- dimensional (3D) cell-based assay platforms have been commercially introduced with more contexts to better mimic the complex in vivo microenvironment. The premise behind these products is that mimicking the in vivo micro- environment yields physiologically more rele- vant cells that provide physiologically more relevant drug responses [6]. 1359-6446/06/$ - see front matter � 2011 Elsevier Ltd. All rights reser Several studies have provided evidence in support of this premise. For example, 3D multi- cellular tumorspheroids(MCTS)havebeen shown to mimic in vivo situations closely [7,8]; they were also able to recapitulate fully the cell adhesion- mediated drug resistance (CAM-DR) of EMT6 tumors [9], which was induced in vivo but was lacking in cultured monolayers of the same cells. Also, Weaver et al. [10] showed the phenotypic transformation of malignant cells in a 3D collagen matrix upon treatment with integrin antibodies; however, this phenomenon, also observed in vivo, was not observed in traditional monolayer cul- tures. In another study, which highlights the relevance of 3D culture format, two tumorigenic cell lines showed protease-independent amoe- boid movement within a 3D collagen, challenging the traditional screening for anti-metastatic ved. doi:10.1016/j.drudis.2011.01.009 agents against proteolytic activity with two- dimensional (2D) cultures [11]. In drug develop- ment, drug-induced liver injury or hepatotoxicity is a major deciding factor behind the approval, non-approval or limitation in the usage of the drug by the FDA. Usage of hepatoblastoma cell line (HepG2) cultured in 2D monolayers is the current gold standard in early screening of drug candidates for hepatotoxicity activity. However, cells in this format either entirely lack or express acutely low levels of many drug metabolizing enzymes [cytochrome P450s (CYPs)] and trans- porters found in hepatocytes in vivo [12]. By contrast, 3Dhepatocyte-likecellculturesachieved in bioreactors have been shown to emulate in vivo characteristics to such a high degree that they can be utilized to fabricate a functional bioartificial organ for transplantation [13,14]. www.drugdiscoverytoday.com 293 http://dx.doi.org/10.1016/j.drudis.2011.01.009 s t l p w w s b c h c h y w B S p o m e l [ n t t a o c c a 3 s PERSPECTIVE Drug Discovery Today �Volume 16, Numbers 7/8 �April 2011 [()TD$FIG] 3D Geometric property (Three-dimensionality) Cell–matrix adhesion (including polarity and mechanical inputs) Growth factors (cytokines and hormones) (a) Soft (b) Chemical matrix composition Physical matrix characteristics Mechanical stress and fluid flow Nutritional status (e.g. O2 and nutrients) Stiff 2D Autoregulation (Niche formation) Cell–cell junctions and communication (including polarity and mechanical inputs) Chemical S p a ti a l/ te m p o ra l (t o p o g ra p y a n d d if fe re n ti a ti o n ) (d) (e) (e .g . m at rix p ro pe rty an d fo rc e) Ph ys ic al (c) Drug Discovery Today FIGURE 1 Microenvironmental cues and cellular phenotype outcomes. (a) The possible cues that might affect the status of a cell. The major players are spatial and/or temporal, physical and chemical cues [4,5]. (b–e) Phalloidin-stained F-actin. (b, d) Show muscle cells cultured on soft and stiff flat surfaces, respectively [20]. (c, e) Show SKBR malignant breast cancer cells grown in 3D lrECM [34] and 2D glass surfaces [35], respectively. Scale bars = 50 mm (c) and 10 mm (d). 2 P e rsp e ctiv e � F E A T U R E An increase in studies highlighting the phy- iological relevance of the 3D culture format over raditional monolayers have formed the intel- ectual basis behind the founding of new com- anies such as 3DBiotek (http:// ww.3dbiotek.com), KIYATEC Inc. (http:// ww.kiyatec.com), CellAsic (http://www.cella- ic.com) and BellBrook Labs (http://www.bell- rooklabs.com), which have since launched ommercial 3D products for cell-based assay igh throughput screening (HTS). Examples of ompanies with HTS application products that ave been marketed for more than at least 3 ears include: Invitrogen Corporation (http:// ww.invitrogen.com/3D-cellculture), Glycosan ioSystems (http://www.glycosan.com) and urModics (http://www.Surmodics.com). More roducts are in the pipeline, increasingly relying n developments in microfabrication and icrofluidics technologies. For example, Huh t al. [15] successfully reconstituted organ-level ung function on a chip and Wang and Kisaalita 16] are developing a nanofibrous intercon- ected microwell platform for neural networks. Although there might be exceptions, intui- ively, the more complex the culture system is, he more cues it can capture and, thus, the more ble it is to mimic the in vivo microenvironment f the cell. However, as the complexity in the ulture system increases, so does the screening ost per compound. Product, cell maintenance nd readout instrumentation can cost more for D platforms. With the industry drive to reduce creening costs, an optimal balance between 94 www.drugdiscoverytoday.com cost and in vivo microenvironment emulation is needed. It might not be prudent to pursue a ‘perfect’ cell culture system that prices itself out of adoption in state-of-the-art screening laboratories. So the question that begs an answer is: among the three cue group dimen- sions outlined in the first paragraph, can any one of them carry more weight and, as such, con- stitute the major driver in designing less com- plex cell-based HTS platforms? To answer this question, we examine the microenvironment cues below in more detail. Microenvironmental cues Chemical cues can be subdivided into short- and long-range categories [3]. Short-range chemistry refers to the chemistry of the surface. To be able to emulate the in vivo phenotype, the approach adopted by most researchers is to present a substrate to the cell that chemically resembles the extracellular matrix (ECM). Long- range chemistry refers to diffusible chemical species that are found in the ECM that contri- bute to the maintenance or control of the cell behavior and/or phenotype outcome. The list includes basic nutrients, growth factors, cyto- kines and other morphogens, as well as metalloproteinases. Although their use in, for example, controlling differentiation is not well understood, long-range chemical cues are the easiest factor to manipulate, typically by including them in culture medium. A key attri- bute of chemical cues is that they can be orthogonal to the other two major cues; that is, they can be altered without affecting other properties of the platform. Physical cues of the material or scaffolding in or on which the cells grow and differentiate have been characterized in terms of softness, stiffness, pliability or Young’s modulus (terms used inter- changeably to express stiffness of an isotropic elastic material). When physical properties were first investigated with hydrogels, they were intertwined with chemical cues; high water content in the hydrogel makes the hydrogel softer, but simultaneously decreases the che- mical ligands density (i.e. the number of adhe- sion ligands for cell surface receptors per unit area of the substrate surface). To address this problem, researchers developed a novel hydro- gel system in which they were able to alter the stiffness of the hydrogel without altering ligand density [17]. This type of culture system made it possible to study the effect of the physical properties of the scaffold without interfering with the chemical effects. However, most studies on substrate physical properties have been conducted on flat surfaces. Initially, ‘3D cell cultures’ mainly referred to culture system scaffolds that constituted com- plex architecture at micro and/or nano scales, which included only the spatial context. The term now has a more comprehensive meaning that entails the three ‘dimensions’ outlined in Fig. 1a. However, closer examination reveals that spatial cues are highly intertwined with the other cues. For example, by introducing micropores and/or nanofibers into the cell culture platform, http://www.3dbiotek.com/ http://www.3dbiotek.com/ http://www.kiyatec.com/ http://www.kiyatec.com/ http://www.cellasic.com/ http://www.cellasic.com/ http://www.bellbrooklabs.com/ http://www.bellbrooklabs.com/ http://www.invitrogen.com/3D-cellculture http://www.invitrogen.com/3D-cellculture http://www.glycosan.com/ http://www.surmodics.com/ onmental factor to achieve more platform Drug Discovery Today �Volume 16, Numbers 7/8 �April 2011 PERSPECTIVE P e rs p e ct iv e � F E A T U R E the surface area of the construct can increase significantly, thus increasing the amount che- mical ligands that can be deposited on it. The thickness of the porous or fibrous scaffolds can also affect the concentration of the culture medium by generating a gradient of long-range chemical cues, such as oxygen and growth fac- tors. Additionally, the porous and fibrous archi- tecture might alter the physical properties of the material, reducing its strength and altering the anisotropism of the material. For example, the strength of fibrous scaffolds along the fiber direction is higher than the direction that is perpendicular to the fiber, and the strength in both directions is lower than the bulk property of the material [18]. Mathematical models have been used to study chemical gradients induced by 3D cell cultures (e.g. see [19]). However, few 3D cell culture studies have included physical property measurements in their studies. It is true that architecture and/or geometry have a profound impact on cell behavior. To be able to engineer a platform that more accurately emulates the in vivo microenvironment, it is important to isolate the contributions of the different cues to the phenotypic outcome. Consider morphology for example; the right panel of Fig. 1 shows F-actin staining from muscle cells on soft (Fig. 1b) and stiff (Fig. 1d) flat surfaces [18], SKBR malignant breast cancer cells in 3D laminin-rich extracel- lular matrix (lrECM) (Fig. 1c) [19] and on glass 2D surfaces (Fig. 1e) [20]. The soft surface and 3D cell cultures can both make cells more roundish and smaller in smaller size when compared with the same cells grown on rigid 2D surfaces. The F- actin organization is also similar between soft surfaces and 3D cell cultures: in both cases, the F- actin does not form stress fibers and is more abundant at the edge of the cells. Why do two different factors have similar effects on a cellular outcome? The Young’s modulus for lrECM (also called Matrigel in other studies) ranges from 10 Pa to 50 Pa as concentration changes from 50% to 100% [21]. In addition, the 3D lrECM culture had a lrECM coating as thick as 500 mm (thick in comparison to the typical cell size of 50 mm). It is reasonable to speculate that the cells cultured in 3D lrECM interacted with a soft tissue-mimicking microenvironment and the morphological changes observed might have resulted mainly from pliability changes. In many 3D culture systems, cells cluster together and form multicellular aggregates [22–24]. Under such circumstances, most individual cells within the multicellular spheres are exposed to their neighboring cell surfaces, which are soft. Therefore, it is possible that the influence of the material physical properties in such cases is limited, which raises the question; if softness is all a cell requires to become round and subse- quently mimic its in vivo counterpart, why is there a need to create a complex 3D architec- ture? Alternatively, the rigidity effect of the material has been bypassed by simple fabrica- tion of micropost arrays (0.97–12.9 mm) onto a polydimethylsiloxane (PDMS) material; the rigidity of the altered material is reflected by the resulting 3D-like cellular morphology from cells cultured on the highest micropost substrate [25]. Above, we have briefly summarized current understanding of microenvironmental cues. Unfortunately, this understanding is not suffi- cient to guide the design of the simplest plat- form that provides the desired emulation of the in vivo situation. A common practice is the trial- and-error approach, in which the platform designer focuses on a complex physiological relevance (CPR) outcome as an indication of the capture of one or several cues sufficient for the desired outcome. CPR here means in vitro emulation of in vivo structure and/or function in 3D that is not possible in 2D cultures, such as bile canaliculi-like structures [26] and albumin secretion [27] by hepatocytes. There are many cells whose phenotypes are altered by culture in 3D structures but, unlike hepatocytes, their CPR is not readily discernible. For such cells, as the platform architecture is simplified, it is difficult to determine the simplest form that maintains the desired emulation of the in vivo situation. There is a need for simple technology to establish the culture status trajectory that results in formation of three-dimensionality that leads to CPR. Such a technology will enable the establishment of a balance between a simplistic architectural design and the complex microenvironmental features. As mentioned above, some cells might not require a complex platform to grow in or on; rather, they might initially need the minimal essential cues to become competent enough to produce endogenous ECM, which then provides the remaining or additional essential cues. This has been observed in several studies; for example, tumor cells grown in porous 3D poly(lactide-co-glycolide) (PLG) and Matrigel scaffolds showed upregulation of interleukin (IL)- 8 compared with 2D cultures [28]. Even though both systems have a varying degree of com- plexity and provide different microenviron- mental cues, the outcome (elevation in IL-8 level) was similar and physiologically justified as it depicts the angiogenic capability of the cells. Alternatively, IL-8 could have been added to the culture systems as a biochemical microenvir- complexity; however, it would have been redundant as the cells can produce it themselves if the essential initial minimal factors are present. This example underscores the need to better understand the required essential cues and their thresholds that yield CPR or in vivo emulation outcomes. This will enable simplification of the platform without giving up the physiologically relevant behavior of the cells. However, the task of specifying such complex microenvironments is daunting. A simpler approach is to generalize the measurement of the developmental com- petence that leads to CPR outcomes in terms of ubiquitous three-dimensionality biomarkers [3]. As the platform architecture design is simplified, it will be possible, with such a measure, to know when the trajectory toward CPR outcomes is being compromised by the introduced simpli- fication(s). To address the question of what the biomar- kers might be, we reviewed the current 3D culture literature with respect to finding che- mical entities that are produced differentially between 2D (minimal or no in vivo-like micro- environment) and 3D (more in vivo-like micro- environment providing CPR) systems across a variety of cells derived from different tissue types and 3D architectural platforms. This search indicates cytokines as the most probable family of compounds to provide the badly needed biomarkers. We summarize the arguments below for the case for cytokines. Searching for three-dimensionality biomarkers Cytokines are soluble, low molecular weight, extracellular protein mediators that usually act at short range between neighboring cells. They are crucial intercellular regulators and mobilizers of cells engaged in innate and adaptive inflam- matory host defenses, cell growth, differentia- tion, cell death, angiogenesis, and development and repair processes. They have been assigned to various family groups based on the structural homologies of their receptors and can be broadly classified into families such as colony stimulating factors, interleukins, interferons, transforming growth factors (TGF), tumor necrosis factors (TNF), platelet-derived growth factors (PDGF) and chemokines. Although cytokines have been extensively studied in the field of immunology and oncology, they have been overlooked by tissue engineers. However, these same small proteins might have the power to revolutionize the field. Although the evidence for their existence in 3D cultures is compelling, they have not yet been looked at as candidates for potential biomarkers. However, we suggest www.drugdiscoverytoday.com 295 PERSPECTIVE Drug Discovery Today �Volume 16, Numbers 7/8 �April 2011 [()TD$FIG] 140 120 100 80 60 40 20 0 900 800 700 600 500 400 300 200 100 0 2D, RGD- alginate 3D, alginate 3D, RGD- alginate 2D, RGD- alginate 3D, alginate 3D, RGD- alginate V E G F ( p g /1 0 0 0 0 c e ll s /2 4 h ) IL -8 ( p g /1 0 0 0 0 c e ll s /2 4 h ) * * * * (a) (b) Drug Discovery Today FIGURE 2 Cytokine production by tumor cells (oral squamous cell carcinoma, OSCC-3) in alginate hydrogels containing covalently coupled RDG peptides at a density that correlates to the number of RDG sites in tumors in vivo. The 3D system resulted in enhanced IL-8 production (a); however, this was not the case for VEGF (b). (*, P � 0.05; **, P � 0.01; black and gray bars represent 5 and 10 days in culture, respectively). Adapted, with permission, from [33]. P e rsp e ctiv e � F E A T U R E that they are the ideal family to explore for identification and validation follow-up studies for the following three main reasons. First, when cells are transitioned from a 2D monolayer to a 3D microenvironment, they become surrounded by homotypic neighbors, forming a loosely bound disorganized aggre- gate. In vivo, such a scenario is encountered only during avascular tumorigenesis or early stages of inflammatory wound healing, which are both similar in nature and regulated by cytokines [29]. Therefore, in vitro, the cells growing in 3D relate to any of those two models, depending on their type (malignant or primary, respectively), therefore explaining the upregulation of their cytokine levels. Second, several 2D–3D comparative tran- scriptomic studies with cells from the four main tissue types (nerve, muscle, connective and epithelial) cultured on a variety of platforms, have reported the upregulation of cytokines and their receptors in 3D cultures. For example, Klapperich and Bertozzi [30] reported upregu- lation of seven cytokines [IL-8, chemokine (C-X-C motif) ligand (CXCL)-1, CXCL2, CXCL3, CXCL5, vascular endothelial growth factor (VEGF) and leukemia inhibitory facto (LIF)] by a human fetal lung fibroblast (IMR-90) cultured in a collagen– glycosaminoglycan (collagen/GAG) 3D mesh. The mesh was prepared by freeze drying and heat enabled cross-linking of the polymer and had an average pore size of 80–100 mm. In addition, Ghosh et al. [31] reported upregulation of six cytokines [CXCL1–3, IL-8, macrophage Inflammatory Protein-3 (MIP-3a) and angiopoe- tin-like 4] by a melanoma cell line (NA8) cultured 296 www.drugdiscoverytoday.com on poly-2-hydroxyethyl methacrylate (poly- HEMA) plates when compared with 2D surfaces. The polymer coating prevented cells from attaching to the plastic surface leading to the formation of MCTS. Third, transcriptomic findings such as those in the above examples have been supported by studies at the protein level. For example, Enzerink et al. [32] has shown that clustering of fibroblasts induces chemokine (CCL2-5, CXCL1-3 and CXCL8) secretion in five different fibroblast cell lines cultured in agarose. In addition, Fischbach et al. [33] cultured tumor cells in a 2D and 3D RGD-alginate system and reported a dramatic increase in IL-8 levels; however, no significant VEGF differences were reported between 2D and 3D cultures (Fig. 2). In the same study, cells grown in alginate gels having RGD peptides (spatial, biophysical and biochemical cues) produced a higher level of IL-8 compared with those lacking the adhesion peptide (only spatial and biophysical cues), although both produced higher levels when compared with 2D (lacking all three cues). This shows that changes in the microenvironment are conveyed directly by the difference in the level of cytokine pro- duction. In another study by the same group, the same cells also showed an upregulation of cytokines when grown in Matrigel (lrBM) com- pared with their 2D counterparts. This compar- ison is particularly important as cells grown on Matrigel have already been shown to produce a CPR outcome (formation of mammary gland acinus and milk-like secretions into the lumen) [10], which mean that Matrigel provides the relevant microenvironmental cues. Taken together, these studies suggest that the upre- gulation of cytokines in 3D compared with 2D cultures is not a random differential response but a potentially ubiquitous biomarker for three- dimensionality. Further studies to establish this view firmly are needed. Cytokines are particularly attractive as bio- markers for several practical reasons. First, they are secreted in the media, making it easy for their detection to be amenable for HTS readout. Second, they are expressed in a wide range of cells from the four tissue types (muscle, con- nective, epithelial and nerve), suggesting the potential for their ubiquity as opposed to being cell or tissue specific. Third, their temporal expression suggests the use of profiles as opposed to single-time measurements, which increases their robustness as biomarkers. Concluding remarks Simplifying 3D cell culture platforms for HTS applications calls for the identification and validation of ubiquitous three-dimensionality biomarkers. Our review of comparative tran- scriptomic and proteomic literature strongly suggests the cytokine family as having the greatest potential to yield compounds that can serve as three-dimensionality biomarkers. How- ever, narrowing the candidates to a practical few still remains to be done. References 1 Weaver, V.M. et al. (1997) Reversion of the malignant phenotype of human breast cells in three-dimensional culture and in vivo by integrin blocking antibodies. J. Cell Biol. 137, 231–245 Drug Discovery Today �Volume 16, Numbers 7/8 �April 2011 PERSPECTIVE F E A T U R E 2 Justice, B.A. et al. (2008) 3D cell culture opens new dimensions in cell-based assays. Drug Discov. Today 14, 102–107 3 Kisaalita, W.S. (2010) 3D Cell-Based Biosensors in Drug Discovery Programs: Microtissue Engineering for High Throughput Screening. CRC Press 4 Yamada, K.M. et al. (2003) Dimensions and dynamics in integrin function. Braz. J. Med. Biol. Res. 36, 959– 966 5 Yamada, K.M. and Cukierman, E. (2007) Modeling tissue morphogenesis and cancer in 3D. Cell 130, 601–610 6 Pampaloni, F. et al. (2007) The third dimension bridges the gap between cell culture and live tissue. Nat. Rev. Mol. Cell Biol. 8, 839–845 7 Mueller-Klieser, W. (2000) Tumor biology and experimental therapeutics. Crit. Rev. Oncol. Hematol. 36, 123–139 8 Kunz-Schughar, L.A. et al. (2004) The use of 3D cultures for high-throughput screening: the multicellular spheroid model. Soc. Biomol. Screen. 9, 273–285 9 Kobayashi, H. et al. (1993) Acquired multicellular- mediated resistance to alkylating agents in cancer. Proc. Natl. Acad. Sci. U. S. A. 90, 3294–3298 10 Weaver, V.M. et al. (1997) Reversion of the malignant phenotype of human breast cells in three-dimensional cultures and in vivo by integrin blocking antibodies. J. Cell Biol. 137, 231–245 11 Wolf, K. et al. (2003) Compensation mechanism in tumor cell migration: mesenchymal-amoebic transition after blocking of pericellular proteolysis. J. Cell Biol. 160, 267–277 12 Wilkening, S. et al. (2003) Comparison of primary hepatocytes and hepatoma cell line HepG2 with regard to their biotransformation properties. Drug Metab. Disp. 31, 1035–1042 13 Morsiani, E. et al. (2001) Long-term expression of highly differentiated functions by isolated porcine hepatocytes perfused in a radial-flow bioreactor. Artif. Organs 25, 740–748 � 14 Miyashita, T. et al. (2000) Development of a bioartificial liver with glutamine synthetase-transduced recombinant human hepatoblastoma cell line, HepG2. Transplant Proc. 32, 2355–2358 15 Huh, D. et al. (2010) Reconstituting organ-level lung functions on a chip. Science 328, 1662–1668 16 Wang, L. and Kisaalita, W.S. (2010) Characterization of micropatterned nanofibrous scaffolds for neural network activity for high-throughput screening. J. Biomed. Mater. Res. B 94, 238–249 17 Saha, K. et al. (2008) Substrate modulus directs neural stem cell behavior. Biophys. J. 95, 4426–4438 18 Baker, S.C. et al. (2006) Characterization of electrospun polystyrene scaffolds for three-dimensional in vitro biological studies. Biomaterials 27, 3136–3146 19 Griffith, L.G. and Swartz, M.A. (2006) Capturing complex 3D tissue physiology in vitro. Nat. Rev. Mol. Cell Biol. 7, 211–224 20 Discher, D.E. et al. (2005) Tissue cells feel and respond to the stiffness of their substrate. Science 310, 1139–1143 21 Zaman, M.H. et al. (2006) Migration of tumor cells in 3D matrices is governed by matrix stiffness along with cell- matrix adhesion and proteolysis. Proc. Natl. Acad. Sci. U. S. A. 103, 10889–10894 22 Albrecht, D.R. et al. (2006) Probing the role of multicellular organization in three-dimensional microenvironments. Nat. Methods 3, 369–375 23 De Bank, P.A. et al. (2007) Accelerated formation of multicellular 3-D structures by cell-to-cell cross-linking. Biotechnol. Bioeng. 97, 1617–1625 24 Wang, L. et al. (2009) SU-8 microstructure for quasi- three-dimensional cell-based biosensing. Sens. Actuators 140, 349–355 25 Fu, J. et al. (2010) Mechanical regulation of cell function with geometrically modulated elastomeric substrates. Nat. Methods 7, 733–736 26 Abu-Absi, S.F. et al. (2002) Structural polarity and functional bile canaliculi in rat hepatocytes spheroids. Exp. Cell Res. 274, 56–67 e 27 Kane, B.J. et al. (2006) Liver-specific functional studies in microfluidic array of primary mammalian hepatocytes. Anal. Chem. 78, 4291–4298 28 Fischbach, C. et al. (2007) Engineering tumors with 3D scaffolds. Nat. Methods 4, 855–860 29 Coussens, L.M. and Werb, Z. (2002) Inflammation and cancer. Nature 420, 860–867 30 Klapperick, C.M. and Bertozzi, C.R. (2004) Global gene expression of cells attached to a tissue engineering scaffold. Biomaterials 25, 5631–5641 31 Gosh, S. et al. (2005) Three-dimensional cultures of melanoma cells profoundly affect gene expression profile: a high density oligonucleotide array study. J. Cell. Physiol. 204, 522–531 32 Enzerink, A. et al. (2009) Clustering of fibroblasts induces proinflammatory chemokines secretion promoting leukocyte migration. Mol. Immunol. 46, 1787–1795 33 Fischbach, C. et al. (2009) Cancer cell angiogenic capability is regulated by 2D culture and integrin engagement. Proc. Natl. Acad. Sci. U. S. A. 106, 399–404 34 Lee, G.Y. et al. (2007) Three-dimensional culture models of normal and malignant breast epithelial cells. Nat. Methods 4, 359–365 35 Stofega, M.R. et al. (2004) Constitutive p21-activated kinase (PAK) activation in breast cancer cells as a result of mislocalization of PAK to focal adhesions. Mol. Biol. Cell 15, 2965–2977 Yinzhi Lai, Amish Asthana, William S. Kisaalita Cellular Bioengineering Laboratory, Faculty of Engineering, Driftmier Engineering Center, University of Georgia, Athens, GA 30602, USA www.drugdiscoverytoday.com 297 P e rs p e ct iv Biomarkers for simplifying HTS 3D cell culture platforms for drug discovery: the case for cytokines Introduction Microenvironmental cues Searching for three-dimensionality biomarkers Concluding remarks References