ArticlePDF Available

Time-lapse 3-D measurements of a glucose biosensor in multicellular spheroids by light sheet fluorescence microscopy in commercial 96-well plates


Abstract and Figures

Light sheet fluorescence microscopy has previously been demonstrated on a commercially available inverted fluorescence microscope frame using the method of oblique plane microscopy (OPM). In this paper, OPM is adapted to allow time-lapse 3-D imaging of 3-D biological cultures in commercially available glass-bottomed 96-well plates using a stage-scanning OPM approach (ssOPM). Time-lapse 3-D imaging of multicellular spheroids expressing a glucose Förster resonance energy transfer (FRET) biosensor is demonstrated in 16 fields of view with image acquisition at 10 minute intervals. As a proof-of-principle, the ssOPM system is also used to acquire a dose response curve with the concentration of glucose in the culture medium being varied across 42 wells of a 96-well plate with the whole acquisition taking 9 min. The 3-D image data enable the FRET ratio to be measured as a function of distance from the surface of the spheroid. Overall, the results demonstrate the capability of the OPM system to measure spatio-temporal changes in FRET ratio in 3-D in multicellular spheroids over time in a multi-well plate format.
Content may be subject to copyright.
Scientific RepoRts | 6:37777 | DOI: 10.1038/srep37777
Time-lapse 3-D measurements
of a glucose biosensor in
multicellular spheroids by light
sheet uorescence microscopy in
commercial 96-well plates
Vincent Maioli1, George Chennell1,2,3, Hugh Sparks1, Tobia Lana2,3, Sunil Kumar1,
David Carling2,3, Alessandro Sardini2,3 & Chris Dunsby1,4
Light sheet uorescence microscopy has previously been demonstrated on a commercially available
inverted uorescence microscope frame using the method of oblique plane microscopy (OPM). In this
paper, OPM is adapted to allow time-lapse 3-D imaging of 3-D biological cultures in commercially available
glass-bottomed 96-well plates using a stage-scanning OPM approach (ssOPM). Time-lapse 3-D imaging
of multicellular spheroids expressing a glucose Förster resonance energy transfer (FRET) biosensor is
demonstrated in 16 elds of view with image acquisition at 10 minute intervals. As a proof-of-principle,
the ssOPM system is also used to acquire a dose response curve with the concentration of glucose in the
culture medium being varied across 42 wells of a 96-well plate with the whole acquisition taking 9 min.
The 3-D image data enable the FRET ratio to be measured as a function of distance from the surface of the
spheroid. Overall, the results demonstrate the capability of the OPM system to measure spatio-temporal
changes in FRET ratio in 3-D in multicellular spheroids over time in a multi-well plate format.
Multicellular spheroids (MCS) provide a 3-D model of in vitro cell culture and are increasingly being used in
in vitro assays1,2. Compared to 2-D cell monolayer culture on a plastic or glass surfaces, MCS provide cell-cell
contacts and gradients of environmental parameters such as oxygen, nutrients and pH similar to those found in
tumours1,2. e oxygen gradient is caused by the rate of consumption of oxygen by cells in the MCS being greater
than the rate of diusion of O2 into the centre. For larger MCS, typically with diameters greater than approxi-
mately 400–500 μ m, the centre is known to become hypoxic2 leading to increased rates of glycolysis and conse-
quently increased lactic acid production in their centre. ere is also increasing evidence that gene expression in
MCS is dierent to that in 2-D cell culture3–5 and is more similar to that found in xenogra tumours compared to
2-D cell culture6. In addition to observations of dierent assay results in 2-D compared to 3-D cell culture7,8, MCS
have been applied to show that certain compounds can be eective at the centres of 3-D MCS but be ineective at
the edges of MCS and in 2-D cell culture9. MCS also provide the opportunity to study diusion rates of a range of
substances into and out of the centre of the MCS.
Genetically expressed Förster resonance energy transfer (FRET) biosensors provide a means to measure a
wide range of cellular parameters10,11. When combined with 3-D cell culture, they have been applied to observe
changes in redox potential12 and the onset of apoptosis13 when the MCS are exposed to pharmacological agents,
and also to readout AMPK activity14. FRET biosensors have also been applied to monitor Rac1, Cdc42 and RhoA
activity in a spheroid invasion assay15.
e instrumentation available for 3-D imaging in MCS using single-photon uorescence excitation has been
recently reviewed16 and multiphoton excitation approaches have also been explored, e.g. refs 4,17–19, but the
1Photonics Group, Department of Physics, Imperial College London, United Kingdom. 2MRC Clinical Sciences Centre
(CSC), Du Cane Road, London, United Kingdom. 3Institute of Clinical Sciences (ICS), Faculty of Medicine, Imperial
College London, Du Cane Road, London, United Kingdom. 4Centre for Pathology, Faculty of Medicine, Imperial
College London, United Kingdom. Correspondence and requests for materials should be addressed to V.M. (email:
Received: 31 August 2016
Accepted: 01 November 2016
Published: 25 November 2016
Scientific RepoRts | 6:37777 | DOI: 10.1038/srep37777
speed of these single point scanning approaches are limited by the maximum peak excitation power that can
be tolerated by the sample in terms of photobleaching and, in the case of live cells, phototoxicity. Light sheet
uorescence microscopy (LSFM) techniques can overcome this limitation and have also been applied to imag-
ing MCS12,20,21, but conventional LSFM require the MCS to be mounted in a manner that generally precludes
high-throughput imaging. is limitation has been addressed using a number of congurations. Galland et al.22
fabricated custom multiwell plates incorporating a 45° mirror within each well to allow light sheet illumina-
tion and imaging with the same objective. However, this approach relies on relatively sophisticated fabrication
approaches and is not compatible with conventional multiwell plates. Strnad et al.23 demonstrated the use of a
v-shaped uorinated ethylene propylene (FEP) channel containing a linear array of 20 zebrash embryos for
time-lapse 3-D imaging of embryo development. While this is a very powerful approach, it does not easily allow
dierent samples within the channel to be exposed to dierent experimental conditions and the sample prepa-
ration process does not easily scale to 100–1000’s samples. Light sheet uorescence microscopy in conventional
multi-well plate arrays has been demonstrated by the use of a uid lled prism to enable illumination and imaging
of the sample at 45° to the plate normal24. is is an elegant approach, however the need for an objective with
a working distance that is suciently long to allow space for the uid-lled prism limits the NA of the illumi-
nation and collection objectives that can be employed. In addition, imaging through the glass coverslip base of
the multi-well plate at 45° to the optical axis introduces astigmatism and higher order aberrations that require
subsequent correction and which increase in severity as the NA employed increases. High throughput light sheet
uorescence microscopy has also been demonstrated using uidic approaches where the sample either ows
through the light sheet using an FEP tube at 45° to the light sheet25, or using a millimetre-scale lab-on-a-chip
device fabricated using femtosecond micromachining where the sample ows up through the light sheet towards
the detection objective26.
is paper concerns the application of LSFM to 3-D imaging of MCS prepared in conventional commercially
available 96-well plates on a commercially available inverted microscope frame. In order to deliver the excitation
light and collect the uorescence at 90° using a conventional inverted uorescence microscope objective and
frame, we have adapted the technique of oblique plane microscopy (OPM)27 so that it can be used for 3-D imag-
ing via a stage-scanning approach, which we refer to as stage-scanning OPM (ssOPM). OPM uses the same high
numerical aperture microscope objective for both illumination and imaging of the specimen without the need
for any modications to the chamber holding the sample. OPM can be considered as a conventional light sheet
microscope setup consisting of perpendicular illumination and detection arms where the light sheet is imaged
into a remote sample and the resulting uorescence is returned using a high numerical aperture image relay
designed to achieve equal lateral and axial magnications.
In this paper we demonstrate time-lapse 3-D multi-well imaging of a glucose FRET biosensor in live MCS
using ssOPM. We present a range of experiments demonstrating the speed of the system, including imaging 16
elds of view at 10 minute intervals over 4 hours and imaging 42 wells of a 96-well plate in 9 min. Together, these
exemplar experiments illustrate the potential of ssOPM to probe a range of experimental questions across a range
of dierent time-scales. Advantages of the full 3-D imaging provided by ssOPM include the ability to calculate
the minimum distance to the surface of a spheroid for every voxel, rather than relying on assumptions typically
required in the analysis of 2-D images of the spheroid’s sphericity and exact selection of the mid-plane when
imaging. We illustrate these advantages by measuring FRET ratio as a function of distance from surface of sphe-
roid for 16 spheroids in a single time-lapse experiment taking 4 hours across 4 dierent experimental conditions.
OPM. e optical setup for OPM, see Fig.1, has been described previously27–29. Briey, light from four excita-
tion sources at dierent wavelengths was combined using dichroic mirrors and controlled using an acousto-op-
tic tunable lter (AOTF) before being coupled into a single-mode polarisation maintaining optical bre. Light
exiting the bre is collimated (L1), focused in the horizontal direction by a cylindrical lens (C1) onto the back
focal plane of spherical lens L2, which results in a vertically oriented light sheet angled at 55° to the optical axis
of objective lens O2. A pair of microscopes arranged back-to-back (formed by O2, compound tube lens TL2,
TL1 and O1) are then used to relay this light sheet to the focal plane of O1, where the light sheet is at 55° to the
optical axis of O1. In Fig.1 it is important to note that the optics inside the grey box (inverted microscope frame)
are oriented at 90° compared to the rest of the gure, i.e. in the direction perpendicular to the plane of the page.
e microscope formed by O1 and TL1 is a commercially available inverted microscope frame (Olympus IX71).
O1 is a 60× /1.2NA water immersion objective and so the overall magnication from the focal plane of O1 to
the focal plane of O2 is chosen to be equal to the refractive index of water to ensure that the overall lateral and
axial magnications between the focal planes of O1 and O2 are equal, see Botcherby et al.30 for more detail. e
excitation light sheet produced across the focal plane of O1 excites uorescence in the sample and the resulting
uorescence is imaged back to the focal plane of O2. O3 is positioned so that its optical axis is at 35° to that of O1
and O2 and therefore, together with tube lenses TL3a and TL3b, focuses uorescence from the excitation light
sheet onto cameras 1 and 2. A dichroic beamsplitter DC (Chroma T510lprxtxt-UF3) and emission lters EM1
(Semrock FF01 550/49) and EM2 (Semrock FF01 482/25) separate the emitted uorescence into two spectral
bands for detection of donor and acceptor uorescence emission respectively. Specications for all of the optical
components, together with a characterisation of the spatial resolution and optical performance of the OPM sys-
tem can be found in reference29.
ssOPM FRET image acquisition protocol. Stage-scanning OPM was implemented here using a motor-
ized stage (SCAN-IM 120 × 80, Marzhäuser) controlled by a driver unit (Tango 2 tted with AUX I/O option,
Marzhäuser) that can be congured to output a TTL trigger each time the stage has travelled a predened dis-
tance, which was set to 2 μ m for the results presented here. is TTL output was connected to a digital acquisition
Scientific RepoRts | 6:37777 | DOI: 10.1038/srep37777
box (National Instruments NI USB-6229) that was congured to output a pattern of signals each time a trigger
signal is received from the x-y stage, see SupplementaryFigureS1. ese signals were used to control the power
and duration of the laser excitation and to trigger the start of the exposure of both cameras. e stage was cong-
ured to scan in the y direction shown in Fig.1a.
We found that the motorized x-y stage produced mechanical vibrations large enough to cause a reduction in
the nal image resolution when operated in the speed range 0.4–0.6 μ m ms1 and therefore a stage scan speed of
0.1 μ m ms1 was used for this work during stage-scan imaging. A faster stage scan speed of 10 μ m ms1 was used
when translating between elds of view.
As shown in SupplementaryFigureS1, aer every 2 μ m (20 ms) of stage travel the system was congured to
provide 2 ms of illumination from the 457 nm laser for excitation of the donor with simultaneous acquisition on
both cameras, i.e. providing images of the donor (camera 2) and the sensitised acceptor emission (camera 1).
10 ms later, 2 ms of 515 nm excitation and simultaneous acquisition on camera 1 provided an image of the directly
excited acceptor emission. e use of 2 ms camera exposure times means that the sample moves 0.2 μ m during
the integration, which is less than the measured 0.5 μ m in-OPM plane spatial resolution29. e details of the laser
power used for each experiment are given in SupplementaryTableS2.
e two sCMOS cameras (PCO.edge, PCO) were both congured to acquire in Global Reset acquisition mode
with 1280× 1000 pixels. e exposure time was dened by the 2 ms laser illumination period. To acquire one eld
of view, the x-y stage was congured to scan 500 μ m in the y direction (see Fig.1a), resulting in a stack of 250
images in each of the 3 channels (donor, sensitised emission, directly excited acceptor). Overall, camera 1 was
triggered at 100 Hz and camera 2 was triggered at 50 Hz during acquisition. Each 3-D volume took 5 s to acquire
and produced 2 GB of raw image data.
Image acquisition was controlled by a HP z800 PC with 96 GB of RAM, 2× 512 MB SSD HDD congured in
RAID 0 and 8× 2TB HDDs congured in RAID 6.
e average background level for each camera when exciting with the 457 nm laser was determined by taking the
average pixel value over a blank eld of view, which we refer to as IDA background for camera 1 (see Fig.1) and IDD background
for camera 2.
e inverted microscope frame was tted with a temperature controlled enclosure set to maintain 37 °C.
Microscope objective O1 was tted with a collar to provide a continuous supply of water immersion liquid and a
heating collar set to maintain 37 °C (0280.036, Pecon).
Image registration. e parameters needed to co-register images acquired on camera 1 and camera 2 (see Fig.1)
were obtained from a single 2-channel uorescence acquisition of a volume of Sphero Multi-Flurophore 0.13 μ m
beads (FP-0257-2, Spherotech Inc.) in 10% agarose using the 457 nm excitation laser. ese data were then used
to determine the x shi, y shi and rotation needed to co-register data acquired on camera 2 to data acquired on
camera 1.
FRET glucose biosensor. Puried FLII12Pglu-700μ δ 631 ECFP-Citrine glucose FRET biosensor plasmid was
received from Addgene (17866, Addgene, UK).
Cell Culture. HEK293T cells were grown in DMEM medium (31966, Gibco, USA) supplemented with 10%
fetal bovine serum (Sigma, UK). Cells were maintained in a humidied incubator set to 37 °C with 5.0% CO2.
Figure 1. Schematic of ssOPM system. (a) system diagram. L – lenses, C – cylindrical lenses, O – microscope
objectives, TL – tube lenses, M – mirror, DC – dichroic beamsplitter, EM – emission lter. (b) schematic of
electrical triggering for image acquisition and laser line control.
Scientific RepoRts | 6:37777 | DOI: 10.1038/srep37777
Formation of stable cell lines expressing FRET biosensor. To provide a uniform expression level
of the biosensor, a clone of stably expressing HEK293T cells was prepared. FLII12Pglu-700μ δ 6 biosensor gene
was incorporated into MSCV retroviral vector aer restriction digest with HindIII(NEB) and BamH1(NEB) fol-
lowed by ligation and sequencing. HEK293 cells stably expressing the viral Gagpol gene were transfected with
the nished biosensor retroviral plasmid and plasmid for 10A1 envelope protein. Cell supernatant was collected
24 hours later, centrifuged and transferred to HEK293T cells to be targeted with biosensor gene and polybrene
(Sigma, UK) added. Aer 48 hours expression was observed using an epiuorescence microscope. 100 cells were
plated sparsely on a 14 cm petri dish and colonies grown for 5 days before selection and expansion.
Formation of Spheroids. Spheroids were formed using an agarose mould with u-shaped bottomed wells.
e Microtissues 12–256 Small Spheroids Kit (Microtissues, USA) was utilised to produce sterile agarose sphe-
roid moulds capable of producing 256 spheroids following the manufacturer’s instructions. Briey, sterile 4% w/v
molten agarose was dispensed into the Microtissues mould, allowed to cool and turned out into a 6 well culture
plate. Cells were suspended to a concentration of 1 × 106 ml1 and 190 μ l was dispensed into the agarose mould.
Aer 30 minutes, normal growth medium was added to the well to achieve a nal volume of 5 ml. 24 hours of
incubation were allowed for spheroid formation.
Imaging medium. Imaging was performed in a home-made medium comprising of 130 mM NaCl, 5 mM
KCl, 0.5 mM MgCl, 2 mM CaCl2, 10 mM HEPES and 0 mM glucose. D-glucose of the desired concentration was
added as indicated in the text. e pH of the medium was adjusted to 7.4 and sterile ltered with a 0.22 μ m bottle
top lter unit.
Preparation of 96-well plates. Spheroids were transferred from the agarose mould by inversion into a
sterile 50 ml sample tube and placed on ice. Aer 5 minutes of settling, the volume of medium was reduced to
approximately 300 μ l by slow aspiration of medium taking care not to disturb spheroids at the bottom of the
tube. 100 μ l of Matrigel (BD Biosciences; Cat. No. 354234) was added to the spheroids and mixed gently. 25 μ l of
spheroid/gel mix was pipetted carefully into individual wells of a glass-bottomed 96-well plate (Greiner 96-well
SensoPlate 655892) and incubated for 30–45 minutes to form a gel. 125 μ l of imaging medium with 25 mM glu-
cose was then added to all wells to give a nal volume of 150 μ l in each well. For experiments where glucose free
conditions were required, medium was subsequently removed and replaced twice with glucose free medium,
with 2 minutes incubation between each wash to allow glucose to perfuse out of the gel before replacement. Plates
were then placed inside the microscope incubator enclosure and maintained at 37 °C for 1 hour prior to the start
of imaging.
Activator Compounds. When adding glucose to wells initially containing 150 μ l glucose free medium, 50 μ l
of imaging medium with 100 mM glucose was added to a well to reach a nal concentration of 25 mM.
e glucose transport inhibitor phloretin (Sigma, UK)32,33 was used as received and dissolved at a concentra-
tion of 400 μ M in imaging medium containing 25 mM glucose. 50 μ l of this solution was added to a well contain-
ing 150 μ l of imaging medium with 25 mM glucose to reach a nal concentration of 100 μ M phloretin without
changing the glucose concentration.
β -escin (Sigma, UK), which is a compound causing permeabilisation of the plasma membrane34, was dissolved
in medium containing glucose and used in experiments to achieve a nal concentration of 50 μ M β -escin and
25 mM g lucose.
Fluorescent glucose analogue. e uorescent glucose analogue 2-NBDG (ermoFisher) was prepared
to an initial concentration of 4 mM in imaging medium containing 200 mM glucose. During imaging, 250 μ l
of this solution was added to 1750 μ l of glucose-free imaging medium to reach a nal concentration of 500 μM
2-NBDG. Imaging was performed using the 488 nm excitation line and a 500 nm long-pass emission filter
(Chroma). Each volume consisted of 250 frames acquired at 50 Hz taking 5 s. Volumes were acquired at 25.5 s
Image processing. Raw OPM image data consists of a set of image planes acquired at 55° to the optical axis
of O1. It is therefore necessary to transform the data into a conventional coordinate system where z is parallel and
x and y are perpendicular to the optical axis of O1. As shown in Fig.1, the light sheet illumination propagation
direction lies in the y-z plane and the stage motion is in the y direction. e reconstruction was performed using
a custom-written bi-linear resampling algorithm implemented in MATLAB.
e background-corrected FRET ratio RFRET was calculated for each pixel as
DA DA background
DD DD background
where IDA and IDD are the signals recorded with excitation at 457 nm with detection in the acceptor and donor
channels respectively. IDA background and IDD background are the background signal levels in these two channels respec-
tively due to the camera oset, camera read noise and background room light, and are determined as described
above in the section ‘ssOPM FRET image acquisition protocol. e colour images presented in this paper were
produced using the parula colormap bar in MATLAB to represent RFRET and with ISE represented using brightness.
Due to the use of the CFP-YFP FRET pair in the FLII12Pglu-700μ δ 6 biosensor, there is inevitable
bleed-through of donor CFP emission into the acceptor detection channel IDA and direct excitation of acceptor
YFP by the 457 nm laser. e donor bleed-through and direct excitation of the acceptor are determined by the
emission spectra of the uorophores, the choice of emission lters and dichroic beamsplitter in the detection
Scientific RepoRts | 6:37777 | DOI: 10.1038/srep37777
beam path and the camera spectral sensitivity prole. e eects of donor bleed-through and direct excitation of
the acceptor on the calculated FRET ratio are therefore the same for all experiments presented in this paper and
therefore all measured FRET ratios can be compared directly.
Calculation of spheroid minimum distance to surface maps (MDS). In order to produce a map of
the minimum distance to the surface of the spheroid for all points within each spheroid, it was rst necessary to
determine the location of the spheroid periphery. is was performed by the following procedure:
(1) reshold all volumes using the same manually determined threshold on ISE. e threshold was chosen to
include the interior of all spheroids within the dataset.
(2) A 3-D binary mask for the interior of the spheroid was then produced by computing a connected compo-
nent map and selecting the connected component containing the centre of the image volume.
For spheroids where there was incomplete imaging of the top and/or bottom of the spheroid, the following
optional steps 3 to 5 were used to estimate the extent of the spheroid by assuming that its shape can be approxi-
mated by an ellipsoid of revolution oriented along the z axis:
(3) Calculate the total sensitised emission intensity within the spheroid mask for each z-plane within the
(4) Fit the resulting curve to an expression for the cross-sectional area A perpendicular to the z axis of an ellip-
soid of revolution as a function of axial position z. e ellipsoid of revolution has radius a in the z axis direction
and radius b perpendicular to the z axis, with an axial centre position z0, i.e.
.Az bzz
() 1()
e three t parameters were a, b and z0 and the t was performed using MATLABs t function. e lateral
coordinates of the centre of the spheroid (x0 and y0) were obtained by calculating the centre of mass of the
spheroid mask in the x and y directions.
(5) e t parameters obtained in step 4 were used to estimate the cross-sectional area of the spheroid for the
top and/or bottom regions that were not captured by the imaging process. A circular mask was then generated
for each z plane in these regions.
e following step was then applied to all spheroids:
(6) Morphological closure with a radius of 10 μ m was then applied to the full mask to ll any holes that may be
present within the mask and to smooth the outer boundary. e resulting binary 3-D mask was then eroded
with a radius of 6 μ m in order to match the visually determined outside perimeter of the spheroid.
For elds of view where a second spheroid was partially imaged at the edge of the eld of view and which was
included in the mask because of contact between the two spheroids, the mask was manually edited to exclude the
second spheroid. When the spheroids are not touching the connected component containing the centre of the
image volume is selected by the algorithm (step 2).
For the experiments carried out with the escin cell membrane permeabilisation agent, the spheroids were
manually classied as ‘spreading’, i.e. spheroids that have spread out and have a large contact area with the covers-
lip and that are not well approximated by an ellipsoid of revolution. In this case, only data from the top half of the
spheroid were used in the tting process described in step 4 when estimating the extent of the spheroid above the
region imaged. In addition, for spreading spheroids the bottom of the spheroid was assumed to not be in contact
with the culture medium when calculating the MDS map.
Calculation of FRET ratio as a function of MDS. For each time-point for each spheroid a MDS map was
calculated. e voxels within each image volume were then binned into MDS intervals of 10 μ m and the average
FRET ratio was calculated for each interval.
Data Availability. e raw image data from this study is available under an open source licence from Imperial
College Londons OMERO server at
Time-lapse 3-D multi-eld of view FRET imaging of MCS in conventional 96-well plates. Plate 1
consisted of a 96-well plate prepared with 16 wells used, see the plate map shown in Fig.2, with each well contain-
ing approximately 10 HEK293T spheroids expressing the FLII12Pglu-700μ δ 6 biosensor. One spheroid per well was
manually selected by choosing a spheroid close to the centre of the well and not close to neighbouring spheroids.
ssOPM was used to acquire 3-D images of all 16 wells in succession, returning to image each well at 10 minute
intervals. Image acquisition was started 73 min prior to manual addition of the substances indicated in Fig.2
and then for a further 174 min aer their addition. A total of 24 volumes were acquired for each of the 16 elds
of view and the total volume of image data acquired in this experiment was 530 GB. Figure3 shows the resulting
data volume for well C5 imaged at t = 174 min. A movie of the full dataset from the same spheroid is shown in
Supplementary Video S3. e spatial resolution achieved can be seen to start to degrade for imaging depths
> ~100 μ m which we attribute to scattering of light by the spheroid.
Scientific RepoRts | 6:37777 | DOI: 10.1038/srep37777
Figure4 shows a montage of all of the spheroids imaged during this experiment at t = 73, 14 and 174 min
for the four conditions studied. A movie of the full time-lapse dataset is shown in Supplementary Video S4. e
0 mM glucose control wells (Fig.4a) exhibited a uniformly low FRET ratio across each spheroid for all time points
as expected. For the 25 mM glucose control wells (Fig.4b), the FRET ratio was higher in the centre of the sphe-
roids than at the edges and the dierence was greater for the larger spheroids (e.g. well E8). For the wells initially
without glucose with 25 mM glucose added at t = 0, see Fig.4c, the FRET ratio is uniformly low at t = 73 min. At
t = 14 min there is an approximately uniform increase in FRET ratio across each spheroid. At t = 174 min there
is a higher glucose concentration in the centre of each spheroid compared to the edges. For the wells initially
exposed to 25 mM glucose and with 100 μ M phloretin added at t = 0, the spheroids initially show similar spatial
variation in FRET ratio to the 25 mM glucose control wells. Following addition of phloretin – a glucose transport
inhibitor32,33 – the FRET ratio gradually decreases to a uniformly low value consistent with that seen in the 0 mM
glucose control wells.
Figure 2. Le, map of plate 1 used for time-lapse 3-D imaging of glucose dynamics in HEK293T MCS. Top
right, description of the symbols. Bottom right, side view of plate showing microscope objective and orientation
of light sheet within sample.
Figure 3. Montage of a sub-set of acquired z-planes at 9 μ m intervals from a single HEK293T FLII12Pglu-
700μδ6 spheroid from plate 1 (well C5, see Fig. 2) at time t = 174 min. e false-colour scale indicates FRET
ratio and brightness indicates the intensity of sensitised emission.
Scientific RepoRts | 6:37777 | DOI: 10.1038/srep37777
Figure 4. Montage of all HEK293T FLII12Pglu-700μδ6 spheroids imaged in plate 1 showing data at
t = 73, 14 and 174 min. (a) & (b) panels show data from wells with 0 mM and 25 mM glucose, respectively.
Panel (c) shows data from wells initially with 0 mM glucose and addition of 25 mM glucose at t = 0. Panel
(d) shows data from wells with 25 mM glucose where 100 μ M phloretin is added at t = 0. An x-y, x-z and y-z
slice is shown through the centre of each spheroid. e false-colour scale shows the FRET ratio and brightness
indicates intensity of the sensitised emission.
Scientific RepoRts | 6:37777 | DOI: 10.1038/srep37777
Analysis of HEK293T FLII12Pglu-700μδ6 3-D data. As the FRET ratios observed vary with distance
from the surface of the spheroid, 3-D minimum distance to surface (MDS) maps were calculated for each sphe-
roid at each time-point, see Materials and Methods. Example 3-way cuts of the resulting MDS maps for wells C5
of plate 1 are shown in Fig.5.
e voxels within each spheroid were then binned into 10 μ m MDS intervals allowing the mean FRET ratio
to be calculated for each interval, see Fig.6. e control data from spheroids exposed to zero glucose for all time
points (wells C6, D5, E6 and F5) show that the FRET ratio does not vary with MDS or time and is uniformly low.
e data from spheroids exposed to 25 mM glucose for all time points (wells C8, D7, E8 and F7) show a higher
FRET ratio in the centre compared to the edges, see Discussion section. ere is a small ripple in the FRET ratio
values around t = 0 for both the zero and 25 mM glucose wells, which we attribute to the change in temperature
within the microscope enclosure caused by opening the door in order to manually add the substances indicated
in Fig.2. e quantum eciency of eCFP, which is the donor in the FLII12Pglu-700μ δ 6 biosensor, has previously
been shown to be sensitive to temperature35.
For spheroids initially exposed to 0 mM glucose and with 25 mM glucose added at t = 0 (wells C5, D6, E5
and F6), the FRET ratio initially increases uniformly for all MDS values. From t = ~20 min onwards, the FRET
ratio at the surface of the spheroids overshoots the nal steady-state value while the FRET ratio for larger MDS
values gradually saturates to a higher value. is can be seen in all 4 spheroids for this condition individually, see
SupplementaryFigureS5. For spheroids initially exposed to 25 mM glucose with 100 μ M phloretin added at t = 0
(wells C7, D8, E7 and F8), for t < 0 the data shows that the FRET ratio in the centre of the spheroid is higher than
Figure 5. Le column, sensitised acceptor emission intensity for x-y, x-z and y-z slices through the centre
of the HEK293T FLII12Pglu-700μδ6 spheroid imaged in well C5 of plate 1 at t = 174 min. Middle column,
corresponding FRET ratio map without intensity merging. Right column, calculated map of the minimum
distance to the surface of the spheroid.
Figure 6. HEK293T FLII12Pglu-700μδ6 FRET ratio as a function of MDS averaged over all four spheroids
for each condition. (a) 0 mM extracellular glucose control, (b) 25 mM glucose control, (c) 0 mM glucose then
25 mM glucose and (d) 25 mM glucose then addition of 100 μ M phloretin. e horizontal dashed red line shows
the mean FRET ratio over all time-points for panel (a).
Scientific RepoRts | 6:37777 | DOI: 10.1038/srep37777
the edges, which is consistent with the glucose control data (wells C8, D7, E8 and F7). For times t > 0 the FRET
ratio falls with time down to the level seen in the in the zero glucose control data (wells C6, D5, E6 and F5).
e results presented for spheroids initially exposed to zero glucose and then with 25 mM glucose added at
t = 0 and for spheroids initially exposed to 25 mM glucose with 100 μ M phloretin added at t = 0 were repeated
independently on a separate plate (plate 2) on a dierent day again with four spheroids per condition. Similar
trends were observed, see SupplementaryFigureS6. e inter- and intra-plate variation in FRET ratio is investi-
gated in more detail below.
To demonstrate the potential for the ssOPM to image a larger number of elds of view automatically, we pre-
pared a 96-well plate (plate 3) where spheroids were seeded into 54 wells. Six dierent glucose concentrations in
the culture medium ranging from 0 to 30 mM were applied across the plate with 9 replicate wells per condition.
Spheroids could only be located close to the coverslip in 42 of these wells. e total acquisition time for 42 elds
of view including translation of the stage between elds was 9 minutes and the FRET ratio images are shown in
Fig.7. An MDS analysis of these data is presented in Fig.8, which shows that there is a gradual increase in FRET
ratio for the surface of the spheroids with increasing concentration of glucose in the culture medium. For inner
regions of the spheroids this increase is more pronounced and has greater heterogeneity.
We compared the FRET ratios obtained at the rst time-point from all three plates described above plus the
results from a further experiment (plate 4), which were all prepared and imaged on dierent days. e data are
shown as a bar chart in SupplementaryFigureS7 and the standard deviation across all measurements and the
intra- and inter-plate standard deviations are shown in Table1. e standard deviation in FRET ratio was lowest
for the outside of the spheroids (MDS 0–10 μ m) with 0 mM glucose in the culture medium for all three meas-
ures of standard deviation. e standard deviations were larger in the presence of 25 mM glucose in the culture
medium. For the interior of the spheroids (MDS 60–70 μ m) the standard deviations were larger than for the
outside but followed similar trends.
In order to establish the dynamic range of the FLII12Pglu-700μ δ 6 biosensor on the ssOPM system, we
performed time-lapse imaging of spheroids in a culture medium containing 25 mM glucose where 50 μM
β -escin – a compound causing permeabilisation of the plasma membrane34 – was added at time t = 0, see
SupplementaryFigureS8. In spheroid 1 the FRET ratio for the outside of the spheroid (MDS 0–10 μ m) saturated
at 10.3 and for spheroid 3 the FRET ratio saturated at 12.3. For spheroid 2 a plateau was not obtained, but the
maximum value recorded was 13. We attribute the variation in response to dierences in cell permeabilisation by
escin in the dierent repeats causing dierences in the glucose inux rate relative to cellular glucose metabolism.
Finally, to check that the dynamics of the glucose response seen in Fig.6 are not determined by the rate at
which glucose diuses into the intercellular (interstitial) space between the cells forming the spheroid, we applied
ssOPM imaging to image diusion of the uorescent glucose analogue 2-NBDG as it is added to the culture
medium of a spheroid that has been starved of glucose. SupplementaryFigureS9 shows that the intercellular
glucose concentration (measured in a region of interest inside the spheroids) plateaus in ~5 min for 3 separate
repeats of the experiment.
e results obtained from HEK293T MCS expressing the FLII12Pglu-700μ δ 6 biosensor revealed that spheroids
cultured in the presence of glucose have a higher FRET ratio in the centre of the spheroid compared to the edges,
see Fig.4b. Although cells are generally thought to be able to maintain a tight regulation of intracellular pH even
when metabolically generating an extracellular acidication, we cannot exclude an intracellular acidication at
the core of the spheroids36, which may be partially responsible for the change in the FRET ratio observed. is is
because both the donor eCFP and acceptor Citrine in the FLII12Pglu-700μ δ 6 biosensor used here are known to
be sensitive to pH37,38. erefore, the increase in FRET ratio in the centres of the spheroids in the presence of glu-
cose in the culture medium may be due to a pH gradient within the spheroid rather than an intracellular glucose
gradient. In order to investigate this observation further, the next step would be to perform repeat measurements
using a near-identical biosensor with no or very low anity for glucose as a negative control39.
Figure 7. Demonstration of ssOPM plate-reading capability on 42 wells for culture media with varying
glucose concentration (plate 3). Glucose concentration in the culture medium increases down the plate as
indicated on the right hand side. For each well, an x-z (top), y-z (le) and x-y (bottom right) slice is shown
through the spheroid.
Scientific RepoRts | 6:37777 | DOI: 10.1038/srep37777
Cells at the edges of the spheroids in our experiments are expected to experience the well-dened extracel-
lular pH of the culture medium (pH 7.4) and therefore for these cells changes in observed FRET ratio can be
ascribed to changes in intracellular glucose with greater condence. Previous work has shown that HEK293T
cells show a low response to external glucose and have a low endogenous glucose transport activity39. is is
consistent with the results that we obtain for the outside of the spheroids, see Fig.8a, i.e. even in the presence
of high glucose in the culture medium the increase in FRET ratio is small compared to saturation of the sensor
For spheroids initially exposed to 0 mM glucose that then have glucose added, see Figs4c and 6c, the initial
increase in FRET ratio is approximately uniform across the spheroids until at least t = ~20 min. Aer this time,
the FRET ratio for the outside of the spheroid decreases slightly and plateaus whereas the FRET ratio for the
interior of the spheroid continues to increase and plateaus at a higher value. is dierence in biosensor response
Figure 8. Quantication of FRET ratio for glucose titration (plate 3). (a) FRET ratio for the surface of each
spheroid (MDS range 0-10 μ m). (b) FRET ratio for an inner region of each spheroid (MDS range 60–70 μ m).
(c) Plot of mean FRET ratio across repeat wells as a function of MDS for all glucose concentrations.
MDS (μm)
Glucose concentration in culture
medium (mM)
Standard deviation in FRET ratio
Across all
spheroids Intra-plate Inter-plate
0–10 0 0.12 0.07 0.09
25 0.25 0.21 0.16
60–70 0 0.13 0.08 0.1
25 0.75 0.55 0.61
Table 1. Standard deviation in measured FRET ratio for dierent MDS values and glucose concentration
in culture medium for: all spheroids, intra-plate and inter-plate.
Scientific RepoRts | 6:37777 | DOI: 10.1038/srep37777
could be due to dierences in gene expression, e.g. of plasma membrane glucose transporter proteins, causing dif-
ferent intracellular glucose uptake dynamics for cells in the centre of the spheroid relative to those on the outside.
We note that prior to the experiment, the spheroids were cultured in 25 mM glucose and so would have time to
establish dierent protein expression levels due to dierent local extracellular environments with distance into the
spheroid. Another possibility is that the addition of glucose causes a pH gradient to be set up in the spheroid that
aects the readout of the biosensor used dynamically over time. Further experiments are required to investigate
this fully. Our results obtained using the uorescent glucose analogue 2-NBDG suggest that diusion of glucose
into the centre of the spheroid is relatively fast (plateaus in ~5 min, SupplementaryFigureS9), and therefore the
dierence in FRET ratio dynamics seen for the inside and outside of the spheroid are not likely to be due to the
dierence in local extracellular glucose concentration.
We compared the inter- and intra-plate variability in FRET ratio for the MDS ranges 0–10 and 60–70 μ m
(see SupplementaryFigureS7). Our results show that the FRET ratio for spheroids cultured in the absence of glu-
cose is the most consistent both within and between plates, see Table1. For spheroids cultured in the presence of
25 mM glucose, in the MDS range 0–10 μ m the variation in FRET ratio was higher than for 0 mM glucose, and the
variation increased further for the MDS range 60–70 μ m. ese trends can also be seen in Fig.8. Cells exposed to
0 mM are likely to have a uniformly low intracellular glucose and therefore the least biological heterogeneity. Cells
cultured with glucose or that are within the spheroid environment are likely to have the greater heterogeneity in
local environment.
In the control wells of our time-lapse experiment (plate 1, Fig.4a&b), we observed a variation in FRET ratio
around t = 0 that we attribute to the change in temperature caused by opening the door of the microscope enclo-
sure in order to add aliquots to the wells. In the future, any temperature change can be avoided by using a system
for remote addition of compounds to the plate and/or the use of an improved biosensor with reduced sensitivity
to temperature.
Our experimental protocol acquires data in three channels (donor, sensitised emission and directly excited
acceptor). In the future, so-called 3-cube FRET calculations40 can therefore be applied to ssOPM data enabling
more quantitative measurements of the status of the biosensor. is can be achieved in future by determining the
necessary correction factors using samples expressing the appropriate control constructs41.
Our ssOPM system is currently not able to image wells on the outermost edge of a 96-well plate due to the
collar supplying water immersion liquid to the tip of the microscope objective obstructing the movement of the
motorised stage. In the future, this problem can be avoided by modifying the mechanical design of the collar or
the stage insert.
We have demonstrated the potential of ssOPM to probe spatio-temporal dynamics in 3-D in MCS expressing a
FRET biosensor in commercially available glass-bottomed 96-well plates. We used HEK293T cells expressing the
FLII12Pglu-700μ δ 6 intercellular glucose biosensor read out using spectral ratiometric measurements. Sub-cellular
resolution imaging was obtained for imaging depths of ~100 μ m into a spheroid. We imaged 16 wells at 10 minute
intervals for 4 hours generating 530 GB of raw image data (plate 1). An independent repeat of this experiment on
a plate prepared and imaged on dierent days illustrates the repeatability of the method (plate 2). We illustrated
the potential of ssOPM to perform a glucose dose-response measurement in 3-D in a 96-well plate using 42 wells
that required 9 min for image acquisition (plate 3).
Our 3-D imaging approach revealed an increase in FRET ratio in the centre of spheroids cultured in the pres-
ence of glucose. is may be due to gradients of glucose or pH within the spheroid and further experiments and/
or the use of an improved biosensor is required to investigate this further. In addition, ssOPM was used to record
the spatio-temporal FRET ratio response as glucose-starved HEK293T spheroids were exposed to glucose. Using
the 3-D image data acquired, we calculated the FRET ratio as a function of the minimum distance of an image
point to the surface of the spheroid as a function of time. In the future, ssOPM may nd application to studying a
wide variety of FRET biosensors in 3-D cell cultures across a wide-range of conditions in commercially available
96-well plates.
1. LaBarbera, D. V., eid, B. G. & Yoo, B. H. e multicellular tumor spheroid model for high-throughput cancer drug discovery.
Expert Opinion on Drug Discovery 7, 819–830, doi: 10.1517/17460441.2012.708334 (2012).
2. unz-Schughart, L. A. e Use of 3-D Cultures for High-roughput Screening: e Multicellular Spheroid Model. Journal of
Biomolecular Screening 9, 273–285, doi: 10.1177/1087057104265040 (2004).
3. Zschener, O., Streichert, T., Hehlgans, S. & Cordes, N. Genome-Wide Gene Expression Analysis in Cancer Cells eveals 3D
Growth to Aect ECM and Processes Associated with Cell Adhesion but Not DNA epair. PLoS ONE 7, e34279, doi: 10.1371/
journal.pone.0034279 (2012).
4. Indovina, P., Collini, M., Chirico, G. & Santini, M. T. ree-dimensional cell organization leads to almost immediate HE activity
as demonstrated by molecular imaging of MG-63 spheroids using two-photon excitation microscopy. Febs Lett 581, 719–726,
doi: 10.1016/j.febslet.2007.01.040 (2007).
5. ostadinova, . et al. A long-term three dimensional liver co-culture system for improved prediction of clinically relevant drug-
induced hepatotoxicity. Toxicol Appl Pharmacol 268, 1–16, doi: 10.1016/j.taap.2013.01.012 (2013).
6. Zietarsa, M. et al. Molecular description of a 3D in vitro model for the study of epithelial ovarian cancer (EOC). Mol Carcinogen 46,
872–885, doi: 10.1002/mc.20315 (2007).
7. Gunness, P. et al. 3D organotypic cultures of human HepaG cells: a tool for in vitro toxicity studies. Toxicol Sci 133, 67–78,
doi: 10.1093/toxsci/021 (2013).
8. Mueller, D., rämer, L., Homann, E., lein, S. & Noor, F. 3D organotypic HepaG cultures as in vitro model for acute and repeated
dose toxicity studies. Toxicol In Vitro 28, 104–112, doi: 10.1016/j.tiv.2013.06.024 (2014).
9. Wenzel, C. et al. 3D high-content screening for the identication of compounds that target cells in dormant tumor spheroid regions.
Exp Cell es 323, 131–143, doi: 10.1016/j.yexcr.2014.01.017 (2014).
Scientific RepoRts | 6:37777 | DOI: 10.1038/srep37777
10. Li, I. T., Pham, E. & Truong, . Protein biosensors based on the principle of uorescence resonance energy transfer for monitoring
cellular dynamics. Biotechnol Lett 28, 1971–1982, doi: 10.1007/s10529-006-9193-5 (2006).
11. Welch, C. M., Elliott, H., Danuser, G. & Hahn, . M. Imaging the coordination of multiple signalling activities in living cells. Nat ev
Mol Cell Bio 12, 749–756, doi: 10.1038/nrm3212 (2011).
12. Wittig, . et al. Biosensor-expressing spheroid cultures for imaging of drug-induced eects in three dimensions. J Biomol Screen 18,
736–743, doi: 10.1177/1087057113480525 (2013).
13. Weber, P. et al. Monitoring of apoptosis in 3D cell cultures by FET and light sheet uorescence microscopy. Int J Mol Sci 16,
5375–5385, doi: 10.3390/ijms16035375 (2015).
14. Chennell, G. et al. Imaging of Metabolic Status in 3D Cultures with an Improved AMP FET Biosensor for FLIM. Sensors (Basel)
16, doi: 10.3390/s16081312 (2016).
15. Hirata, E. et al. In vivo uorescence resonance energy transfer imaging reveals dierential activation of ho-family GTPases in
glioblastoma cell invasion. J Cell Sci 125, 858–868, doi: 10.1242/jcs.089995 (2012).
16. Li, L., Zhou, Q., Voss, T. C., Quic, . L. & LaBarbera, D. V. High-throughput imaging: Focusing in on drug discovery in 3D.
Methods 96, 97–102, doi: 10.1016/j.ymeth.2015.11.013 (2016).
17. Frongia, C. et al. 3D imaging of the response to CDC25 inhibition in multicellular spheroids. Cancer Biol er 8, 2230–2236 (2009).
18. Scherer, . M. et al. Three-dimensional imaging and uptae of the anticancer drug combretastatin in cell spheroids and
photoisomerization in gels with multiphoton excitation. J Biomed Opt 20, doi: 10.1117/1.Jbo.20.7.078003 (2015).
19. Ivanov, D. P. et al. In vitro co-culture model of medulloblastoma and human neural stem cells for drug delivery assessment.
J Biotechnol 205, 3–13, doi: 10.1016/j.jbiotec.2015.01.002 (2015).
20. Lorenzo, C. et al. Live cell division dynamics monitoring in 3D large spheroid tumor models using light sheet microscopy. Cell Div
6, doi: 10.1186/1747-1028-6-22 (2011).
21. Pampaloni, F., Ansari, N. & Stelzer, E. H. High-resolution deep imaging of live cellular spheroids with light-sheet-based uorescence
microscopy. Cell Tissue es 352, 161–177, doi: 10.1007/s00441-013-1589-7 (2013).
22. Galland, . et al. 3D high- and super-resolution imaging using single-objective SPIM. Nat Methods 12, 641–644, doi: 10.1038/
nmeth.3402 (2015).
23. Strnad, P. et al. Inverted light-sheet microscope for imaging mouse pre-implantation development. Nat Methods 13, 139–142,
doi: 10.1038/nmeth.3690 (2016).
24. McGorty, . et al. Open-top selective plane illumination microscope for conventionally mounted specimens. Opt Express 23,
16142–16153, doi: 10.1364/OE.23.016142 (2015).
25. Gualda, E. J. et al. SPIM-fluid: open source light-sheet based platform for high-throughput imaging. Biomed Opt Express 6,
4447–4456, doi: 10.1364/Boe.6.004447 (2015).
26. Paie, P., Bragheri, F., Bassi, A. & Osellame, . Selective plane illumination microscopy on a chip. Lab Chip 16, 1556–1560,
doi: 10.1039/c6lc00084c (2016).
27. Dunsby, C. Optically sectioned imaging by oblique plane microscopy. Optics Express 16, 20306–20316, doi: 10.1364/Oe.16.020306
28. umar, S. et al. High-speed 2D and 3D uorescence microscopy of cardiac myocytes. Optics express 19, 13839–13847 (2011).
29. Siel, M. B. et al. High speed sCMOS-based oblique plane microscopy applied to the study of calcium dynamics in cardiac
myocytes. J Biophotonics 9, 311–323, doi: 10.1002/jbio.201500193 (2016).
30. Botcherby, E. J., Jusaitis, ., Booth, M. J. & Wilson, T. Aberration-free optical refocusing in high numerical aperture microscopy.
Opt Lett 32, 2007–2009, doi: Doi 10.1364/Ol.32.002007 (2007).
31. Taanaga, H., Chaudhuri, B. & Frommer, W. B. GLUT1 and GLUT9 as major contributors to glucose inux in HepG2 cells identied
by a high sensitivity intramolecular FET glucose sensor. Biochim Biophys Acta 1778, 1091–1099, doi: 10.1016/j.bbamem.2007.
11.015 (2008).
32. Afzal, I., Cunningham, P. & Naalin, . J. Interactions of ATP, oestradiol, genistein and the anti-olestrogens, faslodex (ICI 182780)
and tamoxifen, with the human erythrocyte glucose transporter, GLUT1. Biochem J 365, 707–719, doi: Doi 10.1042/Bj20011624
33. Jordan, N. J. & Holman, G. D. Photolabeling of the Liver-Type Glucose-Transporter Isoform Glut2 with an Azitriuoroethylbenzoyl-
Substituted Bis-D-Mannose. Biochem J 286, 649–656 (1992).
34. John, S. A., Ottolia, M., Weiss, J. N. & ibalet, B. Dynamic modulation of intracellular glucose imaged in single cells using a FET-
based glucose nanosensor. Pug Arch Eur J Phy 456, 307–322, doi: 10.1007/s00424-007-0395-z (2008).
35. Villoing, A. et al. Complex Fluorescence of the Cyan Fluorescent Protein: Comparisons with the H148D Variant and Consequences
for Quantitative Cell Imaging†. Biochemistry 47, 12483–12492, doi: 10.1021/bi801400d (2008).
36. Swietach, P., Vaughan-Jones, . D., Harris, A. L. & Huliova, A. e chemistry, physiology and pathology of pH in cancer. Philos
Trans  Soc Lond B Biol Sci 369, 20130099, doi: 10.1098/rstb.2013.0099 (2014).
37. Miyawai, A. & Tsien, . Y. Monitoring protein conformations and interactions by uorescence resonance energy transfer between
mutants of green uorescent protein. Method Enzymol 327, 472–500 (2000).
38. Griesbec, O., Baird, G. S., Campbell, . E., Zacharias, D. A. & Tsien, . Y. educing the environmental sensitivity of yellow
uorescent protein - Mechanism and applications. J Biol Chem 276, 29188–29194, doi: 10.1074/jbc.M102815200 (2001).
39. Taanaga, H. & Frommer, W. B. Facilitative plasma membrane transporters function during E transit. FASEB J 24, 2849–2858,
doi: 10.1096/.09-146472 (2010).
40. Hoppe, A., Christensen, . & Swanson, J. A. Fluorescence resonance energy transfer-based stoichiometry in living cells. Biophys J
83, 3652–3664 (2002).
41. Hou, B. H. et al. Optical sensors for monitoring dynamic changes of intracellular metabolite levels in mammalian cells. Nat Protoc
6, 1818–1833, doi: 10.1038/nprot.2011.392 (2011).
e authors gratefully acknowledge funding from the UK Engineering and Physical Sciences Research Council
(EP/H03238X/1, EP/K503733/1) and Biotechnology and Biological Sciences Research Council (BB/I023801/1,
BB/M006786/1). Vincent Maioli gratefully acknowledges an EPSRC funded PhD studentship and George
Chennell is the recipient of an interdisciplinary, cross-campus collaborative PhD studentship funded jointly by
the MRC Clinical Sciences Centre and Imperial College London.
Author Contributions
V.M. developed the stage-scanning OPM system with assistance from S.K. performed the imaging experiments
and analysed the data. V.M., G.C., A.S., D.C. and C.D. designed the experiments. G.C. and T.L. generated the
stable cell lines used and prepared the multicellular spheroids. H.S. performed the 2-NBDG glucose analogue
experiment and prepared the corresponding gure. C.D. draed the manuscript and V.M. prepared the gures.
All authors critically reviewed and contributed to the manuscript.
Scientific RepoRts | 6:37777 | DOI: 10.1038/srep37777
Additional Information
Supplementary information accompanies this paper at
Competing nancial interests: CD has led a patent on OPM.
How to cite this article: Maioli, V. et al. Time-lapse 3D measurements of a glucose biosensor in multicellular
spheroids by light sheet uorescence microscopy in commercial 96-well plates. Sci. Rep. 6, 37777; doi: 10.1038/
srep37777 (2016).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
is work is licensed under a Creative Commons Attribution 4.0 International License. e images
or other third party material in this article are included in the article’s Creative Commons license,
unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license,
users will need to obtain permission from the license holder to reproduce the material. To view a copy of this
license, visit
© e Author(s) 2016
... LSFM is a powerful tool for imaging biological processes inside of organoids, including drug uptake [10,17], glucose biosensing [18], cell division dynamic monitoring [16] and spheroid formation [19]. LSFM has also been used to analyze microfluidic devices in applications such as drug screening and precision medicine [20] (for a review, see [21]). ...
Full-text available
Three-dimensional imaging of live processes at a cellular level is a challenging task. It requires high-speed acquisition capabilities, low phototoxicity, and low mechanical disturbances. Three-dimensional imaging in microfluidic devices poses additional challenges as a deep penetration of the light source is required, along with a stationary setting, so the flows are not perturbed. Different types of fluorescence microscopy techniques have been used to address these limitations; particularly, confocal microscopy and light sheet fluorescence microscopy (LSFM). This manuscript proposes a novel architecture of a type of LSFM, single-plane illumination microscopy (SPIM). This custom-made microscope includes two mirror galvanometers to scan the sample vertically and reduce shadowing artifacts while avoiding unnecessary movement. In addition, two electro-tunable lenses fine-tune the focus position and reduce the scattering caused by the microfluidic devices. The microscope has been fully set up and characterized, achieving a resolution of 1.50 μm in the x-y plane and 7.93 μm in the z-direction. The proposed architecture has risen to the challenges posed when imaging microfluidic devices and live processes, as it can successfully acquire 3D volumetric images together with time-lapse recordings, and it is thus a suitable microscopic technique for live tracking miniaturized tissue and disease models.
... Unfortunately, mechanical scanning is inherently slow. For example, a single 96 well plate (8 x 12 cm² in size) typically requires approximately 8 minutes to scan [20,21]. In addition, while sequential scanning works well for static samples, it is not an option for rapidly moving samples. ...
Full-text available
This article experimentally examines different configurations of a novel multi-camera array microscope (MCAM) imaging technology. The MCAM is based upon a densely packed array of "micro-cameras" to jointly image across a large field-of-view at high resolution. Each micro-camera within the array images a unique area of a sample of interest, and then all acquired data with 54 micro-cameras are digitally combined into composite frames, whose total pixel counts significantly exceed the pixel counts of standard microscope systems. We present results from three unique MCAM configurations for different use cases. First, we demonstrate a configuration that simultaneously images and estimates the 3D object depth across a 100 x 135 mm^2 field-of-view (FOV) at approximately 20 um resolution, which results in 0.15 gigapixels (GP) per snapshot. Second, we demonstrate an MCAM configuration that records video across a continuous 83 x 123 mm^2 FOV with two-fold increased resolution (0.48 GP per frame). Finally, we report a third high-resolution configuration (2 um resolution) that can rapidly produce 9.8 GP composites of large histopathology specimens.
... The use of 3D cell culture techniques to model aspects of tumour biology not captured in 2D monolayer models has been growing in interest over the last decade and has benefited from the convergence of advanced imaging technologies [14] , 3D bioprinting [15] , organoid tissue culture [16] and microfluidic devices [17] . Examples range from simple 3D spheroid models of free-floating cell aggregates [18] to more complex multicellular organoid and organ-on-a-chip devices. ...
Full-text available
Three dimensional models of cell culture enables researchers to recreate aspects of tumour biology not replicated by traditional two dimensional techniques. Here we describe a protocol to enable automated high throughput phenotypic profiling across panels of patient derived glioma stem cell spheroid models. We demonstrate the use of both live/dead cell end-points and monitor the dynamic changes in the cell cycle using cell lines expressing the FUCCI cell cycle reporter. Together, these assays provide additional insight into the mechanism of action of compound treatments over traditional cell viability assay endpoints.
... The LSM system at NPL uses a pair of water dipping objectives with samples typically mounted within agarose or Matrigel on top of a selective plane illumination microscope (SPIM), where a gel-embedded sample sits in a water-filled "imaging pocket". Single objective variants, such as the oblique plane microscopy method which AZ is currently exploring in collaboration with Imperial College London (Maioli et al. 2016), allow imaging of samples mounted on glass slides and in multiwell plates. These details of the instrument geometries and mounting method need to be captured in the image metadata to provide a meaningful insight into the experiment. ...
Technical Report
Full-text available
The rapid expansion of technology and computing processes are facilitating research on unprecedented scales. The multitude of proprietary file formats and data storage systems make data location and sharing difficult. Many bioimaging modalities generate terabytes of imaging data per day and pose new challenges for data analysis and management. These challenges can be partially addressed by using standardised data descriptors (metadata) to capture scientific, regulatory and business-related features of a dataset. This report is aimed to highlight the importance of metadata, to provide an insight into the different metadata types and to compare the current bioimaging annotation practices at three high profile partner sites: AstraZeneca, GlaxoSmithKline and the National Physical Laboratory. The report is focussed on three life science imaging techniques: high-content screening, mass spectroscopy imaging, and light-sheet microscopy.
... Presently, AZ are exploring whether plate-based LSM (Maioli & Chennell, 2016) provides an advantage over established techniques such as wide-field or confocal microscopy. To our knowledge, GSK does not conduct any LSM imaging. ...
Technical Report
Full-text available
This survey has been conducted between October 2018 and April 2019 as a part of a collaborative project "Federation of Imaging Data for Life sciences research" (FIDL). During the survey, National Physical Laboratory (NPL) Data Science team interviewed scientists from AstraZeneca (AZ), GlaxoSmithKline (GSK) and NPL bio-metrology group to depict the landscape of the instruments and data management methods in three imaging domains: high content screening, mass spectrometry imaging and light-sheet microscopy. This report compares the commonly used instruments, data acquisition, data management and data processing strategies used in AZ, GSK and NPL laboratories.
... This is detrimental since the efficiency of organoid formation from single cells is particularly low (around 15% in the case of murine intestinal organoids 6 ). Furthermore, previous work on the live recording of organoid dynamics has focused either on specific biological questions 6,8,11 , organoid wide phenotype-driven screening approaches [12][13][14] , or on specific isolated tools 15 without a more generalized yet in-depth approach on light-sheet imaging and data analysis. In another work which aimed at creating a light-sheet organoid imaging platform 16 the focus was mainly on the determination of culture-wide heterogeneities through a combination of both lightsheet and wide-field techniques. ...
Full-text available
Organoids provide an accessible in vitro system to mimic the dynamics of tissue regeneration and development. However, long-term live-imaging of organoids remains challenging. Here we present an experimental and image-processing framework capable of turning long-term light-sheet imaging of intestinal organoids into digital organoids. The framework combines specific imaging optimization combined with data processing via deep learning techniques to segment single organoids, their lumen, cells and nuclei in 3D over long periods of time. By linking lineage trees with corresponding 3D segmentation meshes for each organoid, the extracted information is visualized using a web-based “Digital Organoid Viewer” tool allowing combined understanding of the multivariate and multiscale data. We also show backtracking of cells of interest, providing detailed information about their history within entire organoid contexts. Furthermore, we show cytokinesis failure of regenerative cells and that these cells never reside in the intestinal crypt, hinting at a tissue scale control on cellular fidelity. Live imaging of organoid growth remains a challenge: it requires long-term imaging of several samples simultaneously and dedicated analysis pipelines. Here the authors report an experimental and image processing framework to turn long-term light-sheet imaging of intestinal organoids into digital organoids.
... Epi-illumination SPIM (ssOPM) enabled parallel 3D volume acquisition of a single cell at the single molecule level in 96-well plates 19 . It also allowed the imaging of living oncospheres in 3D seeded in agarose to detect glucose uptake 20 , but only during a short period of time (3 hours). Moreover, the random positioning of the spheroids in the agarose limited the study to 42 oncospheres. ...
Full-text available
Current imaging approaches limit the ability to perform multi-scale characterization of three-dimensional (3D) organotypic cultures (organoids) in large numbers. Here, we present an automated multi-scale 3D imaging platform synergizing high-density organoid cultures with rapid and live 3D single-objective light-sheet imaging. It is composed of disposable microfabricated organoid culture chips, termed JeWells, with embedded optical components and a laser beam-steering unit coupled to a commercial inverted microscope. It permits streamlining organoid culture and high-content 3D imaging on a single user-friendly instrument with minimal manipulations and a throughput of 300 organoids per hour. We demonstrate that the large number of 3D stacks that can be collected via our platform allows training deep learning-based algorithms to quantify morphogenetic organizations of organoids at multi-scales, ranging from the subcellular scale to the whole organoid level. We validated the versatility and robustness of our approach on intestine, hepatic, neuroectoderm organoids and oncospheres. A method for high-content 3D imaging of organoids.
Fast volumetric imaging of large fluorescent samples with high-resolution is required for many biological applications. Oblique plane microscopy (OPM) provides high spatiotemporal resolution, but the field of view is typically limited by its optical train and the pixel number of the camera. Mechanically scanning the sample or decreasing the overall magnification of the imaging system can partially address this challenge, albeit by reducing the volumetric imaging speed or spatial resolution, respectively. Here, we introduce a novel dual-axis scan unit for OPM that facilitates rapid and high-resolution volumetric imaging throughout a volume of 800 × 500 × 200 microns. This enables us to perform volumetric imaging of cell monolayers, spheroids and zebrafish embryos with subcellular resolution. Furthermore, we combined this microscope with a multi-perspective projection imaging technique that increases the volumetric interrogation rate to more than 10 Hz. This allows us to rapidly probe a large field of view in a dimensionality reduced format, identify features of interest, and volumetrically image these regions with high spatiotemporal resolution.
Conference Paper
Dual-view Oblique Plane Microscopy (dOPM) enables single-objective multi-view light-sheet fluorescence microscopy. This talk introduces the dOPM concept and demonstrates optical resolution performance with exemplar 3D datasets of fluorescence bead samples and fixed multicellular spheroids.
Full-text available
A bstract Aberrations in cell geometry are linked to cell signalling and disease. For example, metastatic melanoma cells alter their shape to invade tissues and drive disease. Despite this, there is a paucity of methods to quantify cell shape in 3D and little understanding of the shape-space cells explore. Currently, most descriptions of cell shape rely on predefined measurements of cell regions or points along a perimeter. The adoption of 3D tissue culture and imaging systems in medical research has recently created a growing need for comprehensive 3D shape descriptions of cells. We have addressed this need using unsupervised geometric deep learning to learn shape representations of cells from 3D microscopy images of metastatic melanoma cells embedded in collagen tissue-like matrices. We used a dynamic graph convolutional foldingnet autoencoder with improved deep embedded clustering to simultaneously learn lower-dimensional representations and classes of 3D cell shapes from a dataset of more than 70,000 drug-treated melanoma cells imaged by high throughput light-sheet microscopy. We propose describing cell shape using 3D quantitative morphological signatures, which represent a cell’s similarity to shape modes in the dataset, and are a direct output from our model. We used the extracted features to reveal the extent of the cell shape landscape and found that the shapes learned could predict drug treatment (up to 86% accuracy) and cell microenvironment, and are explainable. In particular, we found strikingly similar deep learning shape signatures between cells treated with microtubule polymerisation inhibitors and branched actin inhibitors. Finally, we implemented our methods as a Python package for ease of use by the medical research community.
Full-text available
We describe an approach to non-invasively map spatiotemporal biochemical and physiological changes in 3D cell culture using Forster Resonance Energy Transfer (FRET) biosensors expressed in tumour spheroids. In particular, we present an improved Adenosine Monophosphate (AMP) Activated Protein Kinase (AMPK) FRET biosensor, mTurquoise2 AMPK Activity Reporter (T2AMPKAR), for fluorescence lifetime imaging (FLIM) readouts that we have evaluated in 2D and 3D cultures. Our results in 2D cell culture indicate that replacing the FRET donor, enhanced Cyan Fluorescent Protein (ECFP), in the original FRET biosensor, AMPK activity reporter (AMPKAR), with mTurquoise2 (mTq2FP), increases the dynamic range of the response to activation of AMPK, as demonstrated using the direct AMPK activator, 991. We demonstrated 3D FLIM of this T2AMPKAR FRET biosensor expressed in tumour spheroids using two-photon excitation.
Full-text available
Light sheet fluorescence microscopy has recently emerged as the technique of choice for obtaining high quality 3D images of whole organisms/embryos with low photodamage and fast acquisition rates. Here we present an open source unified implementation based on Arduino and Micromanager, which is capable of operating Light Sheet Microscopes for automatized 3D high-throughput imaging on three-dimensional cell cultures and model organisms like zebrafish, oriented to massive drug screening.
Full-text available
Oblique plane microscopy (OPM) is a form of light sheet microscopy that uses a single high numerical aperture microscope objective for both fluorescence excitation and collection. In this paper, measurements of the relative collection efficiency of OPM are presented. An OPM system incorporating two sCMOS cameras is then introduced that enables single isolated cardiac myocytes to be studied continuously for 22 seconds in two dimensions at 667 frames per second with 960 × 200 pixels and for 30 seconds with 960 × 200 × 20 voxels at 25 volumes per second. In both cases OPM is able to record in two spectral channels, enabling intracellular calcium to be studied via the probe Fluo-4 AM simultaneously with the sarcolemma and transverse tubule network via the membrane dye Cellmask Orange. The OPM system was then applied to determine the spatial origin of spontaneous calcium waves for the first time and to measure the cell transverse tubule structure at their point of origin. Further results are presented to demonstrate that the OPM system can also be used to study calcium spark parameters depending on their relationship to the transverse tubule structure.
Full-text available
The uptake of E -combretastatins, potential prodrugs of the anticancer Z -isomers, into multicellular spheroids has been imaged by intrinsic fluorescence in three dimensions using two-photon excited fluorescence lifetime imaging with 625-nm ultrafast femtosecond laser pulses. Uptake is initially observed at the spheroid periphery but extends to the spheroid core within 30 min. Using agarose gels as a three-dimensional model, the conversion of Z(trans)→E(cis) via two-photon photoisomerization is demonstrated and the location of this photochemical process may be precisely selected within the micron scale in all three dimensions at depths up to almost 2 mm. We discuss these results for enhanced tissue penetration at longer near-infrared wavelengths for cancer therapy and up to three-photon excitation and imaging using 930-nm laser pulses with suitable combretastatin analogs.
Full-text available
Single-objective selective-plane illumination microscopy (soSPIM) is achieved with micromirrored cavities combined with a laser beam-steering unit installed on a standard inverted microscope. The illumination and detection are done through the same objective. soSPIM can be used with standard sample preparations and features high background rejection and efficient photon collection, allowing for 3D single-molecule-based super-resolution imaging of whole cells or cell aggregates. Using larger mirrors enabled us to broaden the capabilities of our system to image Drosophila embryos. © 2015 Nature America, Inc. All rights reserved.
Selective plane illumination microscopy can image biological samples at a high spatiotemporal resolution. Complex sample preparation and system alignment normally limit the throughput of the method. Using femtosecond laser micromachining, we created an integrated optofluidic device that allows obtaining continuous flow imaging, three-dimensional reconstruction and high-throughput analysis of large multicellular spheroids at a subcellular resolution.
Despite its importance for understanding human infertility and congenital diseases, early mammalian development has remained inaccessible to in toto imaging. We developed an inverted light-sheet microscope that enabled us to image mouse embryos from zygote to blastocyst, computationally track all cells and reconstruct a complete lineage tree of mouse pre-implantation development. We used this unique data set to show that the first cell fate specification occurs at the 16-cell stage. © 2015 Nature Publishing Group, a division of Macmillan Publishers Limited. All Rights Reserved.
We have developed a new open-top selective plane illumination microscope (SPIM) compatible with microfluidic devices, multi-well plates, and other sample formats used in conventional inverted microscopy. Its key element is a water prism that compensates for the aberrations introduced when imaging at 45 degrees through a coverglass. We have demonstrated its unique high-content imaging capability by recording Drosophila embryo development in environmentally-controlled microfluidic channels and imaging zebrafish embryos in 96-well plates. We have also shown the imaging of C. elegans and moving Drosophila larvae on coverslips.
Single-objective selective-plane illumination microscopy (soSPIM) is achieved with micromirrored cavities combined with a laser beam-steering unit installed on a standard inverted microscope. The illumination and detection are done through the same objective. soSPIM can be used with standard sample preparations and features high background rejection and efficient photon collection, allowing for 3D single-molecule-based super-resolution imaging of whole cells or cell aggregates. Using larger mirrors enabled us to broaden the capabilities of our system to image Drosophila embryos.