Article Text

Fourier transform infrared spectroscopy imaging of live epithelial cancer cells under non-aqueous media
  1. JunYi Soh1,
  2. Adeline Chueng1,
  3. Aminat Adio1,
  4. Alan J Cooper1,
  5. Brian R Birch2,
  6. Bashir A Lwaleed3
  1. 1School of Pharmacy and Biomedical Sciences, Portsmouth University, Portsmouth, Hampshire, UK
  2. 2Department of Urology, University Hospital Southampton NHS Foundation Trust, Southampton, Hampshire, UK
  3. 3Faculty of Health Sciences, University of Southampton, Southampton, Hampshire, UK
  1. Correspondence to Dr Bashir A Lwaleed, Faculty of Health Sciences, University of Southampton, South Academic and Pathology Block (MP 11), Southampton General Hospital, Tremona Road, Southampton, Hampshire SO16 6YD, UK; bashir{at}


Aims Fourier transform infrared (FT-IR) imaging is increasingly being applied to biomedical specimens, but strong IR absorption by water complicates live cell imaging. This study investigates the viability of adherent epithelial cells maintained for short periods under mineral oils in order to facilitate live cell spectroscopy using FT-IR with subsequent imaging.

Methods The MGH-U1 urothelial or CaCo2 colorectal cancer cell lines were grown on plastic surfaces or mid-range infrared transparent windows. Medium in established cultures was replaced with paraffin mineral oil, or Fluorolube, for up to 2 h, and viability assessed by supravital staining. Drug handling characteristics were also assessed. Imaging of preparations was attempted by reflectance and transmission using a Varian FT-IR microscope.

Results Cells covered by mineral oil remained viable for 2 h, with recovery into normal medium possible. MTT ((3-(4,5-dimethylthlazol-2-yl)-2,5-diphenyl tetrazolium) conversion to crystalline formazan and differential patterns of drug uptake were maintained. The combination of a calcium fluoride substrate, Fluorolube oil, and transmission optics proved best for spectroscopy. Spectral features were used to obtain images of live cells.

Conclusions The viability of cells overlaid with IR transparent oils was assessed as part of a technique to optimise conditions for FT-IR imaging. Images of untreated cells were obtained using both reflectance and transmission. This represents an effective means of imaging live cells by IR spectroscopy, and also means that imaging is not necessarily a terminal event. It also increases options for producing images based on real-time biochemistry in a range of in vitro experimental and ‘optical biopsy’ contexts.

Statistics from


Imaging cells has been of key importance in biomedical science, not least in the development of new drugs.1 Limitations of conventional bright-field cytochemistry are that it is usually a terminal event requiring fixation, and that extrinsic labels may alter the handling of compounds of interest and may be transferred or lost.

Mid-range infrared is identified by wave-numbers between 4000–400 cm−1 (wavelengths 2.5–25 μm), and is used for compound identification due to the molecular fingerprint region being located around 1450–600 cm−1 (λ6.9–16 667 μm). Fourier transform infrared (FT-IR) spectroscopy originated as a tool for use by chemists and in materials sciences.2 ,3 However, for nearly two decades, it has been applied to biomedical problems4 ,5 including three-dimensional structural studies of cellular molecules in vitro.5 Incorporating a microscope into the FT-IR instrument has permitted small areas to be examined and the collection of arrays of scans, so spatial information at the tissue level6 can be obtained. FT-IR has thus been used on fixed dried cells with a view to clinical application7 including disease diagnosis.8 Applications with practical potential have included observing spectral changes in virally transformed cells,9 ,10 identification of the multidrug-resistance phenotype in cancer cells11 and the discrimination of different prostate cancer cell lines.12 Further recent publications include subcellular localisation in fixed/dried material at least to the level of differentiating cytoplasmic and nuclear compartments13 and a relatively detailed image of a single cell.14

As with fluorescence microscopy, studying and imaging live cells is currently an active area of research. However, the intense IR absorption of the excess water inevitably present in culture medium poses a problem for those attempting live cell analysis. Both Kuimova10 and Miyamoto11 minimised this by growing cells directly onto components of the instrument. Hoi-Ying used high-resolution Synchrotron Radiation-based FT-IR microspectroscopy to obtain spectra from live cells demonstrating the non-toxicity of the radiation to cells but did not image them.15 A more recent description of an advanced technique, the application of IR surface plasmon spectroscopy by Yashunsky et al16 yielded real-time images of Madin-Darby canine kidney (MDCK) cells settled on a gold surface in a flow-chamber.

Another limitation of mid-range FT-IR is that due to the relatively long wavelength, there is limited spatial resolution, which is at best 3–10 microns.15 Concern about the ability of FT-IR to detect subtle pathological changes in cellular constituents17 may be valid but would not preclude its application to time-lapse studies of the cellular uptake of exogenous material, such as drugs. This is the application envisaged for the technique in our group, but the work presented here aims simply to investigate optical arrangements for imaging live cells by reflection or transmission using non-aqueous media to overlay the cells and, thereby, prevent desiccation on the instrument stage. The initial concept was to use paraffin mineral oil (liquid paraffin), used by IR spectroscopists as a ‘mulling’ agent to mix with compounds of interest and equally familiar, medicinally, as liquid paraffin, where it is clearly not toxic to buccal mucosa. Partial success with this material led us to try Fluorolube, which has superior IR transparency, albeit a more viscous liquid. For studies where it is necessary to layer medium below the oil, liquid paraffin remains the oil of choice.

Materials and methods

Culture conditions and viability assays

The cell lines used were the urothelial cancer cell line (MGH-U1) and, for two assays, the CaCo2 colorectal cancer cell line. Both are epithelial, show contact inhibition and are anchorage dependent. The media are RPMI1640 and Medium199 respectively, both supplemented with 10% fetal calf serum and antibiotics with glutamine (Sigma, UK). Incubation was at 37°C in a humidified atmosphere containing 5% CO2 in air. To determine the duration of cell viability in non-aqueous media, cells grown in 40 mm diameter culture-grade petri dishes had their medium replaced with paraffin mineral oil or Fluorolube for up to 120 min. The oils were replaced with 5 ml of conventional aqueous media before adherent cells were subjected to trypan blue or acridine orange/propidium iodide viability analysis according to previously published protocols.18 MTT (3-(4,5-dimethylthlazol-2-yl)-2,5-diphenyl tetrazolium; 500 µg/ml final concentration) was used as an alternative viability assay giving quantitative, but not spatial information. Formazan product was measured spectrophotometrically at 570 nm. The differential cytoplasmic/nuclear uptake between epirubicin and idarubicin19 ,20 was assessed after 2 h under oil in 35 mm petri dishes by underlying medium with 25 µg/ml drug between the oil and the cells before examining by confocal fluorescence microscopy. To maintain pH during the microscopy (in air), as much medium/oil as practical without delay was removed from the dishes and replaced by drug-free phosphate buffered saline (PBS).

Assessing the possibility of cell growth on crystalline windows (BaF2, CaF2 and ZnSe, Crystran, Poole, UK) was achieved by placing the windows in a 60 mm diameter culture-grade petri dish, and covering with cell suspension in medium. Gold foil was also tested as a relatively inexpensive and flexible substrate for reflectance microscopy, as were Nunc slide chambers sputter coated with gold-palladium using equipment in an electron microscopy laboratory.


To image live cells by FT-IR in non-aqueous media, chambers were detached from Nunc 4-well chamber slides and sealed to MirrIR (Kevley Technologies, Chesterland, Ohio, USA) reflective slides with a hot glue gun. The seal was tested with water, and residual leaks plugged from outside with silicone grease. For transmission FT-IR, cells were grown in individual Nunc chambers (cut with a hot scalpel) and fixed in place on 1 mm thick CaF2 windows. Ethanol (70%) was used for sterilisation. All culture assembly interiors were sterilised with 70% ethanol and washed exhaustively with medium. MGH-U1 cells were seeded onto the reflective slides. Live cultures were overlaid with paraffin mineral oil or Fluorolube immediately before imaging by reflectance or transmission FT-IR. Imaging was performed on the Varian 660-IR FT-IR microscope system.



Cells attached, assumed their normal morphology and proliferated on CaF2 and MirrIR surfaces. On BaF2 and ZnSe crystals cell adherence was generally radically impaired, with some batch variation. Cells did attach to gold foil and to sputter-coated surfaces, but imaging using white light gave poor results, and we were unable to get reflectance IR images. These surfaces could not be produced with a smoothness to match the MirrIR slides. Cells on CaF2 viewed with a standard inverted microscope are illustrated in figure 1 to illustrate adhesion and general morphology (figure 1).

Figure 1

Cell attachment, growth and morphology on calcium fluoride windows. MGH-U1 cells in medium, showing adherence and typical morphology. The image quality is adversely affected by the thickness and optical properties of the window.

Viability under mineral oils

Supravital staining showed cells surviving incubation at 37°C in a gassed incubator under either paraffin mineral oil or Fluorolube for up to 2 h. Paraffin mineral oil was easier to use, being less viscous, but had somewhat more intrusive intrinsic absorption characteristics. The FT-IR imaging illustrated here, therefore, was under Fluorolube. At 1 mm thick and 1 cm in diameter, CaF2 windows were unsatisfactory platforms for the confocal microscope, so conventional plastic dishes were used. Cells were grown on culture-grade plastic, but incubated under Fluorolube for 2 h before replacing this with PBS containing acridine orange and propidium iodide. Confocal fluorescence microscopy was then carried out. Figure 2 shows that both morphology and uptake of acridine orange is characteristic of viable, healthy cells despite oil immersion. A comparison of fluorescence intensity due to epirubicin accumulation over 2 h, with and without prior immersion under oil, including an untreated control dish, is illustrated in figure 3. Both drug distribution and intensity are maintained in the oil-treated cultures.

Figure 2

Viability of MGH-U1 cells after 2 h under fluorolube. Image from an upright Zeiss confocal microscope using a ×50 water immersion lens, split into four panels: (A) acridine orange fluorescence (green=live), (B) propidium iodide fluorescence (red=dead), (C) Differential interference contrast (DIC); note the moiré pattern in the DIC image from residual oil, (D) overlay of A–C. Supravital stains dissolved in phosphate buffered saline. This figure is only reproduced in colour in the online version.

Figure 3

Drug uptake after incubation under oil. CaCo2 cells grown in 35 mm plastic culture dishes; (A) incubated with epirubicin (25 µg/ml, 1 h); (B) incubated under oil for 2 h, then epirubicin introduced under the oil layer for 1 h; (C) untreated control plate (no drug or oil). The fluid was largely replaced with PBS (to maintain pH) for confocal microscopy. Excitation at 488 nm: detection using a propidium iodide filter set. This figure is only reproduced in colour in the online version.

The microplate MTT technique permits the quantification of replicate cultures of CaCo2 cells. Using a 6×10 format in a 96-well plate, with the outer wells water-filled to minimise evaporation effects, the optical density readings (ie, formazan formation) were stable between 0.67±0.030 standard error of the mean (SEM) and 0.73±0.023 over time periods from 0–3 h; at 4 h it had declined slightly to 0.58±0.019 (n=12 for all series).

The documented relative intracellular distribution of the anthracyclines epirubicin and idarubicin, nuclear uptake versus nuclear exclusion respectively, were replicated (figure 4) after a 2 h incubation of CaCo2 cells under oil.

Figure 4

Contrasting nuclear uptake of anthracyclines after incubation under oil. CaCo2 cells grown in 35 mm plastic culture dish; incubated under mineral oil (2 h), medium containing idarubicin (A) or epirubicin (B) (both at 25 µg/ml) introduced under oil and incubated (1 h). Fluid largely replaced with PBS (to maintain pH) for confocal microscopy. Excitation at 488 nm, detection using a propidium iodide filter set.

Infrared imaging

Reflectance mode on silver-coated slides

The images in figure 5 were acquired simultaneously on MirrIR slides under Fluorolube. The IR intensity look-up table runs from blue (lowest absorbance) to red (highest). The area of confluent cells tends to show at all wave-numbers, but with lower background at 1542 and 1644 cm−1 (λ6.49 and 6.08 μm), the so-called amide peaks.11 ,12 The visible light image is crude and unfamiliar, being reflectance. The correspondence of high FT-IR absorption areas and cells in the light image is approximate.

Figure 5

Reflectance images of live MGH-U1 cells growing in a chamber on a MirrIR slide under Fluorolube. The left panel is a white-light image showing location of cells. The block of four FT-IR images taken using peaks at the wave numbers specified above each. 1542 and 1644 cm−1 represent amide peaks, 961 and 1120 cm−1 are Fluorolube peaks and not directly representative of cellular material. The axes depict the detector number of the focal-plane array imaging system. This figure is only reproduced in colour in the online version.

Transmission mode on calcium fluoride

FT-IR imaging of live cells by transmission under Fluorolube on CaF2 windows (figure 6) gave a more familiar bright-field image; the effect is, however, of somewhat rounded cells, probably due to the length of time in the uncontrolled environment of the open unheated microscope stage during set-up and data acquisition, the latter taking >15 min). The FT-IR images at 1542 and 1644 cm−1 (λ6.49 and 6.08 μm) show a high signal-to-noise ratio, and correspond well with cells on the bright-field panel.

Figure 6

Transmission images of live MGH-U1 cells growing in a chamber on a calcium fluoride window under Fluorolube. The left panel is a bright-field image showing the location of the cells and some morphology. The block of four FT-IR images taken using peaks at the wave numbers specified above each. 1542 and 1644 cm−1 represent amide peaks, 961 and 1120 cm−1 are Fluorolube peaks and not directly representative of cellular material. The axes depict the detector number of the focal-plane array imaging system. This figure is only reproduced in colour in the online version.


The use of live cells in culture is well established in experimental biomedicine, and is becoming a feature of ex vivo drug sensitivity testing for individual tailoring of treatment regimens.21 ,22 FT-IR imaging has potential in this context, if the problems of specimen preparation can be adequately addressed. This study addresses the issue of minimising extraneous water in live preparations for analysis.

Of the crystalline windows examined as substrate, only CaF2 supported cell adhesion, good morphology and growth, although BaF2 and ZnSe are normally favoured for FT-IR work as transmitting a wider spectrum. We confined ourselves to unmodified windows; arguably coating BaF2 and ZnSe with agents such as poly-L-lysine or collagen might improve matters. CaF2 is a less favourable mineral in that it is not transparent to IR below 100 cm−1 (λ100 μm). However, its useful range takes in the amide peaks and the absorption peaks characteristic of the molecules of interest to our research group. These include epirubicin, which is our reference compound, as its uptake and localisation by autofluorescence are well characterised and can be used for comparative purposes.23 This suggests a practical application of the technique in the study of modes of multidrug resistance. Although uptake studies are not part of this paper, it is where the research leads, will be presented elsewhere, and is thus relevant to this methodological study.

We also acknowledge that gold is another substrate which has been used—for surface plasmon spectroscopy, with imaging.16 Here, real-time images of MDCK were derived from cells applied to a gold surface in a flow-chamber. The cells appeared to settle and spread within 2 h. Our experience with gold surfaces has been less satisfactory, possibly due to the cells used. The MDCK cell line is very long established (since 1958, only 8 years after HeLa cells) and probably equally robust. We have variously used RT112 and MGH-U1 urothelial cancer cells and the LnCaP prostate cancer cell line. These are also old, though some 20 years younger than MDCK and the use of MGH-U1 is certainly subject to frequent criticism due to likely contamination over time and overlapping characteristics with alternatively named cell lines.

The potential for increasing resolution to discriminate between subcellular compartments is currently confined to imaging dead material using attenuated total reflection FT-IR microscopy, requiring contact between the instrument and the specimen. Its applicability to live cells under oil remains untested, although adherent MGH-U1 cells survive microinjection well24 which involves rather more than contact, albeit at a single fine point. The limiting factor is probably the ability to touch the crystal down onto the specimen with minimal force.

Cells survive under the oils used for periods in excess of those needed solely for imaging; albeit in the controlled environment of an incubator (a heated stage was not available for imaging). The viability staining was performed after recovery into an aqueous environment, the images were therefore considered to be of live cells. This view is enhanced by the ability of the cells to metabolise MTT appropriately after contact with oil, and with the oil still acting as a barrier between medium and gas phase. More especially, contact with the oil does not necessarily distort internal cellular processes, as exemplified by the appropriate distribution of the anthracyclines epi- and idarubicin after immersion under oil. Furthermore, as exposure to oil is not necessarily a terminal event; recovery of cultures is possible. Sequential imaging is therefore practical, although the time required to acquire mid-range infrared images may preclude much time-lapse work. Most imaging applications are terminal events for cultures; however, this system could be used serially in some studies, if infection risks can be satisfactorily managed during microscopy.

Fluorolube is ideal for IR samples to be run from 4000 to 1360 cm−1 (λ2.50–7.35 μm) where it is non-absorbing, except at 2321.9 cm−1. Below 1360 cm−1 (λ above 7.35 μm), three further identifying peaks for Fluorolube are at 1191, 1120 and 961 cm−1. The amide bands I and II at 1644 and 1542 cm−1, away from these potentially interfering peaks, again detect the presence of cells. There was a particularly good association between the amide-based FT-IR and light images on CaF2, although this may be partly because transmission images of cells are more familiar than reflectance techniques. Images also obtained from Fluorolube absorption peaks did not pick out cells but showed an uneven background perhaps reflecting different thicknesses of the viscous Fluorolube layer. It was not practical to overlay the light and FT-IR images—a problem with instrument design still informed by chemistry more than biology. The system is a practical and low-cost way of examining live cells by FT-IR absorption. Furthermore, transmission gives better or more familiar results than reflectance. The temporary use of non-aqueous media holding live cells for analysis is an advantage not only applicable to imaging, but can enhance spot spectroscopy by eliminating the artefacts from fixation and drying in preserved preparations.

The method described here is one of a range of solutions to the water absorption problem, as alluded to in the introduction. We decline to refer to it as ‘better’, but it is simple to use and economical, compared with growing cells on crystalline lenses or suboptimal substrates such as gold, or indeed using more complex imaging equipment with, for instance, synchrotron sources.

In conclusion, this study demonstrates that the biochemical information available through the use of FT-IR can be applied to live adherent epithelial cells to yield data related to the spatial distribution of selected features with the additional advantage of subcellular definition.

Take-home message

▸ Aqueous culture media prove an impediment to mid-range FT-IR imaging of live cells. Holding cells temporarily under mineral oils does not compromise their viability. This procedure, in combination with CaF2 as substrate, can be used to image live adherent cells based on biochemical information available through IR spectroscopy, and is applicable to clinical material, such as exfoliated cells.

What the paper adds

▸ This study validates an effective and inexpensive solution to the problems of excess water causing large, confounding absorption peaks when attempting to image live cells in aqueous medium using mid-range infrared wavelengths. The opportunities mid-range infrared would give to the array of practical imaging modes applicable to live cells in experimental and clinical settings is that it allows for spatial real-time distribution of the uptake of exogenous agents to be visualised, based on their biochemical features.


This work would not have been possible without the donated use of instruments and technical assistance in their laboratories of Varian Inc (Oxford), and Perkin Elmer UK (Beaconsfield).

View Abstract


  • Contributors JS, AC and AA made a substantial contribution to the design, organisation and conduct of the study (including acquisition of study data). AJC made a substantial contribution to the conception, design, organisation and conduct of the study. Also drafted and critiqued the output for important intellectual content. BRB helped with critiquing the output for important intellectual content. He constitutes our reference point for clinical orientation and relevance of our research effort. BAL made a substantial contribution to the conception, design, organisation and conduct of the study. Also critiqued the output for important intellectual content.

  • Funding The study was funded internally. Access to specialist equipment was free and unconditional.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Additional viability studies including MTT (3-(4,5-dimethylthlazol-2-yl)-2,5-diphenyl tetrazolium) and regrowth of cultures after imaging, which may be touched on but not expanded in the article, can be elaborated upon. Spectroscopic and/or imaging studies involving other exogenous agents, including lycopene, mitomycin-c and cisplatin have also been carried out. Manuscripts detailing related results using near-infrared spectroscopy are in preparation and, while we would not wish to pre-empt or anticipate the definitive articles, we would be happy to share in general terms our experiences in this area.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.