Infrared Microspectroscopy of Live Cells in Microfluidic Devices (MD

Apr 23, 2012 - Therefore, only IRMS of live cells can provide reliable information on both .... Bulletin of the Chemical Society of Japan 2018 91 (5),...
0 downloads 0 Views 4MB Size
Article pubs.acs.org/ac

Infrared Microspectroscopy of Live Cells in Microfluidic Devices (MDIRMS): Toward a Powerful Label-Free Cell-Based Assay L. Vaccari,*,† G. Birarda,†,‡ L. Businaro,§ S. Pacor,∥ and G. Grenci‡ †

Elettra Synchrotron Light Laboratory, SISSI beamline, S.S. 14 Km 163.5, 34149 Basovizza, Trieste, Italy IOM-CNR, TASC Laboratory, S.S. 14 Km 163.5, 34149 Basovizza, Trieste, Italy § CNR−Institue for Photonics and Nanotechnologies, via Cineto Romano, 42, 00156 Roma, Italy ∥ Life Science Department, Trieste University, via Licio Giorgieri 7-9, 34127, Trieste, Italy ‡

S Supporting Information *

ABSTRACT: Until nowadays most infrared microspectroscopy (IRMS) experiments on biological specimens (i.e., tissues or cells) have been routinely carried out on fixed or dried samples in order to circumvent water absorption problems. In this paper, we demonstrate the possibility to widen the range of in-vitro IRMS experiments to vibrational analysis of live cellular samples, thanks to the development of novel biocompatible IR-visible transparent microfluidic devices (MD). In order to highlight the biological relevance of IRMS in MD (MD-IRMS), we performed a systematic exploration of the biochemical alterations induced by different fixation protocols, ethanol 70% and formaldehyde solution 4%, as well as air-drying on U937 leukemic monocytes by comparing their IR vibrational features with the live U937 counterpart. Both fixation and air-drying procedures affected lipid composition and order as well as protein structure at a different extent while they both induced structural alterations in nucleic acids. Therefore, only IRMS of live cells can provide reliable information on both DNA and RNA structure and on their cellular dynamic. In summary, we show that MD-IRMS of live cells is feasible, reliable, and biologically relevant to be recognized as a label-free cell-based assay.

L

technologies. Fourier transform infrared (FTIR) spectroscopy has been largely employed from the 1950s for the investigation of the most relevant biological molecules5−7 and, since mid1980s, with the development of commercial visible-infrared (Vis-IR) microscopes, infrared microspectroscopy (IRMS) established itself as a mature tool for life science studies. IRMS provides the advantage to correlate the morphological features of a biosample with its vibrational local pattern, that is the biomarker for structure and/or composition of cellular lipids, proteins, carbohydrates and nucleic acids.6,7 Sensitive, noninvasive, and nonradiation damaging8 IRMS has been mainly performed on fixed biosamples: tissue slides and cells have been studied following a comparative approach based on the technique capability to distinguish between different cellular

abel-free analytical methodologies for real-time monitoring of live cells represent the new frontier of cell-based assays, offering unique opportunities to cellular biology. Aiming to understand cell function and structure through the study of the complex network of macromolecular interactions at the base of cell life, modern cell biology is looking with increasing interest for noninvasive investigation tools. As a matter of fact, the use of exogenous probes, such as radioactive, colorimetric, fluorescent labels,1 or others, inevitably alters the cell environment possibly tuning the cellular response2 while fixation protocols affect cellular morphology and/or biochemistry,3 therefore potentially limiting the capabilities of any bioanalytical tool. Label-free cell-based assays are generally established on optical, acoustic, magnetic, or electrical biosensors able to transduce the response of living cells in measurable signals.4 However, their interpretation is often troublesome and the biocommunity is looking for more straightforward readout © 2012 American Chemical Society

Received: February 1, 2012 Accepted: April 23, 2012 Published: April 23, 2012 4768

dx.doi.org/10.1021/ac300313x | Anal. Chem. 2012, 84, 4768−4775

Analytical Chemistry

Article

states, such as normal and cancer tissue areas,9,10 different cell cycle stages,11,12 prion-infected and normal cells13 or brain regions,14−16 as well as drug-treated and untreated samples,17−20 and many others.21 The possibility to monitor live cells under physiological conditions for tracking their biochemical modification either induced or naturally occurring would be a turning point in the full exploitation of IRMS diagnostic capabilities. However, in-vitro IRMS experiments are still in their infancy, due to (i) the unavailability of suitable fluidic devices and (ii) the lack of standardized procedures for the compensation of water spectral features prominent in live cell spectra. Some groups already succeeded in building demountable liquid devices for live cell IRMS by spacing apart two IR optical windows, usually CaF2, ZnSe, or diamond, with plastic spacers 22−29 or micromachining thin reservoirs onto them,30,31 and an open-channel geometry was recently proposed by Holman and co-workers.32,33 While the former approach does not guarantee any flexibility in the design of the fluidic device, limiting its layout to a “pond” for the cells, the latter exploits advanced fabrication techniques and seems to be less stressful for the sample; however, due to the materials employed, it is more suitable for transflectance measurements, that could introduce spectral artifacts, consequently complicating the data postprocessing with respect to pure transmission measurements.34−36 For overcoming these issues, we extended microfabrication concepts and technologies to Vis-IR transparent materials,37,38 and in this paper, we present the first example of a fully sealed CaF2 microfluidic device (MD) for live cell analysis in transmission mode by IRMS. This approach has the clear advantage to allow monitoring a live cell in real-time and finally answer a fundamental question on the biochemical alterations induced by fixation as revealed by IRMS. In this paper, we address this problem by performing a systematic comparison of the spectral features of live U937 leukemic monocytes with the same cells differently fixed or unfixed air-dried. Our results demonstrate that formalin fixative is preserving proteins and lipids in a status closer to the one of viable cells while ethanol fixation and air-drying are not. However, the nucleic acid structure is affected by all fixation protocols; thus MD-IRMS of live cells is the only pursuable approach for achieving reliable information on nucleic acid structure and dynamic evolution.

day after. Cell concentration has been set in order to produce a densely packed monolayer of cells. Air Drying. Once the complete medium with phosphate buffered saline buffer (PBS) was substituted by centrifugation (400g, 5 min), 30 μL of the cell suspension 10 × 106 cells/mL were dropped onto a CaF2 window 2 mm thick and U937 monocytes were allowed to settle at 4 °C for 2 h, then at room temperature for 1 h. The PBS excess was then removed; the cells were briefly rinsed twice with sterile physiological solution and dried at room temperature for 2 h before data collection. The cell concentration has been set in order to produce a densely packed monolayer of cells. Live Cells. Just before the measurement, U937 monocytes were collected from the culture medium, centrifuged at 400g for 1 min and resuspended in fresh medium. The procedure was repeated twice achieving a final concentration of 10 × 106 cells/ mL. The cell suspension was used as it was for the experiments and injected into the microfluidic device by a syringe pump. The cell concentration was set in order to fill almost homogenously the inner chamber of the MD (see Figure 1).

Figure 1. Sketch of the layout of the fluidic device employed for the measurements. On the top of the 2 mm thick CaF2 window (gray disk), a layer of XARP 3100/10 resist (orange features) was spun at 8.5 μm and lithographed in order to define two concentric chambers, b and c, separated by a porous septum. The inner chamber c accommodates the cell suspension while the b contains fresh culture medium. The a regions are isolated, water-free wells where air background is collected.

Microfluidic Device and Experimental Setup. Circular CaF2 windows (Crystran UK), 30 mm in diameter, 2 and 1 mm thick, were used as substrate and lid, respectively. The device is made by two concentric chambers (b and c regions in Figure 1), defined in XARP 3100/10 resist (AllResist GmbH) 8.5 μm thick on the substrate window, separated by a porous septum (3 μm interspace). Within the device, four regions water-free were lithographed for collecting air background (a wells in Figure 1). The lid was thermomechanically sealed on top of the bottom window using a hot-press in order to obtain a fully sealed microfluidic device. For more details on the fabrication steps and on the device connection with the pumping and thermalizing systems, see Supporting Information section “Microfluidic device fabrication and connection”, Figures S-1 and S-2. The biocompatibility of the materials used was also tested before the experiments, as detailed in Supporting Information section “Cell viability analysis”. IRMS Data Collection and Preprocessing. U937 cell monolayers were investigated at the end-station of the SISSI beamline40 biobranch at Elettra Synchrotron (Trieste, Italy) equipped with a Vis-IR Hyperion 3000 microscope coupled with a Vertex 70 interferometer (Bruker Optics GmbH, Ettlingen, Germany). Both interferometer and microscope,



MATERIALS AND METHODS Cell Culture and Fixation Protocols. Monocyte cell line U937 (American Type Culture Collection, Rockville, Md.)39 was chosen for the experiments. Cells were cultured in RPMI medium (RPMI 1640, 2 mM L-glutamine and 10% FBS), with 100 U/mL penicillin and 100 μg/mL streptomycin, in an incubator at 37 °C with 5% CO2. Formalin and Ethanol Fixation. Once the complete medium with phosphate buffered saline buffer (PBS) was substituted by centrifugation (400g, 5 min), 30 μL of the cell suspension 10 × 106 cells/mL were dropped onto a CaF2 window 2 mm thick and U937 monocytes were allowed to settle for 2 h at 4 °C, in order to lower membrane fluidity and promote the layering of these circulating cells. The liquid excess was then removed, the fixative solution was added (4% formaldehyde solution or ethanol 70%) and let to act at room temperature for 20 min. Once the fixative excess was removed, the cells were briefly rinsed twice with ultrapure water, dried at room temperature overnight, and measured the 4769

dx.doi.org/10.1021/ac300313x | Anal. Chem. 2012, 84, 4768−4775

Analytical Chemistry

Article

Table 1. Band Integrals and Their Assignment wavenumber range (cm−1)

integral type

2990−2830 2995−2947 2947−2882 2882−2863 2863−2840 1705−1582 1582−1480

peak area massif massif massif massif peak area peak area

baseline range (cm−1) straight straight straight straight straight straight straight

baseline baseline baseline baseline baseline baseline baseline

assignment

3000−2800 3000−2800 3000−2800 3000−2800 3000−2800 1800−1000 1800−1000

symmetric and asymmetric stretching of methyl and methylene moieties asymmetric stretching of methyl moieties, CH3asym asymmetric stretching of methylene moieties, CH2asym symmetric stretching of methyl moieties, CH3sym symmetric stretching of methylene moieties, CH2sym amide I band amide II band

cm−1, and carbonyl ester of phospholipids, 1750−1720 cm−1) and nucleic acids−carbohydrates (1280−1000 cm−1). HCA was run on vector normalized second derivatives of spectra (nine smoothing points, Savitsky−Golay algorithm), chosen in order to enhance the spectral resolution and minimize baseline and offset variations between spectra.

sealed with an in-house designed box, were purged with nitrogen in order to reduce spectral contributions for environmental water vapor and carbon dioxide. Densely packed 50 μm × 50 μm cellular monolayer areas were first visually chosen and then infrared spectra were collected in transmission mode with conventional Globar source by using a 15× Schwarzschild objective (NA = 0.4, working distance = 24 mm) matched with a 15× condenser in the Mid-IR regime with a single point Hg−Cd−Te detector (MCT, 250 μm). For each spectrum, 512 scans have been averaged with a spectral resolution of 4 cm−1 (zero-filling factor = 2). For all fixed cellular samples, background spectrum was acquired on a clean window portion and final absorbance spectra were corrected for water vapor and carbon dioxide contributions running the “atmospheric compensation” routine of OPUS NT 6.5 software (Bruker Optics GmbH, Ettlingen, Germany). These spectra were used for data postprocessing. Concerning live cell measurements, the air background was collected on the water-free regions of the device and the spectrum of a 50 μm × 50 μm area of the pure medium (Buf ferSpectrum) was taken with the same instrumental settings as close as possible to the acquired group of live cells (RawCellSpectrum). The RawCellSpectrum of a group of live U937 cells and the corresponding Buf ferSpectrum are shown in Figure S-3 in Supporting Information section “Water subtraction strategies”. Liquid water features dominate a RawCellSpectrum;41 in order to disclose the cellular bands, the amide I in particular, the Buf ferSpectrum has been subtracted by applying an appropriate scaling factor, Coef f (0.8 ≤ Coef f ≤ 1), chosen by running an inhouse automated algorithm that optimizes the baseline flatness of the corrected cell spectrum (CorrCellSpectrum) in the 1850− 2500 cm−1 range. For a more exhaustive explanation of the subtraction criterion and algorithm, see Supporting Information section “Water subtraction strategies”. IR Data Postprocessing: Statistical and Chemometric Analysis. Spectral band assignment was done accordingly to peer-reviewed papers.6,7 The integrals reported in Table 1 have been calculated by using OPUS 6.5 NT software (Bruker Optics GmbH, Ettlingen, Germany). In order to perform semiquantitative comparison between calculated quantities, integral values relative to each acquisition have been normalized to the number of cells measured, as obtained by the inspection of the optical images. For the chemometric analysis, the HyperSpecJSS program was employed42,43 (http://hyperspec.r-forge.r-project.org). The Hierarchical Cluster Analysis (HCA), based on Euclidean distances and Ward’s classification algorithm, was applied independently for three different regions, characteristic of cellular lipids (3050−2800 cm−1: mainly symmetric and asymmetric stretching bands of methyl and methylene groups), cellular proteins−phospholipids (1750−1480 cm−1: mainly amide I band, 1705−1582 cm−1, amide II band, 1582−1480



RESULTS AND DISCUSSION

The device used for collecting IR spectra of live cells presented in this article is the first example of a fully sealed microfluidic device employed for this aim. It is beyond the purpose of this manuscript to detail both the technical problems faced for the device microfabrication and the microfabrication steps, described in Supporting Information “Microfluidic device fabrication and connection”, while we want to highlight the advantages offered by fully sealed microfluidic devices with respect to demountable liquid ones, almost exclusively employed for IRMS of live cells up to now. Demountable liquid cells do not guarantee a perfect closure of the device, limiting long-term measurements, and do not ensure an optical path precisely controlled and constant through all the device, a fundamental prerequisite for increasing the water subtraction accuracy that is primarily affected by changes in path length between sample and buffer spectra.44 Our device does not suffer any of these limitations, while it obviously guarantees all the advantages offered by microfabrication, nonetheless the design flexibility and the possibility of integration within the chip of other functionalities that make the proposed strategy useful for almost any desired device design and experiment. The specific layout sketched in Figure 1 is made by an inner chamber where the U937 monocyte suspension in complete medium is injected using a syringe pump at a temperature of 37 °C. The cells are separated by a porous septum from an external chamber where a constant gentle flow of fresh medium is maintained in order to provide nutrients and oxygen to the cells through medium diffusion. U937 monocytes are round shaped cells with an average diameter of 8−10 μm:39,45 both shape and dimension are preserved within the device, that has a thickness of 8.5 μm, as can be appreciate from the optical image of a group of measured live cells (L-U937 hereafter) in Figure 2a, proving that no mechanical stress inducing cellular deformation and consequent biochemical alterations is suffered during the measurement.37 Once formalin fixed, U937 monocytes (referred as F-U937 henceforth) preserve their spherical shape but clearly look shrunken and smaller in size, as evident from the scanning electron microscopy (SEM) micrograph in Figure 2b. Ethanol fixed U937 cells (referred as E-U937 henceforth) have dimension comparable with L-U937, but they appear flattened and with irregular profiles at a closer view (see SEM micrograph in Figure 2c) while unfixed air-dried U937 (ADU937) have shape and dimensions similar to E-U937 but 4770

dx.doi.org/10.1021/ac300313x | Anal. Chem. 2012, 84, 4768−4775

Analytical Chemistry

Article

can be recognized (see Figure 3a): E-U937 greatly differ from L-U937, clustering closer to AD-U937 and F-U937, that group together. Such classification can be interpreted as claiming that ethanol fixation deeply affects the cellular lipids, while formalin and air drying preserve them better. The centroids of each class are shown in Figure 3d. The signal profiles of L-U937 and (FU937, AD-U937) centroids look quite similar in both intensity, position, and width of the second derivative peaks: methyl asymmetric and symmetric stretching bands are centered at 2960 ± 2 cm−1 and 2873 ± 2 cm−1, respectively, while methylene asymmetric and symmetric stretching are at 2923 ± 2 cm−1 and 2853 ± 2 cm−1. A pronounced shift toward higher wavenumbers of the methylene asymmetric stretching band at 2931 cm−1 as well as an increased bandwidth can be easily appreciated for E-U937, in agreement with Hastings et al.50 Both spectroscopic evidences, as well as the higher standard deviation exhibited by E-U937 with respect to all the other samples investigated, demonstrate an increased cellular membrane permeability and disorder. 53,54 Moreover, a remarkable decrement of methylene moieties with respect to methyl ones upon ethanol fixation can also be inferred from Figure 3d55 as well as from the comparison of representative spectra for each class in Figures 2e−h. The ratio of methylene to methyl asymmetric stretching, decreased by 27.8 ± 2.7% in comparison to L-U937, reflects an altered membrane composition of E-U937, richer in shorter and/or less saturated acyl chains, again responsible for an increased membrane permeability.56,57 The normalized total lipid content also dramatically changes upon ethanol fixation from 0.183 ± 0.028 to 0.064 ± 0.008 a.u. while it remaines almost unaltered upon formalin fixation (0.184 ± 0.020 a.u.) and air drying (0.194 ± 0.023 a.u.). The lipid-depletion induced by ethanol on phospholipids is even more evident, as can be appreciated from Figures 2e−h as well as Figure 3e: the carbonyl band of phospholipids, centered at 1746 ± 2 cm−1 for L- and AD-U937 and at 1742 ± 2 cm−1 F−U937, falls below the detection limit for E-U937, supporting the findings of Gazi et al.47 and Hastings et al.50 Actually, the position of this band reveals a prevalent trans conformation of the CC bond adjacent to ester CO group, also sensitive to the lipid hydration level as proven by the band broadening toward lower wavenumbers in L-U937.53 Taken together, this spectroscopic evidence allow us to draw a comprehensive picture of the effects of the diverse fixatives on cellular lipids. Ethanol is a rather polar solvent that quickly and easily penetrates cellular membranes. Membrane-associated lipids are mostly polar phospholipids with whom ethanol interacts solvating their hydrophilic heads similarly to water but disrupting the van der Waals interactions between adjacent acyl chains moving them away, consequently altering the membrane architecture and order.58 At the concentration used for fixation, higher than 50−60%, ethanol acts also as a solvent for the lipids, extracting them from cellular membranes, as highlighted by the decreased total lipid content. Conversely, formalin is preserving lipid content, order, and composition. It is known that formalin is influencing chemo-physical properties of lipids: it combines with unsaturated lipids at the double bond as well as it hydrolyzes phosphatides.59 However, the extent of decomposition is dependent on the fixation conditions (formalin concentration, pH of the solution, temperature and time of fixation)60 as well as on the cell type61 and cannot be appreciated when measuring F-U937 freshly prepared following the well-established protocol we have chosen. Air drying is not

Figure 2. (a) Optical image of a group of L-U937 cells within the microfluidic device (scale bar 10 μm). (b−d) SEM micrographs of F-, E-, and AD-U937, respectively (scale bar 5 μm). (e) IR spectrum of a group of twenty L-U937 as obtained after water contribution subtraction. (f−h) IR spectra of groups of F-, E-, and AD-U937, respectively, 20 cells each.

preserve clear defined thicker nuclear regions (see Figure 2d). These results are not surprising, since the effects of the most commonly employed cell fixatives as well as air-drying on both cellular matrix and morphology have been long studied and characterized.3,46 Several reports have also been published on the effects of different fixation methods on the biochemistry and architecture of both cells47 and tissues48−52 as revealed by IRMS, while the corresponding infrared vibrational pattern modifications with respect to live cells have not been yet, due to the lack of a valuable control, represented by live cell spectra. This represents the gap we aimed to fill by exploiting the capabilities of MD-IRMS. In particular, hiearchical cluster analysis (HCA) on second derivative of spectra was employed for highlighting the spectral heterogeneity induced by the different sample preparations. Within the 3050−2800 cm−1 lipid region, three major clusters 4771

dx.doi.org/10.1021/ac300313x | Anal. Chem. 2012, 84, 4768−4775

Analytical Chemistry

Article

Figure 3. (a−c) Dendrograms of the classification of vector normalized second derivatives of spectra from L-, F-, E-, and AD-U937 monocytes as obtained by HCA (Euclidean distances, Wards’ algorithm) in the regions of (a) lipids, (b) proteins−phospholipids, and (c) nucleic acids− carbohydrates; (d−f) centroids of the major classes identify by HCA in the (d) lipids, (e) proteins−phospholipids, and (f) nucleic acids− carbohydrates regions. Line thickness is proportional to standard deviation.

a standard procedure for dehydration of cellular samples, and it has been considered in this paper since it is often employed for probing biosamples using vibrational spectroscopies50,62−66 being a fast and chemical-free technique. When AD-U937 cells are measured immediately after the drying, that is within few hours from the preparation for limiting the autolysis events, the lipid spectral profile is not more discernible from F-U937. The moderate cellular swelling and the good preservation of the nuclear envelope of AD-U937 shown by SEM micrographs are indirect evidence of the spectroscopic results. U937 cells cluster pretty well in accordance with the sample preparation in the 1760−1480 cm−1 range, as can be seen in the dendrogram of Figure 3b. Even if the proteins−phospholipids region is the most affected by water subtraction procedure in bands’ intensity, the spectral frequencies and consequently also the results drawn out from second derivative analysis are not.30 The amide I band of L-U937 is centered at 1657 cm−1, a contribution that can be assigned to α-helix segments,6 and a second component at ∼1640 cm−1 is discerned. The proper assignment of the latter is not trivial. It probably arises from the overlapping of multiple contributions:67 the random coiled segments of cellular proteins, the intramolecular β-sheet structures (a hypothesis supported also by the presence of the component at 1682 cm−1) and the bending of water, both free (extracellular water) and bound to proteins.68 F-U937 has a similar profile of the average second derivative while it exhibits a much lower standard deviation with respect to LU937, as a consequence of the uncertainty on water subtraction that mainly affects the bandwidth in this spectral region. Both

AD-U937 and E-U937 centroids clearly show two extra components, centered at 1628 and 1695 cm−1, that can be assigned to extended intermolecular β-structures and proteins aggregates.69 The dependence of the shape of the amide II band on protein structure is more complex than amide I even if equally sensitive.70 L-U937 centroid shows two major components in the 1582−1480 cm−1 range, centered at 1550 and 1515 cm−1, accounting for the α-helix contribution70 and tyrosine amino acid,71 respectively. A broad band centered at 1545 cm−1 characterizes F-, AD-, and E-U937 as well, due to both α-helix and β-sheet protein segments. Any attempt to draw out quantitative information on protein conformation content is inappropriate, considered the huge complexity of the system under investigation; however, from a qualitative point of view, it is possible to conclude that formalin fixation preserves the conformation of cellular proteins in a state closer to the one in physiological environment while both air-drying and ethanol induce protein misfolding and consequent protein aggregation and precipitation. Formaldehyde is a coagulant fixative that acts by forming methylene bridges between reactive groups, such as amino, amido, guanidine, thiol, phenolic, imidazolyl, and indolyl, but the most frequent type of cross-linking is the one between the amino terminal groups of lysine and the amido groups of the peptide backbone. This mechanism of action results in the loss of integrity of the quaternary and tertiary structure of proteins and may possibly alter also secondary protein conformations. However, the chosen fixation conditions prevent the protein denaturation, as revealed by cluster analysis and previously highlighted by Mason et al.49 4772

dx.doi.org/10.1021/ac300313x | Anal. Chem. 2012, 84, 4768−4775

Analytical Chemistry

Article

The position of PO2−sym stretching mode is almost insensitive to nucleic acid conformation,87 and it falls at 1086 cm−1 for all the samples investigated. However, such band is more intense and sharper for live cells, as already highlighted by Milijkovic et al.,88 probably reflecting the phosphate backbone fragmentation induced by denaturing agents such as formalin and ethanol or by autolysis for unfixed cells. Interestingly, nonliving U937 cells exhibit an intense contribution in the 1060−1057 cm−1 region, shifted toward 1052 cm−1 for L-U937, that could be assigned to DNA/RNA backbone vibrations with a strong C−O stretching contribution, whose intensity is increased in the case of Z-form helices.89 The biological role of Z-DNA, if it exists in-vivo and how its formation is regulated are questions still debated and is beyond the aim of this paper to dissert on this topic.

Furthermore, formalin induces a decrement in amide I integral intensity, whose averaged normalized value is 0.634 ± 0.061 a.u. versus 0.896 ± 0.195 a.u. for L-U937. This aspect has been already highlighted using Raman microspectroscopy,49,72 and it was attributed to the mechanism of action of formalin that produces a loss of secondary amide groups through the formation of methylene bridges. Indeed, it could account also for a higher order of cellular proteins in hydrated environment thanks to the stabilizing effects on aminoacid lateral side chains induced by bound water, in agreement with Ishida and Griffiths.73 AD-U937 exhibits protein content similar to FU937 (0.681 ± 0.085 a.u.) but differences in protein structure, due to protein denaturation promoted by cell autolysis and consequent aggregation. The phenomenon of protein precipitation is even more pronounced for E-U937, due to the disruption of the hydrogen bonding network induced by dehydration. Moreover, a leak of soluble cytoplasmic proteins during ethanol fixation, as a consequence of membrane permeabilization, can be deduced by a more pronounced lowering of the E-U937 normalized protein content (0.499 ± 0.063 a.u.) in comparison with the other samples considered. It is in the nucleic acid−carbohydrates low frequency region, below 1300 cm−1, where living cells show the most prominent differences with respect to fixed and unfixed air-dried ones (see Figure 3c). The most intense bands characteristic of animal cells in such region are the asymmetric and symmetric stretching of phosphate moieties, PO2−asym and PO2−sym respectively. PO2−asym is sensitive to nucleic acid conformation and clearly exhibits distinctive characteristics for L-U937. LU937 PO2−asym band originates from the overlapping of two major components, centered at 1240 and 1220 cm −1 respectively. The 1220 cm−1 component arises from the Bhelical form of the both DNA and RNA, overlapped with a RNA ribose vibration.74,75 The 1240 cm−1 band is diagnostic for the A-helical form of nucleic acids. However, the A-form of DNA has been detected only in dehydrated DNA samples or for high ionic strength solutions,76 while RNA A-form is stable also under physiological conditions. Therefore, we assigned this band univocally to RNA A-form of L-U937. Asymmetric stretching phosphate bands of F- and E-U937, centered at 1242 and 1237 cm−1 respectively, are much less resolved. Ethanol is a solvent commonly employed for the purification of nucleic acid through precipitation, exploiting the lower solubility of both DNA and RNA in alcohol with respect to more polar water.77 Since dehydration is favoring the transition from B- to A-helical form of DNA76,78 as well as, in extreme conditions like solvent evaporation, the denaturation of nucleic acids,79−83 it is possible to postulate that the broad PO2−asym band of E-U937 is generated by the overlapping of both A and B nucleic acids conformers and unpaired/uncoiled structures. As a matter of fact, the displacement of intracellular water by ethanol negatively impacts the balance of cell osmotic environment and, consequently, the cellular morphology, as evidenced from the disappearance of the nuclear structure (See SEM micrograph in Figure 2c). Formaldehyde is a well-known denaturizing agent of DNA, even if, at the fixation conditions, it is believed to preserve its structure.83−85 However, our studies show that a partial denaturation is still occurring. This hypothesis is supported also by the suppression of the band centered at 1717 cm−1 upon drying or fixation, which reflects the base paring of nucleic acids in the B-form,86 particularly pronounced in L-U937 and almost absent in F-, E-, and ADU937.



CONCLUSIONS From the presented results, it is possible to conclude that the IR spectra of both formalin fixed and unfixed air-dried U937 monocytes are good snapshots of the cellular status at the time of fixation, as deduced by comparison with the IR spectra of live U937 cells acquired in MD. Both lipid order and composition as well as protein conformation are quite well preserved for both protocols, while alcohol dehydration is dramatically affecting both cellular macromolecular content and architecture. However, the nucleic acid structure is much less defined and for sure affected by denaturation when any fixation protocol or simple air-drying without fixation are employed. A clear and valuable distinction between the different DNA and RNA conformers is possible only operating under physiological conditions. This advantage, coupled with the possibility offered by MD-IRMS to monitor in real time the biochemical modifications undergone by live cells, opens new and exciting opportunities in the field of IRMS for life sciences and promote its upgrade to a powerful label-free cell-based assays.



ASSOCIATED CONTENT

S Supporting Information *

Microfluidic device fabrication and connection; cell viability analysis; and water subtraction strategies. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Mailing address: Sincrotrone Trieste S.CpA di interesse nazionale, Strada Statale 14 - km 163,5 in AREA Science Park, 34149 Basovizza, Trieste, Italy. Phone. +39 040 3758465. Fax: +39 040 9380902. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS The authors want to thank Dr. Diana Bedolla for the proofreading. REFERENCES

(1) Dixit, R.; Cyr, R. Plant J. 2003, 36 (2), 280−90. (2) Cunningham, B. T.; Karlsson, R.; Rich, R. L.; Myszka, D. G.; Cooper, M. A.; Markey, F.; Huber, W.; Fang, Y.; Fang, J.; Tran,E.; Xie, X.; Hallstrom, M.; Frutos, A. G.; Brown,R. K.; Brandts, M.; O’Brien, R.; Peters, W. B.; McGuinness, R. ; Verdonk, E. Label-free biosensors: techniques and applications, 1st ed.; Cambridge University Press: Cambridge, New York, 2009; p 300. 4773

dx.doi.org/10.1021/ac300313x | Anal. Chem. 2012, 84, 4768−4775

Analytical Chemistry

Article

(3) Hayat, M. A. Principles and Techniques of Electron Microscopy: Biological Applications, 4th ed.; Cambridge University Press: Cambridge, New York, 2000. (4) Matthew, A, C. Drug Discovery Today 2006, 11 (23−24), 1061− 1074. (5) Barth, A.; Haris, P. I. Biological and Biomedical Infrared Spectroscopy; IOS Press: Amsterdam, 2009. (6) Stuart, B. H. Biological Applications of Infrared Spectroscopy, 2nd ed.; John Wilwy & Sons: New York, 2004. (7) Mantsch, H. H.; Chapman, D. Infrared Spectroscopy of Biomolecules, 3rd ed.; Wiley-Liss, Inc.: 1996. (8) Holman, H.-Y. N.; Martin, M. C.; McKinney, W. R. J. Biol. Phys. 2003, 29 (2−3), 275−286. (9) Sahu, R. K.; Mordechai, S. Future Oncol. 2005, 1 (5), 635−47. (10) Yano, K.; Ohoshima, S.; Gotou, Y.; Kumaido, K.; Moriguchi, T.; Katayama, H. Anal. Biochem. 2000, 287 (2), 218−225. (11) Holman, H.-Y. N.; Martin, M. C.; Blakely, E. A.; Bjornstad, K.; Mckinney, W. R. Biopolymers 2000, 57, 329−335. (12) Boydston-White, S.; Gopen, T.; Houser, S.; Bargonetti, J.; Diem, M. Biospectroscopy 1999, 5 (4), 219−27. (13) Didonna, A.; Vaccari, L.; Bek, A.; Legname, G. ACS Chem. Neurosci. 2011, 2 (3), 160−174. (14) Kneipp, J. Biochim. Biophys. Acta 2003, 1639 (3), 152−158. (15) Kneipp, J.; Miller, L. M.; Spassov, S.; Sokolowski, F.; Lasch, P.; Beekes, M.; Naumann, D. Biopolymers 2004, 74 (1−2), 163−7. (16) Kretlow, A.; Wang, Q.; Kneipp, J.; Lasch, P.; Beekes, M.; Miller, L.; Naumann, D. Biochim. Biophys. Acta 2006, 1758 (7), 948−959. (17) Gasper, R.; Dewelle, J.; Kiss, R.; Mijatovic, T.; Goormaghtigh, E. Biochim. Biophys. Acta 2009, 1788 (6), 1263−70. (18) Draux, F.; Jeannesson, P.; Gobinet, C.; Sule-Suso, J.; Pijanka, J.; Sandt, C.; Dumas, P.; Manfait, M.; Sockalingum, G. D. Anal. Bioanal. Chem. 2009, 395 (7), 2293−301. (19) Flower, K. R.; Khalifa, I.; Bassan, P.; Demoulin, D.; Jackson, E.; Lockyer, N. P.; McGown, A. T.; Miles, P.; Vaccari, L.; Gardner, P. Analyst 2010, 136, 498−507. (20) Bellisola, G.; Della Peruta, M.; Vezzalini, M.; Moratti, E.; Vaccari, L.; Birarda, G.; Piccinini, M.; Cinque, G.; Sorio, C. Analyst 2010, 135 (12), 3077−86. (21) Lasch, P.; Kneipp, J. Biomedical Vibrational Spectroscopy; John Wiley & Sons: New York, 2008. (22) Gierlinger, N.; Goswami, L.; Schmidt, M.; Burgert, I.; Coutand, C.; Rogge, T.; Schwanninger, M. Biomacromolecules 2008, 9 (8), 2194−201. (23) Nasse, M. J.; Ratti, S.; Giordano, M.; Hirschmugl, C. J. Appl. Spectrosc. 2009, 63 (10), 1181−6. (24) Heraud, P.; Wood, B. R.; Tobin, M. J.; Beardall, J.; McNaughton, D. FEMS Microbiol. Lett. 2005, 249 (2), 219−225. (25) Tobin, M. J.; Puskar, L.; Barber, R. L.; Harvey, E. C.; Heraud, P.; Wood, B. R.; Bambery, K. R.; Dillon, C. T.; Munro, K. L. Vibrational Spectrosc. 2010, 53 (1), 34−38. (26) Moss, D.; Keese, M.; Pepperkok, R. Vibrational Spectrosc. 2005, 38 (1−2), 185−191. (27) Quaroni, L.; Zlateva, T.; Normand, E. Anal. Chem. 2011, 83 (19), 7371−7380. (28) Holman, H. Y.; Martin, M. C.; Blakely, E. A.; Bjornstad, K.; McKinney, W. R. Biopolymers 2000, 57 (6), 329−35. (29) Zhao, R.; Quaroni, L.; Casson, A. G. Analyst 2010, 135 (1), 53− 61. (30) Marcsisin, E. J.; Uttero, C. M.; Miljkovic, M.; Diem, M. Analyst 2010, 135 (12), 3227−3232. (31) Mourant, J. R.; Yamada, Y. R.; Carpenter, S.; Dominique, L. R.; Freyer, J. P. Biophys. J. 2003, 85, 1938−1947. (32) Holman, H.-Y. N.; Miles, R.; Hao, Z.; Wozei, E.; Anderson, L. M.; Yang, H. Anal. Chem. 2009, 81 (20), 8564−8570. (33) Holman, H.-Y. N.; Bechtel, H. A.; Hao, Z.; Martin, M. C. Anal. Chem. 2010, 82 (21), 8757−8765. (34) Bassan, P.; Byrne, H. J.; Lee, J.; Bonnier, F.; Clarke, C.; Dumas, P.; Gazi, E.; Brown, M. D.; Clarke, N. W.; Gardner, P. Analyst 2009, 134 (6), 1171−5.

(35) Romeo, M.; Diem, M. Vibrational Spectrosc. 2005, 38 (1−2), 129−132. (36) Filik, J.; Frogley, M. D.; Pijanka, J. K.; Wehbe, K.; Cinque, G. Analyst 2012, 37 (4), 853−861. (37) Birarda, G.; Grenci, G.; Businaro, L.; Marmiroli, B.; Pacor, S.; Piccirilli, F.; Vaccari, L. Vibrational Spectrosc. 2010, 53 (1), 6−11. (38) Birarda, G.; Grenci, G.; Businaro, L.; Marmiroli, B.; Pacor, S.; Vaccari, L. Microelectron. Eng. 2010, 87 (5−8), 806−809. (39) Sundström, C.; Nilsson, K. Int. J. Cancer 1976, 17 (5), 565−577. (40) Lupi, S.; Nucara, A.; Perucchi, A.; Calvani, P.; Ortolani, M.; Quaroni, L.; Kiskinova, M. JOSA B 2007, 24 (4), 959−964. (41) Venyaminov, S.; Prendergast, F. G. Anal. Biochem. 1997, 248 (2), 234−45. (42) Bonifacio, A.; Beleites, C.; Vittur, F.; Marsich, E.; Semeraro, S.; Paoletti, S.; Sergo, V. Analyst 2010, 135 (12), 3193−3204. (43) Beleites, C.; Geige, r. K.; Kirsch, M.; Sobottka, S.; Schackert, G.; Salzer, R. Anal. Bioanal. Chem. 2011, 400 (9), 2801−2816. (44) Powell, J. R.; Wasacz, F. M.; Jakobsen, R. J. Appl. Spectrosc. 1986, 40 (3), 339−344. (45) Wang, S. Y.; Mak, K. L.; Chen, L. Y.; Chou, M. P.; Ho, C. K. Immunology 1992, 77 (2), 298−303. (46) Weyn, B.; Kalle, W.; Kumar-Singh, S.; Van Marck, E.; Tanke, H.; Jacob, W. J. Microsc. 1998, 189 (Pt 2), 172−80. (47) Gazi, E.; Dwyer, J.; Lockyer, N. P.; Miyan, J.; Gardner, P.; Hart, C.; Brown, M.; Clarke, N. W. Biopolymers 2005, 77 (1), 18−30. (48) Ofaolain, E.; Hunter, M.; Byrne, J.; Kelehan, P.; McNamara, M.; Byrne, H.; Lyng, F. Vibrational Spectrosc. 2005, 38 (1−2), 121−127. (49) Mason, J. T.; O’Leary, T. J. J. Histochem. Cytochem. 1991, 39 (2), 225−9. (50) Hastings, G.; Wang, R.; Krug, P.; Katz, D.; Hilliard, J. Biopolymers 2008, 89 (11), 921−930. (51) Pleshko, N.; Boskey, A.; Mendelsohn, R. Calcif. Tissue Int. 1992, 51 (1), 72−77. (52) Gazi, E.; Gardner, P. In Vibrational Spectroscopic Imaging for Biomedical Applications; Srinivasan, G., Ed.; McGraw Hill: New York, 2010; pp 59−94. (53) Mantsch, H. H.; McElhaney, R. N. Chem. Phys. Lipids 1991, 57 (2−3), 213−26. (54) Casal, H. L.; Mantsch, H. H. Biochim. Biophys. Acta 1984, 779 (4), 381−401. (55) Susi, H.; Byler, D. M. Methods Enzymol. 1986, 130, 290−311. (56) Tsvetkova, N. M.; Horvath, I.; Torok, Z.; Wolkers, W. F.; Balogi, Z.; Shigapova, N.; Crowe, L. M.; Tablin, F.; Vierling, E.; Crowe, J. H.; Vigh, L. Proc. Natl. Acad. Sci. U.S.A. 2002, 99 (21), 13504−9. (57) Yuk, H.-G.; Marshall, D. L. Appl. Environ. Microbiol. 2003, 69 (9), 5115−5119. (58) Patra, M.; Salonen, E.; Terama, E.; Vattulainen, I.; Faller, R.; Lee, B. W.; Holopainen, J.; Karttunen, M. Biophys. J. 2006, 90 (4), 1121−1135. (59) Wolman, M.; Greco, J. Biotech. Histochem. 1952, 27 (6), 317− 324. (60) JW, C.; DS, T.; Thompson, D. L. Biopolymers 2009, 91 (2), 132−139. (61) Gazi, E.; Gardner, P.; Lockyer, N. P.; Hart, C. A.; Brown, M. D.; Clarke, N. W. J. Lipid Res. 2007, 48 (8), 1846−1856. (62) Mourant, J. R.; Gibson, R. R.; Johnson, T. M.; Carpenter, S.; Short, K. W.; Yamada, Y. R.; Freyer, J. P. Phys. Med. Biol. 2003, 48 (2), 243−57. (63) Lasch, P.; Boese, M.; Pacifico, A.; Diem, M. Vibrational Spectrosc. 2002, 28, 147−157. (64) Lasch, P.; Pacifico, A.; Diem, M. Biopolymers 2002, 67 (4−5), 335−338. (65) H.Y., H.; Martin, M. C.; Blakely, E. A.; Bjornstad, K.; McKinney, W. R. Biopolymers 2000, 57 (6), 329−335. (66) Salman, A.; Ramesh, J.; Erukhimovitch, V.; Talyshinsky, M.; Mordechai, S.; Huleihel, M. J. Biochem. Biophys. Methods 2003, 55 (2), 141−153. 4774

dx.doi.org/10.1021/ac300313x | Anal. Chem. 2012, 84, 4768−4775

Analytical Chemistry

Article

(67) Dong, A.; Huang, P.; Caughey, W. S. Biochemistry 1990, 29 (13), 3303−8. (68) Abbott, T. P.; Nabetani, H.; Sessa, D. J.; Wolf, W. J.; Liebman, M. N.; Dukor, R. K. J. Agric. Food Chem. 1996, 44 (8), 2220−2224. (69) Zandomeneghi, G.; Krebs, M. R. H.; McCammon, M. G.; Fändrich, M. Protein Sci. 2004, 13 (12), 3314−3321. (70) Goormaghtigh, E.; Ruysschaert, J.; Raussens, V. Biophys. J. 2006, 90 (8), 2946−2957. (71) Barth, A. Prog. Biophys. Mol. Biol. 2000, 74 (3−5), 141−73. (72) Meade, A.; Clarke, C.; Draux, F.; Sockalingum, G.; Manfait, M.; Lyng, F.; Byrne, H. Anal. Bioanal. Chem. 2010, 396 (5), 1781−1791. (73) Ishida, K. P.; Griffiths, P. R. Appl. Spectrosc. 1993, 47 (5), 584− 589. (74) Banyay, M.; Sarkar, M.; Graslund, A. Biophys. Chem. 2003, 104 (2), 477−88. (75) Banyay, M.; Sandbrink, J.; Strömberg, R.; Gräslund, A. Biochem. Biophys. Res. Commun. 2004, 324 (2), 634−639. (76) Pastor, N. Biophys. J. 2005, 88 (5), 3262−3275. (77) Shie, M.; Kharitonenkov, I. G.; Tikhonenko, T. I.; Chirgadze, Y. N. Nature 1972, 235 (5338), 386−388. (78) Lee, C. H.; Mizusawa, H.; Kakefuda, T. Proc. Natl. Acad. Sci. U.S.A. 1981, 78 (5), 2838−2842. (79) Svaren, J.; Inagami, S.; Lovegren, E.; Chalkley, R. Nucleic Acids Res. 1987, 15 (21), 8739−8754. (80) Lee, S. L.; Debenedetti, P. G.; Errington, J. R.; Pethica, B. A.; Moore, D. J. J. Phys. Chem. B 2004, 108 (9), 3098−3106. (81) Herskovits, T. T.; Harrington, J. P. Biochemistry 1972, 11 (25), 4800−4811. (82) Herskovits, T. T.; Singer, S. J.; Geiduschek, E. P. Arch. Biochem. Biophys. 1961, 94 (1), 99−114. (83) Theodore, T, H. Arch. Biochem. Biophys. 1962, 97 (3), 474−484. (84) Freifelder, D.; Davison, P. F. Biophys. J. 1963, 3, 49−63. (85) Blake, R.; Delcourt, S. G. Nucleic Acids Res. 1996, 24 (11). (86) Taillandier, E.; Liquier, J. Methods Enzymol. 1992, 211, 1−619. (87) Liquier, J.; Akhebat, A.; Taillandier, E.; Ceolin, F.; Dinh, T. H.; Igolen, J. Spectrochim. Acta A Mol. Biomol. Spectrosc. 1991, 47 (2), 177−186. (88) Miljkovic, M.; Romeo, M.; Matthus, C.; Diem, M. Biopolymers 2004, 74 (1−2), 172−175. (89) Taillandier, E.; Peticolas, W. L.; Adam, S.; Huynh-Dinh, T.; Igolen, J. Spectrochim. Acta A Mol. Biomol. Spectrosc. 1990, 46 (1), 107−112.

4775

dx.doi.org/10.1021/ac300313x | Anal. Chem. 2012, 84, 4768−4775