Sage Journals: Discover world-class research

Abstract

Tissue dynamics spectroscopy combines dynamic light scattering with short-coherence digital holography to capture intracellular motion inside multicellular tumor spheroid tissue models. The cellular mechanical activity becomes an endogenous imaging contrast agent for motility contrast imaging. Fluctuation spectroscopy is performed on dynamic speckle from the proliferating shell and hypoxic core to generate drug–response spectrograms that are frequency versus time representations of the changes in spectral content induced by an applied compound or an environmental perturbation. A combination of 28 reference compounds and conditions applied to rat osteogenic UMR-106 spheroids generated spectrograms that were crosscorrelated in a similarity matrix used for unsupervised hierarchical clustering of similar compound responses. This work establishes the feasibility of tissue dynamics spectroscopy for three-dimensional tissue-based phenotypic profiling of drug response as a fully endogenous probe of the response of tissue to reference compounds.

Keywords

multicellular tumor spheroids dynamic light scattering digital holography intracellular motility apoptosis and necrosis drug response spectrograms hierarchical clustering label-free endogenous coherence gating optical coherence imaging motility contrast imaging tissue dynamics spectroscopy speckle holography multi-parameter phenotypic profiling

Three-dimensional Tissues for Screening

Living tissue is a complex system that responds to its environment in complex ways. In this sense, living tissue is a biosensor that can provide a multiparameter phenotypic profile of a pharmaceutical. For instance, multicellular tumor spheroids^1,2 are three-dimensional (3D) living tissue, which allow cells to exist in a natural 3D environment with normal cell–cell contacts and contacts to extracellular matrix (ECM). These tissues are more likely to display natural intercellular signaling and, hence, respond more naturally to applied pharmaceuticals. It is now known that cells in contact with flat hard surfaces do not behave the same as cells embedded in the ECM.^3,4 In response to this, a major conceptual and experimental shift has occurred in recent years away from two-dimensional (2D) cell cultures to 3D tissue cultures.^{5

–8} Despite the advantages, one obstacle to wide adoption of the tumor spheroid model has been the difficulty to extract meaningful phenotypic responses from deep inside, especially in a high-content manner and without the need for labels.

Light scattering from tissue has become an important probe of cellular structure and dynamics.^9,10 Probing tissues with light is label free, nonionizing, and low cost. The challenge for light scattering from tissues has been how to extract high-content and context information from scattered light that can compete with the highly specific optical signals from fluorescence techniques. This has been partially achieved using low-coherence enhanced backscatter^11,12 that probes subtle structural changes in tissue layers. Light scattered from a selected depth can be partially localized by applying a coherence gate on the scattered light^13,14 as in optical coherence tomography.^15,16 Full-field coherence gating was demonstrated using dynamic holography^17,18 and was applied to multicellular tumor spheroids¹⁹ and the mouse eye.²⁰ Digital holography²¹ has been applied in the single-scattering regime with numerical focusing to study the refractive index profiles of cells^22,23 and in the multiple-scattering regime as a coherence gate to study the structure of tumors and effect of antimitotic drugs on tissue,²⁴ which is the approach we describe and use in this article.

Dynamic light scattering (DLS) in tissues falls into two regimes: quasi-elastic light scattering when light is predominantly singly scattered²⁵ and diffusing-wave spectroscopy^{26
–28} when light is multiply scattered. The transition from single scattering to multiple scattering is particularly important for backscatter applications,^{29

–32} such as in the work presented in this article. Coherence-gated DLS lies between the regimes of single and multiple scatterings. The coherence gate preferentially selects singly scattered ballistic photons.^29,32,33 Coherence gating has been applied to DLS to provide 3D information on the intracellular motions of cells.³⁴ This is the principle of motility contrast imaging (MCI), which has been applied to living tissue and used to study the effects of antimitotic cytoskeletal drugs on tissue dynamics.²⁴

In this article, we present holographic MCI and tissue dynamics spectroscopy (TDS) applied to multicellular tumor spheroids. These two related techniques use volumetric imaging based on DLS to extract high-content information from the active motion occurring inside the cells of the spheroids. MCI spatially resolves and quantifies the amount of motion in different parts of the tissue, whereas TDS performs fluctuation spectroscopy to identify the microscopic mechanisms of motion. Our specific approach is to generate drug–response spectrograms of living tissue responding to xenobiotic and environmental perturbations. These drug–response spectrograms display the changes in the fluctuating frequency content of the DLS as a function of time after the dose—much like voiceprint spectrograms. Different drugs elicit different drug–response spectrograms. The different features in the spectrograms can be related back to phenotypic effects of the drugs on the tissue. We have obtained drug–response spectrograms for several pharmaceutical compounds. By correlating the spectrograms, classes of drugs with common mechanisms can be identified.

Multicellular Tumor Spheroids

Multicellular spheroids of normal cells or neoplastic cells are balls of cells that may be easily cultured up to 1 mm in size in vitro.^{35

–40} The spheroids can be used to simulate the histological and metabolic features of small nodular tumors in the early avascular stages of growth⁴⁰ and may be used to simulate the optical properties of a variety of tissues such as the epidermis and various epithelial tissues.

Beyond a critical size (about 200 μm), most spheroids develop a necrotic core surrounded by a shell of viable proliferating cells, with a thickness varying from 100 to 300 μm. The development of necrosis has been linked to deficiencies in the metabolites related to energy generation and transfer. The limiting factor for necrosis development is oxygen consumption and transport that reflects the status of the spheroid.^36,41 Early work on spheroids⁴² launched the study of therapeutic strategies for cancer, especially the spheroid response to different drugs. The response to drug therapy was quantified from an analysis of spheroid volume growth delay, increase in the necrotic area, and change in survival capacity. This work focused on hypoxia and its induction by chemical agents.⁴³

Tumor spheroids of permanent cell lines offer a reliable model for systematic study of tumor response to therapy.^38,44 In vitro monitoring of tissue response to drugs is an area of strong interest to pharmaceutical companies.⁴⁵ Although the in vitro environment is artificial, the biochemistry, metabolism, and cell signaling response of cells grown as 3D constructs closely simulates in vivo tissue^{46

–51} that would more accurately simulate in vivo growth of tumor cells in terms of their pathophysiology and response to therapy.^{52

–55} This is in part because genomic profiles are preserved in spheroids but not in primary monolayer cultures.^{56
–58} There have been several comparative transcriptomic studies that have tracked the expression of genes associated with cell survival, proliferation, differentiation, and resistance to therapy that are expressed differently in spheroids versus 2D cultures. For example, this was observed for cell lines from epithelial ovarian cancer,^59,60 hepatocellular carcinoma,^{61
–63} or colon cancer⁶⁴ in which the expression profiles from the spheroids were more like those from tumor tissues. In addition, the 3D environment of the spheroids presents different pharmacokinetics than 2D monolayer culture and produces differences in cancer drug sensitivities between the monolayers and the spheroids.^{65

–68} Therefore, in vitro experiments are a validated (and inexpensive) surrogate for in vivo response and have been used to test cancer therapeutics, such as metabolic and chemical gradients, hypoxia, cell–cell and cell–matrix contacts, and chemoresistance.^{43,69

–73}

Although microscopic imaging of cellular motility and motility-related gene expression is well established in 2D,⁷⁴ none of these studies considered cellular and subcellular motility as a diagnostic of cellular vitality, despite the obvious utility of this diagnostic, because there was no means of detecting motility nondestructively throughout a volume. MCI²⁴ provides this capability, and we have obtained motility information up to a millimeter deep in tumor spheroids.

Holographic Optical Coherence Imaging

Holographic TDS is the latest stage in an evolution of techniques that began with holographic optical coherence imaging (OCI), which is a form of full-field optical coherence tomography.^75,76 This soon progressed to MCI that spatially resolved the autocorrelation functions and used overall motion as a novel contrast agent.^24,34 Now it has evolved to TDS⁷⁷ that resolves overall motion into at least three broad frequency ranges with a tentative assignment of microscopic intracellular motion mechanisms.

The optical system for holographic OCI is shown in Figure 1A. We used a mode-locked Ti:sapphire laser (100-fs pulse duration and 100-MHz repetition rate) with a center wavelength of 840 nm and a bandwidth of 17 nm. The lens performs the Fourier transform of the object beam. The charge-coupled device (CCD) camera with a 7.4 mm pixel pitch is placed at the Fourier plane where the object beam interferes with the zero-path-matched reference beam that passes through the computer-controlled delay line. The typical object intensity for living tissue at the object plane is 5 mW/mm², and an 8-bit CCD camera with one-megapixel resolution was used with an exposure time of 10 ms. Digital holograms were reconstructed by fast Fourier transform. An example of a 3D reconstructed tumor spheroid is shown in Figure 1B and C. The full tumor is shown in Figure 1B resting on the Petri dish. The data in Figure 1C are eroded to show only the brightest reflections. These come from the optically heterogeneous necrotic core, which casts a “shadow” on the Petri dish.

Figure 1.

(A) Holographic OCI system and reconstruction of a multicellular tumor spheroid from OCI data. The 3D reconstructions at the bottom show all the data (B), or the high-reflectance voxels (C) that display the necrotic core and the shadow the core casts on the Petri dish.

Holographic OCI is based on off-axis digital holography using a CCD camera. The nominal speckle size on the CCD chip is approximately six pixels with approximately three optical fringes per speckle using a reference beam crossing angle of 3° at a wavelength of 840 nm. The intensity I _H(x′, y′) is electronically captured by the CCD camera. A simple Fourier transform is used to reconstruct the object from the digital hologram. The Fourier transform of the intensity is

\begin{matrix} F T (I_{H}) = F T ({| ψ_{R} |}^{2}) + F T ({| ψ_{O F} |}^{2}) + F T (ψ_{R} * ψ_{O F}) + F T (ψ_{R} ψ_{O F^{*}}) \\ = F_{1} + F_{2} + F_{3} + F_{4} \end{matrix}

(1)

where the third term produces a holographic image, the fourth term produces a conjugate holographic image, and the first and second terms produce a zero-order image. The third and fourth terms in Eq. (1) are

F_{3} (v_{x^{'}}, v_{y^{'}}) = i λ f ψ_{RO} exp (- \frac{i 4 π f}{λ}) ψ_{0} (- λ f v_{x^{'}} + λ f v_{x 0}, - λ f v_{y^{'}} + λ f v_{y 0})

(2)

F_{4} (v_{x^{'}}, v_{y^{'}}) = - i λ f ψ_{RO} exp (\frac{i 4 π f}{λ}) ψ_{0}^{*} (λ f v_{x^{'}} + λ f v_{x 0}, λ f v_{y^{'}} + λ f v_{y 0}) .

(3)

The holographic image F ₃ is located at λfv_x ₀, λfv_y ₀, and the conjugate image F ₄ is located at − λfv_x ₀, − λfv _y0 on the image plane, which are spatially separated from the zero-order image. The numerical Fourier transformation of the digital hologram acquired at the Fourier plane provides both amplitude and phase information of the multiply scattered light from the coherence-gated object depth.

To remove the zero-order image, we approximate a digital hologram without interference by averaging the interference fringes over the fringe spacing interval. The fringe spacing for our crossing angle is approximately three pixels. Therefore, the digital hologram without interference is approximated by three-pixel averaging in the horizontal direction. Most of the zero-order images are removed after this zero-order suppression, the data are more stable, and it is less time consuming than acquiring a non-zero-path-matched image for zero-order subtraction.⁷⁶

The structure reconstructed in 3D is not static in living tissue, but is highly dynamic, reflecting the multitude of motions that occur on the subcellular scale. These motions include organelle transport by molecular motors, active motions of the membrane, membrane shape changes and restructuring of cell–cell and cell–ECM contacts, mitosis and cytokinesis, and necrosis and apoptosis. These motions span a range of nanometers per second to micrometers per second and cause dynamic shifts in the speckle from scattered light in the range of sub-Hertz to tens of Hertz. By analyzing the statistics of the fluctuating laser speckle during OCI, we are able to quantify the total amount of motion and use this measure as a novel contrast agent. The resulting images are called motility contrast images, and examples are shown in Figure 2. The sections are color coded by the total amount of motion, and the sections are shown for several selected depths from the top of the tumor through the midsection to the bottom. In the midsection, the motility metric provides a clear imaging contrast between the proliferating outer shell (with high degree of motion) and the necrotic core. The 3D reconstruction of the motion is displayed as a cutaway in Figure 2B.²⁴

Figure 2.

(A) MCI of an 800-μm-diameter tumor. The sections are color coded to total motion, displaying the proliferating outer shell and necrotic core. (B) The 3D MCI image that is a composite of successive sections.

Tissue Dynamics Spectroscopy

MCI provides a single metric value for the mechanical activity of a 3D voxel, which is used as the imaging contrast for Figure 2. However, the fluctuating speckle intensities have time–frequency information that is extracted by Fourier analysis into fluctuation power spectra. An example of power spectra from a tumor spheroid is shown in Figure 3 for the proliferating shell and hypoxic core of a 500-μm-diameter tumor. The shell power spectrum shows a distinct knee frequency near 0.1 Hz, but the hypoxic core appears more like 1/f noise down to the noise floor at the Nyquist sampling frequency of 5 Hz. The spectra have two to three orders of magnitude of detection dynamic range across three orders of magnitude of frequency.

Figure 3.

Fluctuation power spectra of a tumor spheroid for the proliferating outer shell and necrotic core. The shell shows higher frequency content and a clear knee frequency. The necrotic core displays characteristic 1/ f behavior. Both spectra converge on a noise floor at the Nyquist frequency at 5 Hz.

The power spectrum for TDS is

S (ω) = FT (exp (- q_{eff}^{2} Δ x^{2} (τ)))

(4)

where FT stands for Fourier transform, q _eff ≈ 4π/λ is the light scattering parameter, and Δx ² is the mean-squared displacement. The normalized spectral power density can be expressed approximately as

S (f) = \frac{1}{1 + {(\frac{f}{f * (β^{'})})}^{1 + β^{'}}}

(5)

where f*(β′) is a characteristic frequency that depends on the exponent β′ that is closely related to, but not identical with, the anomalous diffusion exponent β. The value for β′ is 0.2 for the proliferating shell and close to zero for the hypoxic core, possibly reflecting weaker diffusion in the core. However, experimental spectral power densities usually show important deviations from Eq. (5) because many contributions to the DLS overlap in frequency to produce the approximately 1/f behavior. Also, there is a notable deviation near the Nyquist sampling frequency at which higher frequency motion can be “freeze framed” or “strobed” by a fast sampling time (wide detection bandwidth) that is much shorter than the Nyquist period.⁷⁸

When perturbations are applied to a living tissue sample, the response of the tissue dynamics can be subtle, but consistent. Because both the autocorrelation and spectral density functions span a wide dynamic range over several orders of magnitude, small changes in dynamics are de-emphasized in logarithmic plots. To capture the time course of subtle changes in tissue dynamics requires a differential relative measure, which we take to be the differential relative spectrum, called a spectrogram.

D (z; ω, t) = \frac{S (z; ω, t) - S (z; ω, 0)}{S (z; ω, 0)}

(6)

This differential spectrogram is a 2D function of frequency and time at fixed depth z, which captures the changes in the spectral power density as a function of time normalized by an appropriate spectral density, usually taken as the spectrum before the perturbation. The differential spectrogram shows positive and negative deviations from the nominal values across a wide frequency range.

As examples, two spectrograms are shown in Figure 4. The top spectrogram is for the metabolic drug iodoacetate, and the bottom spectrogram is for the antiactin drug cytochalasin D. Frequency is along the vertical axis, and time is along the horizontal axis, with the drug applied at t = 0. The color of the spectrogram is proportional to the quantity in Eq. (6). These drug–response spectrograms are like voiceprints, showing the frequency content as a function of time. Because they are differential, only the changes in the frequency content are shown. The baseline is set before the application of the drug.

Figure 4.

Drug–response spectrograms obtained from TDS for iodoacetate (top) and cytochalasin D (bottom). These are differential normalized spectrograms of Eq. (6). The baseline is set before the application of the drug, and only the changes in the fluctuation frequencies are displayed in color (red: enhanced motion; blue: suppressed motion). Note the striking differences in the spectrogram signatures for these two drugs.

To interpret the spectrograms, it is necessary to establish which structural features and which dynamical mechanisms are involved in observed changes in the various spectral bands. The lowest frequency in our experimental spectrograms is 0.005 Hz, and the highest frequency is 5 Hz. To make the connection between detection frequency and motion, the general relationships for single backscattering are q ² D = ω_D for diffusion and qv = ω_d for directed, where ω_c is a knee frequency, Γ is the relaxation rate, D is the diffusion coefficient, and v is a directed speed. The smallest and largest frequencies define the physical ranges for directed transport and diffusion, respectively, which are 0.003< ν <1 μm/sec and 2 × 10⁻⁴ < D < 0.05 μm²/sec, and 5 nm <Δx _max < 2 μm. The velocity range is well within the range of intracellular motion in which molecular motors move organelles at speeds between 0.1 and 1 μm/sec.^{79

–83} Diffusion of very small organelles, as well as molecular diffusion, is too fast to be resolved by our maximum frame rate of 10 fps. Membrane undulations are a common feature of cellular motions, leading to the phenomenon of flicker.^{84

–88} The characteristic frequency for membrane undulations tends to be in the range around 0.1–0.5 Hz.^82,85,89 Much larger shape changes of the membrane, related to cytokinesis or bleb formation during necrosis or apoptosis, tend to be even slower, with backscatter frequencies around 0.01–0.1 Hz.

The general aspect of these characteristic frequencies is the connection of spatial scale with temporal scale. Cell shape changes at the scale of tens of micrometers occur on frequency scales of 0.01–0.1 Hz. Smaller membrane undulations with spatial wavelengths of micrometers occur on frequency scales of 0.1–1 Hz. Even smaller organelle and cytoskeletal motions on submicrometer scales occur on frequency scales of 1–10 Hz. This spatial–temporal trend is only semiquantitative, but it provides a general principle that may help disentangle the mixtures of frequencies that arise from multiple DLS experiments.

The effect of temperature on the spectrogram is shown in Figure 5. The baseline is set for room temperature at 24 °C. At t = 0, the temperature is increased to physiological temperature at 37 °C and held there for almost 2 h, after which it is increased to 43 °C and held again for almost 2 h. When the temperature is returned to 37 °C, the spectrogram does not return to its prior form but shows an increased high-frequency content and a new low-frequency band. Temperatures above about 42 °C are lethal to cells, and the response is likely a heat-shock response, including cell membrane blebbing (the low-frequency signature) and enhanced organelle transport (the high-frequency signature).

Figure 5.

Differential spectrogram response to temperature. The baseline is established at room temperature, and the temperature is raised to physiological temperature of 37 °C, and then to 43 °C, after which it is returned to 37 °C.

The dose–response to the drug nocodazole, an antitubulin antimitotic drug, is shown in Figure 6 as a function of dose. Each dose is for a separate tumor spheroid, and the individual spectrograms of the different doses are shown in the panels on the left. The dose–response function on the right was obtained by crosscorrelating (unnormalized) the individual spectrograms with the 10-μg/mL spectrogram. This leads to the response curve that saturates above 1 μg/mL. The sensitivity is approximately 10 ng/mL with a 50% effective concentration EC₅₀ = 60 ng/mL. The saturated response has a variability of approximately 10%, which reflects the tumor-to-tumor variability. Although we have not yet performed a multiwell statistics experiment, this level of tumor variability is anticipated to produce a z-factor of approximately 70%.

Figure 6.

Dose response to nocodazole from 10 ng/mL to 10 μg/mL. Each dose is applied to a different tumor spheroid. The dose response has an EC₅₀ = 60 ng/mL with a sensitivity of 10 ng/mL.

Spectrogram Correlations

We applied 28 different drugs, concentrations, and conditions to 28 different tumor spheroids and obtained the spectrograms in each case. These spectrograms are complex, showing many different types of features in response to the applied drugs. Similar drugs produced similar spectrograms, whereas widely differing spectrograms could be elicited from drugs with very different mechanisms of action. To capture the similarity or dissimilarity of the drug–response spectrograms, we calculated the cross-correlation coefficient between spectrograms specified by

C (x, y) = \frac{\sum_{v, t} S^{(x)} (v, t) S^{(y)} (v, t)}{\sqrt{\sum_{v, t} {(S^{(x)} (v, t))}^{2}} \sqrt{\sum_{v, t} {(S^{(y)} (v, t))}^{2}}}

(7)

where S ^(x)(ν, t) is the spectrogram of condition x, and the sum is over both frequency and time. The correlation coefficient was calculated among the 28 conditions, leading to a 28-by-28 similarity matrix of correlation coefficients, shown in Figure 7, in which red denotes positive correlation and blue denotes negative correlations.

Figure 7.

Similarity matrices after hierarchical clustering on the shell spectrograms. Red corresponds to positive correlation, and blue to negative correlation. (A) The shell–shell correlations exhibit a nearly block-diagonal structure with approximately five groups of drugs/conditions after clustering. (B) Core–core correlations for the sequence clustered by the shell. (C) Shell–core correlations for the same sequence.

The similarity matrix provides the basis for a hierarchical clustering algorithm that can group different drugs and conditions by their similar drug–response spectrograms. Each row of the N × N similarity matrix is an N-dimensional vector. The inner product of each row defines a measure of distance

Dist (A, B) = \frac{1}{N} \sum_{y = 1}^{N} C (A, y) C (B, y)

(8)

Inner products near unity correspond to “close” associations, and inner products near negative unity correspond to “far” associations.

The compounds, doses, and conditions are given in Table 1, grouped according to the hierarchical clustering of the drug responses based on the spectrograms of the proliferating shell. In a hierarchical clustering algorithm, at each stage, the two closest vectors are identified and grouped into an average vector, and the number of vectors decreases by one. The sequence of the vectors that are grouped in this way is retained until all vectors have been grouped. The sequential grouping of vectors produces a clustering of similar drug responses. The similarity matrix is rearranged according to the sequence of grouping according to the shell–shell correlations, and the result is shown in Figure 7A. The similarity matrix shows structure, with an approximately “block diagonal” appearance. If the matrix were truly block diagonal, the clustering would consist of groups of unique behaviors. However, in the case of the drug responses, there is considerable off-diagonal structure, which indicates that different groups of drug responses do share some aspects in common.

Table 1.

Reference compounds, doses and conditions

Compound or condition	Action
Temperature 24–37 °C	Increased motility
pH 6	Weak acidic
KCN 20 μg/mL	Inhibit electron transport
Iodoacetate 1 μg/mL	Inhibits glycolysis
Osmolarity 508	Hypertonic cell desiccation
Osmolarity 428	Hypertonic cell desiccation
Iodoacetate 10 μg/mL	Inhibit glycolysis
KCN 200 μg/mL	Inhibit electron transport
Tumor necrosis factor 5 μg/mL	Cytokine apoptosis induction
Cytochalasin 10 μg/mL	Antiactin, antimitotic
Cytochalasin 1 μg/mL	Antiactin, antimitotic
Iodoacetate 40 μg/mL	Inhibit glycolysis
pH 9	Strong basic
Nocodazole 0.01 μg/mL	Antitubulin, antimitotic
Osmolarity 154	Strong hypotonic cell swelling
Osmolarity 77	Strong hypotonic cell swelling
Cytochalasin 50 μg/mL	Antiactin, antimitotic, apoptosis
pH 5	Strong acidic
Colchicine 10 μg/mL	Antitubulin, antimitotic
Iodoacetate 20 μg/mL	Inhibit glycolysis
Nocodazole 1 μg/mL	Antitubulin, antimitotic
Taxol 10 μg/mL	Tubulin stabilization, antimitotic
Cycloheximide	Apoptosis induction
Nocodazole 0.1 μg/mL	Antitubulin, antimitotic
Colchicine 1 μg/mL	Antitubulin, antimitotic
Nocodazole 10 μg/mL	Antitubulin, antimitotic
Taxol 1 μg/mL	Tubulin stabilization, antimitotic
pH 8	Weak basic

The compounds were applied at multiple different doses, some below and some above EC₅₀. Therefore, the clustering does not automatically group similar drugs together, but instead clusters responses. If a higher dose induces apoptosis, but a lower dose does not, then the higher dose will be grouped with other drugs or conditions that induce apoptosis. In a future analysis, each drug response can be referenced to its own EC₅₀, which would remove the dose dependence in the clustering.

The multicellular tumor spheroids have different conditions for the proliferating shell relative to the core, which tends to be hypoxic and ATP depleted. Furthermore, large tumors (larger than 500-μm diameter) have increasingly necrotic cores. TDS is volumetric (depth gated), so the response of the core can be compared with the different response of the shell. The core–core correlations are shown in Figure 7B for the same ordering as for the clustered data using the shell–shell correlations. There are many similarities in the core–core correlations, but with notable differences, and several of the blocks are missing. This is because the shell and core respond differently to the applied conditions. This is seen more clearly in Figure 7C, which is the correlation of the shell spectrograms with the core spectrograms. The diagonal elements of this similarity matrix are not all near unity, showing significant differences in the response of the core relative to the shell, even responding in the same tumor to the same conditions.

Because the core is under a different condition than the shell, the 28 conditions are doubled to 56 conditions, which can be crosscorrelated in a similarity matrix and clustered according to all 56 conditions. The resulting clustered similarity matrix is shown in Figure 8. Many of the shell and core spectrograms are clustered together (highly similar), but there are many other pairs that are separated in the clustering process because their respective spectrograms are not highly similar. This is likely a consequence of the ATP-depleted conditions of the core relative to the proliferating shell. For instance, an apoptotic response requires normal ATP concentrations and can occur in the shell, whereas the cells in the core would be forced to follow a necrotic path. This difference between apoptosis and necrosis based on ATP concentrations may be one of the differentiators between the spectrograms of the shell relative to the core. Clearly, a more systematic study of these differences between the shell and the core may uncover specific spectrogram signatures for processes such as apoptosis.

Figure 8.

Similarity matrix for all 56 conditions, including the difference between the shell and the core. The clustering is unsupervised but tends to pair shell and core spectrograms. On the other hand, many shell-core pairs are not clustered, which may reflect different processes that occur in the hypoxic core relative to the proliferating shell.

Conclusions

The 28 drug–response spectrograms share some features in common but with additional distinguishing features that enable phenotypic profiling across a range of reference compounds and conditions that span wide functionality. The size of the test space was sufficient to perform hierarchical clustering of the drugs into groups that share responses in common. The known mechanisms of action of these reference compounds enable approximate assignments to be made for the basis of clustering in each case.

These preliminary results open the way to more detailed analysis of the spectrogram features. Just as no two voiceprints in voice recognition are identical, we hypothesize that the same will be true for drug–response spectrograms—at least for drugs with different mechanisms of action. Feature recognition software that has been developed for voice recognition may be modified for feature analysis of the drug–response spectrograms. Features become the elements of multidimensional feature vectors that can be used to construct higher fidelity similarity matrices for clustering. More importantly, specific features can be selected—for instance apoptotic signatures—and drugs can be clustered with emphasis on these features. By increasing the number of cell types and reference compounds to include standard of care, a broad phenotypic profiling library can be assembled against which to compare new drugs and their tissue dynamics responses.

Footnotes

Acknowledgments

This work was supported by the National Science Foundation grant CBET-0756005. KJ gratefully acknowledges support from the Korea Military Academy Hwarangdae Research Institute.

Competing Interests Statement: The authors certify that they have no relevant financial interests in this manuscript and that any/all financial and material support for this research and work are clearly identified in the Acknowledgments section of this manuscript.

References

Kunz-Schughart

L.A.

Freyer

J.P.

Hofstaedter

Ebner

The use of 3-D cultures for high-throughput screening: the multicellular spheroid model

J. Biomol. Screen 9 2004 273–285

Hirschhaeuser

Menne

Dittfeld

West

Mueller-Klieser

Kunz-Schughart

L.A.

Multicellular tumor spheroids: an underestimated tool is catching up again

J. Biotechnol 148 2010 3–15

Cukierman

Pankov

Stevens

D.R.

Yamada

K.M.

Taking cell-matrix adhesions to the third dimension

Science 294 2001 1708–1712

Webb

D.J.

Horwitz

A.F.

New dimensions in cell migration

Nat. Cell Biol 5 2003 690–692

Keller

P.J.

Pampaloni

Stelzer

E.H.K.

Life sciences require the third dimension

Curr. Opin. Cell. Biol 18 2006 117–124

Pampaloni

Reynaud

E.G.

Stelzer

E.H.K.

The third dimension bridges the gap between cell culture and live tissue

Nat. Rev. Mol. Cell. Biol 8 2007 839–845

Voytik-Harbin

S.L.

Three-dimensional extracellular matrix substrates for cell culture

Methods Cell Biol 63 2001 561–581

Lee

Cuddihy

M.J.

Kotov

N.A.

Three-dimensional cell culture matrices: state of the art

Tissue Eng. Part B Rev 14 2008 61–86

Scheffold

Cerbino

New trends in light scattering

Curr. Opin. Colloid Interface Sci 12 2007 50–57

10.

Perelman

L.T.

Optical diagnostic technology based on light scattering spectroscopy for early cancer detection

Expert Rev. Med. Devices 3 2006 787–803

11.

12.

13.

Huang

Swanson

E.A.

Lin

C.P.

Schuman

J.S.

Stinson

W.G.

Chang

Hee

M.R.

Flotte

Gregory

Puliafto

C.A.

Fujimoto

J.G.

Optical coherence tomography

Science 254 1991 1178–1181

14.

15.

Fujimoto

J.G.

Brezinski

M.E.

Tearney

G.J.

Boppart

S.A.

Bouma

Hee

M.R.

Southern

J.F.

Swanson

E.A.

Optical biopsy and imaging using optical coherence tomography

Nat. Med 1 1995 970–972

16.

Fercher

A.F.

Drexler

Hitzenberger

C.K.

Lasser

Optical coherence tomography—principles and applications

Rep. Prog. Phys 66 2003 239–303

17.

Hyde

S.C.W.

Jones

Barry

N.P.

Dainty

J.C.

French

P.M.W.

Kwolek

K.M.

Nolte

D.D.

Melloch

M.R.

Depth-resolved holography through turbid media using photorefraction

IEEE J. Sel. Top. Quant. Electron 2 1996 965–975

18.

Tziraki

Jones

French

P.M.W.

Melloch

M.R.

Nolte

D.D.

Photorefractive holography for imaging through turbid media using low coherence light

Appl. Phys. B Laser. Optic 70 2000 151–154

19.

Mustata

Peng

L.L.

Turek

J.J.

Melloch

M.R.

French

P.M.W.

Nolte

D.D.

Holographic optical coherence imaging of rat osteogenic sarcoma tumor spheroids

Appl. Opt 43 2004 4862–4873

20.

Jeong

Peng

Turek

J.J.

Melloch

M.R.

Nolte

D.D.

Fourier-domain holographic optical coherence imaging of tumor spheroids and mouse eye

Appl. Opt 44 2005 1798–1806

21.

Cuche

Marquet

Depeursinge

Simultaneous amplitude-contrast and quantitative phase-contrast microscopy by numerical reconstruction of Fresnel off-axis holograms

Appl. Opt 38 1999 6994–7001

22.

Marquet

Rappaz

Magistretti

P.J.

Cuche

Emery

Colomb

Depeursinge

Digital holographic microscopy: a noninvasive contrast imaging technique allowing quantitative visualization of living cells with subwavelength axial accuracy

Opt. Lett 30 2005 468–470

23.

Rappaz

Marquet

Cuche

Emery

Depeursinge

Magistretti

P.J.

Measurement of the integral refractive index and dynamic cell morphometry of living cells with digital holographic microscopy

Opt. Express 13 2005 9361–9373

24.

25.

Bloomfield

V.A.

Quasi-elastic light-scattering applications in biochemistry and biology

Annu. Rev. Biophys. Bioeng 10 1981 421–450

26.

Maret

Wolf

P.E.

Multiple light-scattering from disordered media—the effect of Brownian-motion of scatterers

Z. Phys. B Condens. Matter 65 1987 409–413

27.

Stephen

M.J.

Temporal fluctuations in wave propagation in random media

Phys. Rev. B Condens. Matter 37 1988 1–5

28.

Pine

D.J.

Weitz

D.A.

Chaikin

P.M.

Herbolzheimer

Diffusing-wave spectroscopy

Phys. Rev. Lett 60 1988 1134–1137

29.

Wax

Yang

C.H.

Dasari

R.R.

Feld

M.S.

Path-length-resolved dynamic light scattering: modeling the transition from single to diffusive scattering

Appl. Opt 40 2001 4222–4227

30.

31.

Lemieux

P.A.

Vera

M.U.

Durian

D.J.

Diffusing-light spectroscopies beyond the diffusion limit: the role of ballistic transport and anisotropic scattering

Phys. Rev. E Stat. Nonlin. Soft Matter Phys 57 1998 4498–4515

32.

Bizheva

K.K.

Siegel

A.M.

Boas

D.A.

Path-length-resolved dynamic light scattering in highly scattering random media: the transition to diffusing wave spectroscopy

Phys. Rev. E Stat. Nonlin. Soft Matter Phys 58 1998 7664–7667

33.

Boas

D.A.

Bizheva

K.K.

Siegel

A.M.

Using dynamic low-coherence interferometry to image Brownian motion within highly scattering media

Opt. Lett 23 1998 319–321

34.

Peng

Mustata

Turek

J.J.

Melloch

M.R.

Nolte

D.D.

Time-dependent speckle in holographic optical coherence imaging and the state of health of tumor tissue

Opt. Lett 29 2004 68–70

35.

de Ridder

Autologous confrontation of brain tumor derived spheroids with human dermal spheroids

Anticancer Res 17 1997 4119–4120

36.

Groebe

Mueller-Klieser

On the relation between size of necrosis and diameter of tumor spheroids

Int. J. Radiat. Oncol. Biol. Phys 34 1996 395–401

37.

Hamamoto

Yamada

Kamihira

Iijima

Differentiation and proliferation of primary rat hepatocytes cultured as spheroids

J. Biochem 124 1998 972–979

38.

Hamilton

Multicellular spheroids as an in vitro tumor model

Cancer Lett 131 1998 29–34

39.

Hargrave

Nicholson

P.W.

Delpy

D.T.

Fribank

Optical properties of multicellular tumour spheroids

Phys. Med. Biol 41 1996 1067–1072

40.

Kunz-Schughart

L.A.

Kreutz

Knuechel

Multicellular spheroids: a three-dimensional in vitro culture system to study tumour biology

Int. J. Exp. Pathol 79 1998 1–23

41.

Mueller-Klieser

Method for the determination of oxygen consumption rates and diffusion coefficients in multicellular spheroids

Biophys. J 46 1984 343–348

42.

Freyer

Schor

P.L.

Jarrett

K.A.

Neeman

Sillerud

L.O.

Cellular energetics measured by phosphorous nuclear magnetic resonance spectroscopy are not correlated with chronic nutrient deficiency in multicellular tumor spheroids

Cancer Res 51 1991 3831–3837

43.

Mueller-Klieser

Three-dimensional cell cultures: from molecular mechanisms to clinical applications

Am. J. Physiol. Cell Physiol 1997 1109–1123

44.

Kunz-Schughart

L.A.

Multicellular tumor spheroids: intermediates between monolayer culture and in vivo tumor

Cell Biol. Int 23 1999 157–161

45.

Tyson

C.A.

Frazier

J.M.

In vitro toxicity indicators. Series Title:Methods in Toxicology, 1B 1994

Academic Press

San Diego

46.

Weaver

V.M.

Petersen

O.W.

Wang

Larabell

C.A.

Briand

Damsky

Bissell

M.J.

Reversion of the malignant phenotype of human breast cells in three-dimensional culture and in vivo by integrin blocking antibodies

J. Cell. Biol 137 1997 231–245

47.

Wang

Weaver

V.M.

Petersen

O.W.

Larabell

C.A.

Dedhar

Briand

Lupu

Bissell

M.J.

Reciprocal interactions between beta 1-integrin and epidermal growth factor receptor in three-dimensional basement membrane breast cultures: a different perspective in epithelial biology

Proc. Natl. Acad. Sci. U.S.A 95 1998 14821–14826

48.

Jacks

Weinberg

R.A.

Taking the study of cancer cell survival to a new dimension

Cell 111 2002 923–925

49.

Abbott

Cell culture biology's new dimension

Nature 424 2003 870–872

50.

Griffith

L.G.

Swartz

M.A.

Capturing complex 3D tissue physiology in vitro

Nat. Rev. Mol. Cell Biol 7 2006 211–224

51.

Yamada

K.M.

Cukierman

Modeling tissue morphogenesis and cancer in 3D

Cell 130 2007 601–610

52.

Gudjonsson

Ronnov-Jessen

Villadsen

Bissell

M.J.

Petersena

O.W.

To create the correct microenvironment: three-dimensional heterotypic collagen assays for human breast epithelial morphogenesis and neoplasia

Methods 30 2003 247–255

53.

Nelson

C.A.

Bissell

M.J.

Modeling dynamic reciprocity: engineering three-dimensional culture models of breast architecture, function, and neoplastic transformation

Semin. Cancer Biol 15 2005 342–352

54.

Lee

G.Y.

Kenny

P.A.

Lee

E.H.

Bissell

M.J.

Three-dimensional culture models of normal and malignant breast epithelial cells

Nat. Methods 4 2007 359–365

55.

Friedrich

Ebner

Kunz-Schughart

L.A.

Experimental anti-tumor therapy in 3-D: spheroids—old hat or new challenge?

Int. J. Radiat. Biol 83 2007 849–871

56.

Hamer

Leenstra

Van Noorden

C.J.F.

Zwinderman

A.H.

Organotypic glioma spheroids for screening of experimental therapies: how many spheroids and sections are required?

Cytometry A 75 2009 528–534

57.

Poland

Sinha

Siegert

Schnolzer

Korf

Hauptmann

Comparison of protein expression profiles between monolayer and spheroid cell culture of HT-29 cells revealed fragmentation of CK18 in three-dimensional cell culture

Electrophoresis 23 2002 1174–1184

58.

Dardousis

Voolstra

Roengvoraphoj

Sekandarzad

Mesghenna

Winkler

Hescheler

Sachinidis

Identification of differentially expressed genes involved in the formation of multicellular tumor spheroids by HT-29 colon carcinoma cells

Mol. Ther 15 2007 94–102

59.

60.

Zietarska

Maugard

C.M.

Filali-Mouhim

Alam-Fahmy

Tonin

P.N.

Provencher

D.M.

Mes-Masson

A.M.

Molecular description of a 3D in vitro model for the study of epithelial ovarian cancer (EOC)

Mol. Carcinog 46 2007 872–885

61.

Chang

T.T.

Hughes-Fulford

Monolayer and spheroid culture of human liver hepatocellular carcinoma cell line cells demonstrate distinct global gene expression patterns and functional phenotypes

Tissue Eng. Part A 15 2009 559–567

62.

Shimada

Yamashita

Tanaka

Shirabe

Nakazawa

Ijima

Sakiyama

Fukuda

Funatsu

Sugimachi

Characteristic gene expression induced by polyurethane foam/spheroid culture of hepatoma cell line, Hep G2 as a promising cell source for bioartificial liver

Hepatogastroenterology 54 2007 814–820

63.

Yamashita

Shimada

Harimoto

Tanaka

Shirabe

Ijima

Nakazawa

Fukuda

Funatsu

Maehara

cDNA microarray analysis in hepatocyte differentiation in Huh 7 cells

Cell Transplant 13 2004 793–799

64.

Gaedtke

Thoenes

Culmsee

Mayer

Wagner

Proteomic analysis reveals differences in protein expression in spheroid versus monolayer cultures of low-passage colon carcinoma cells

J. Proteome Res 6 2007 4111–4118

65.

Frankel

Buckman

Kerbel

R.S.

Abrogation of taxol-induced G(2)-M arrest and apoptosis in human ovarian cancer cells grown as multicellular tumor spheroids

Cancer Res 57 1997 2388–2393

66.

Hazlehurst

L.A.

Landowski

T.H.

Dalton

W.S.

Role of the tumor microenvironment in mediating de novo resistance to drugs and physiological mediators of cell death

Oncogene 22 2003 7396–7402

67.

Serebriiskii

Castello-Cros

Lamb

Golemis

E.A.

Cukierman

Fibroblast-derived 3D matrix differentially regulates the growth and drug-responsiveness of human cancer cells

Matrix Biol 27 2008 573–585

68.

David

Dulong

Le Cerf

Cazin

Lamacz

Vannier

J.P.

Hyaluronan hydrogel: an appropriate three-dimensional model for evaluation of anticancer drug sensitivity

Acta Biomater 4 2008 256–263

69.

Ballangrud

A.M.

Yang

W.H.

Charlton

D.E.

McDevitt

M.R.

Hamacher

K.A.

Panageas

K.S.

D.S.

Bander

N.H.

Scheinberg

D.A.

Sgouros

Response of LNCaP spheroids after treatment with an alpha-particle emitter (Bi-213)-labeled anti-prostate-specific membrane antigen antibody (J591)

Cancer Res 61 2001 2008–2014

70.

Carlsson

Acker

Relations between pH, oxygen partial pressure and growth in cultured-cell spheroids

Int. J. Cancer 42 1988 715–720

71.

Dubessy

Merlin

J.L.

Marchal

Guillemin

Spheroids in radiobiology and photodynamic therapy

Crit. Rev. Oncol. Hematol 36 2000 179–192

72.

Durand

R.E.

Olive

P.L.

Resistance of tumor cells to chemo- and radiotherapy modulated by the three-dimensional architecture of solid tumors and spheroids

Methods Cell Biol 64 2001 211–233

73.

74.

Pollard

T.D.

Borisy

G.G.

Cellular motility driven by assembly and disassembly of actin filaments

Cell 112 2003 453–465

75.

Mustata

Turek

J.J.

French

P.M.W.

Melloch

M.R.

Nolte

D.D.

Holographic optical coherence imaging of tumor spheroids

Appl. Phys. Lett 83 2003 575–577

76.

Jeong

Turek

J.J.

Nolte

D.D.

Fourier-domain digital holographic optical coherence imaging of living tissue

Appl. Opt 46 2007 4999–5008

77.

78.

Parthasarathy

A.B.

Tom

W.J.

Gopal

Zhang

X.J.

Dunn

A.K.

Robust flow measurement with multi-exposure speckle imaging

Opt. Express 16 2008 1975–1989

79.

Suissa

Place

Goillot

Freyssingeas

Internal dynamics of a living cell nucleus investigated by dynamic light scattering

Eur. Phys. J. E Soft Matter 26 2008 435–448

80.

Nan

X.L.

Sims

P.A.

Xie

X.S.

Organelle tracking in a living cell with microsecond time resolution and nanometer spatial precision

ChemPhysChem 9 2008 707–712

81.

Karnaky

K.J.

Garretson

L.T.

Oneil

R.G.

Video-enhanced microscopy of organelle movement in an intact epithelium

J. Morphol 213 1992 21–31

82.

Brazhe

N.A.

Brazhe

A.R.

Pavlov

A.N.

Erokhova

L.A.

Yusipovich

A.I.

Maksimov

G.V.

Mosekilde

Sosnovtseva

O.V.

Unraveling cell processes: interference imaging interwoven with data analysis

J. Biol. Phys 32 2006 191–208

83.

Trinczek

Ebneth

Mandelkow

Tau regulates the attachment/detachment but not the speed of motors in microtubule-dependent transport of single vesicles and organelles

J. Cell Sci 112 1999 2355–2367

84.

Brochard

Lennon

J.F.

Frequency spectrum of flicker phenomenon in erythrocytes

J. Phys. France 36 1975 1035–1047

85.

Strey

Peterson

Measurement of erythrocyte-membrane elasticity by flicker eigenmode decomposition

Biophys. J 69 1995 478–488

86.

Zilker

Ziegler

Sackmann

Spectral-analysis of erythrocyte flickering in the 0.3–4-Mu-M-1 regime by microinterferometry combined with fast image-processing

Phys. Rev. A 46 1992 7998–8002

87.

Peterson

M.A.

Strey

Sackmann

Theoretical and phase-contrast microscopic eigenmode analysis of erythrocyte flicker: amplitudes

J. Phys. II France 2 1992 1273–1285

88.

Yoon

Y.Z.

Hong

Brown

Kim

D.C.

Kang

D.J.

Lew

V.L.

Cicuta

Flickering analysis of erythrocyte mechanical properties: dependence on oxygenation level, cell shape, and hydration level

Biophys. J 97 2009 1606–1615

89.

Evans

Gratzer

Mohandas

Parker

Sleep

Fluctuations of the red blood cell membrane: relation to mechanical properties and lack of ATP dependence

Biophys. J 94 2008 4134–4144

Tissue Dynamics Spectroscopy for Three-Dimensional Tissue-Based Drug Screening

Abstract

Keywords

Three-dimensional Tissues for Screening

Multicellular Tumor Spheroids

Holographic Optical Coherence Imaging

Tissue Dynamics Spectroscopy

Spectrogram Correlations

Conclusions

Footnotes

Acknowledgments

References