Research article Received: 3 September 2013,

Revised: 3 November 2013,

Accepted: 20 November 2013,

Published online in Wiley Online Library: 8 January 2014

(wileyonlinelibrary.com) DOI: 10.1002/nbm.3061

Effects of hypotonic stress and ouabain on the apparent diffusion coefficient of water at cellular and tissue levels in Aplysia Ileana Ozana Jelescua, Luisa Ciobanua, Françoise Geffroya, Pierre Marquetb and Denis Le Bihana* There is evidence that physiological or pathological cell swelling is associated with a decrease of the apparent diffusion coefficient (ADC) of water in tissues, as measured with MRI. However the mechanism remains unclear. Magnetic resonance microscopy, performed on small tissue samples, has the potential to distinguish effects occurring at cellular and tissue levels. A three-dimensional diffusion prepared fast imaging with steady-state free precession sequence for MR microscopy was implemented on a 17.2 T imaging system and used to investigate the effect of two biological challenges known to cause cell swelling, exposure to a hypotonic solution or to ouabain, on Aplysia nervous tissue. The ADC was measured inside isolated neuronal soma and in the region of cell bodies of the buccal ganglia. Both challenges resulted in an ADC increase inside isolated neuronal soma (+31 ± 24% and +30 ± 11%, respectively) and an ADC decrease at tissue level in the buccal ganglia (12 ± 5% and 18 ± 8%, respectively). A scenario involving a layer of water molecules bound to the inflating cell membrane surface is proposed to reconcile this apparent discrepancy. Copyright © 2014 John Wiley & Sons, Ltd. Additional supporting information may be found in the online version of this article at the publisher’s web site. Keywords: biophysical mechanisms of MR diffusion; microscopy; cellular and molecular CNS imaging; Aplysia; cell swelling; digital holographic microscopy

INTRODUCTION

280

Diffusion MRI is exquisitely sensitive to tissue structure, such as cell size or anisotropy (1). In particular, events leading to cell swelling have been shown to result in a decrease of the water apparent diffusion coefficient (ADC) in tissues, as during acute cerebral ischemia (2), hypotonic challenges (3) or cortical excitation by potassium ions or bicuculline (4,5). Yet, the exact mechanism of this ADC decrease has remained elusive. A popular view is that cell swelling causes the extracellular fraction to shrink in favor of intracellular space, where diffusion is assumed to be slow (6,7). Other explanations invoke the arrest of a cytoplasmic streaming or an increase in cytoplasmic viscosity (8,9), as well as an increase in extracellular tortuosity (10). Neurite beading has also been suggested as a source of ADC decrease in brain tissue following ischemia (11), but the ADC decrease pattern with cell swelling has also been observed in tissues without neurites (12). Magnetic resonance microscopy (MRM) provides access down to micrometer ranges (13,14) and has the potential to shed light on the diffusion changes associated with cell swelling both at cellular and tissue levels. MRM diffusion measurements in single cells have been reported in the past, but with limited spatial resolution (15,16) or long acquisition times (17). Three-dimensional (3D) acquisitions with high isotropic resolution are better suited, as they provide higher signal-to-noise ratios, but scan time should be kept as short as possible in order to insure stable tissue conditions. In this work, we implemented a 3D DP-FISP (diffusion prepared fast imaging with steady-state free precession) sequence (18) on a 17.2 T MRI scanner with suitable timings and resolution for

NMR Biomed. 2014; 27: 280–290

MR microscopy, and validated it on various phantoms against a standard 3D diffusion-weighted spin-echo (DW-SE) sequence. The 3D DP-FISP was then used to investigate water ADC behaviors at two structural levels in an Aplysia californica model: inside large spherical neuronal soma mechanically isolated from the abdominal or pleural ganglia, and in the region of cell bodies of the buccal ganglia (19). This region is essentially constituted of very large neuronal bodies (at least 40% volumetric fraction, as measured in the current study), some glial cells and extracellular space, but is almost devoid of neuronal processes (19). We quantified the change in ADC within soma and ganglion tissue following exposure to two challenges known to cause cell swelling: * Correspondence to: D. Le Bihan, NeuroSpin, Commissariat à l’Energie Atomique et aux Energies Alternatives, Bât 145, CEA-Saclay Center, 91191 Gif-sur-Yvette, France. E-mail: [email protected] a I. O. Jelescu, L. Ciobanu, F. Geffroy, D. Le Bihan NeuroSpin, Commissariat à l’Energie Atomique et aux Energies Alternatives, Gif-sur-Yvette, France b P. Marquet Centre des Neurosciences Psychiatriques Département de Psychiatrie, DP-CHUV, Prilly-Lausanne, Switzerland Abbreviations used: 2D/3D, two/three dimensional; ADC, apparent diffusion coefficient; ASW, artificial sea water; DAPI, 4’,6-diamidino-2-phenylindole; DHM, digital holographic microscopy; DP-FISP, diffusion prepared fast imaging with steady-state free precession; DW-SE, diffusion-weighted spin-echo; HEPES, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; ID, inner diameter; IRI, intracellular refractive index; MRM, magnetic resonance microscopy; PVEs, partial volume effects; RARE, rapid acquisition with refocused echoes; ROI, region of interest.

Copyright © 2014 John Wiley & Sons, Ltd.

EVOLUTION OF CELLULAR AND TISSUE WATER DIFFUSION DURING CELL SWELLING hypotonic stress or ouabain. Hypotonic shock is a simple and robust means of inducing cell swelling through osmotic equilibration, while ouabain is a powerful K+/Na+-ATPase inhibitor, which induces excitotoxicity in addition to swelling (20). Unlike ischemia however, hypotonic media or ouabain do not lead to complete energy depletion. Cell volume changes induced by these challenges were also estimated with optical microscopy and digital holographic microscopy (DHM). The latter provides a quantitative estimate of cytoplasmic dilution through the measurement of the intracellular refractive index (IRI).

ASW and from 65 to 7500 ms for the other two phantoms. The 3D DP-FISP sequence parameters for ADC measurement were Δ = 10 ms; δ = 2.5 ms; b = 10/100/200/300/400/500/600 s/mm2; matrix, 190 × 32 × 24; 25 μm isotropic resolution; flip angle = 20°; TE/TR = 2.6/5.2 ms; TRPE2 = 6 s; NA = 4; TA = 9 min 36 s/b-value. The 3D DW-SE parameters were chosen to match those of the DP-FISP, only differing in the following: matrix, 128 × 32 × 24; TE = 18.45 ms; TR = 3 s; NA = 1; TA = 38 min 24 s/b-value. The ADC was calculated in each experiment by fitting ln S = b ADC + constant. Single cell diffusion MR imaging

EXPERIMENT Tissue preparation Fourteen Aplysia californica (National Resource for Aplysia, Miami, FL, USA) were used in total: ten for the MR diffusion measurements and for the optical and DHM volume measurements, three for the estimation of soma fraction in the buccal ganglion tissue, and one for histological evidence of the composition of the region of cell bodies. Animals were anesthetized by injection of an isotonic magnesium chloride solution (MgCl2, 360 mM; HEPES, 10 mM; pH = 7.5). The buccal, abdominal and left pleural ganglia were resected and placed in artificial sea water (ASW) (NaCl, 450 mM; KCl, 10 mM; MgCl2, 30 mM; MgSO4, 20 mM; CaCl2, 10 mM; HEPES, 10 mM; pH = 7.5; osmolarity = 1090 mOsm/L) – all salts from Sigma-Aldrich (Saint-Quentin Fallavier, France). The abdominal and left pleural ganglia were unsheathed and the larger neuronal bodies, with diameters between 100 and 450 μm, were mechanically isolated. These cells were typically R2 and L2–L11 in the abdominal ganglion, and LPl1 in the left pleural ganglion (21,22). MR microscopy All experiments were performed at 19 °C on a 17.2 T system (Bruker BioSpin, Ettlingen, Germany) equipped with 1 T/m gradients. RF transceivers were home-built microcoils, the design of which has been described elsewhere (23,24). The coil inner diameters (IDs) were 700 μm for the phantom and single cell experiments and 2.4 mm for the buccal ganglia experiments.

Cell integrity was verified under the microscope. Each cell was inserted into a 500 μm ID glass capillary filled with ASW and slid inside the transceiver. The cell underwent an initial (pre-challenge) 3D DP-FISP imaging session. Imaging parameters were identical to those in the phantom study. The cell was then removed from the capillary and placed inside a glass cell culture dish (CELLview, Greiner Bio-One, Courtaboeuf, France) filled with the new medium. This medium was either a 33% hypotonic ASW (obtained by mixing two-thirds ASW and one-third distilled water; osmolarity, 727 mOsm/L), ASW containing 1 mM ouabain octahydrate (Sigma-Aldrich) or regular ASW (control solution). The purpose of the control solution experiments was to ensure that our experimental set-up had no effect on the cell ADC and volume. The cell was allowed to equilibrate for 30 min inside the new medium, a period split into 20 min inside the culture dish and 10 min inside a glass capillary while setting up for the post-challenge imaging session. Cell integrity was again verified under the microscope at times t = 0 and t = 20 min in the culture dish. The second (post-challenge) imaging session was identical to the first one. The total duration of the experiment from initial cell isolation to the end of the post-challenge imaging was approximately 3 h, consistent with expected cell survival time (15). In addition, the contrast between ASW, cytoplasm and nucleus on the diffusion-weighted images allowed us to determine whether the cell appeared stable or whether it had undergone a process of nucleus expulsion. Data on cells with compromised integrity was not retained. Six cells were thus exposed to a hypotonic shock, six others to ouabain and three to the control ASW.

3D DP-FISP implementation and phantom validation

NMR Biomed. 2014; 27: 280–290

Buccal ganglia diffusion MR imaging Ten pairs of buccal ganglia were scanned. The bilateral ganglia were inserted into a 2.0 mm ID glass capillary filled with ASW and then slid inside the transceiver. Each pair of ganglia underwent a protocol similar to that for the single cell. A pre-challenge 3D DPFISP imaging session was performed. The imaging parameters that differed from the single cell experiments were the following: matrix, 128 × 44 × 40; 50 μm isotropic resolution; TE/TR = 1.7/ 3.4 ms; NA = 2; TA = 8 min/b-value. Following this initial imaging session, the ganglia were removed from the capillary and placed inside a Petri dish filled with the new medium. This medium was either hypotonic ASW (33% – 727 mOsm/L – for two ganglia and then 67% – 363 mOsm/L, obtained by mixing one-third ASW with two-thirds distilled water – for two others), ASW with ouabain (1 mM for two ganglia and then 3 mM for two others) or regular ASW (control solution – two ganglia). As for the single cell, the ganglia were allowed 30 min of equilibration inside the new medium. The second imaging session was identical to the first one. The total duration of the experiment from initial ganglion

Copyright © 2014 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/nbm

281

The diffusion preparation consisted in a driven equilibrium module: non-selective (90x–180y–90x) pulses with pulsed diffusion gradients placed in between. The effective diffusion time was tdiff = 9.2 ms, with an associated mean diffusion distance of 10 μm in pure water and approximately 7 μm in tissue, much shorter than both the size of Aplysia neurons and the highest spatial resolution employed (25 μm). The ultra-high field reduces the characteristic T1 bias on the diffusion-prepared magnetization through prolonged T1 relaxation times. In an attempt to further limit the T1 contamination, the FISP acquisition was set up with centric encoding and very short overall timing (18). The ADC measurements using 3D DP-FISP and 3D DW-SE were compared on phantoms. The latter were 500 μm ID glass capillaries (VitroCom, Mountain Lakes, NJ, USA) filled with three different solutions: ASW, a saline solution doped with CuSO4 and dodecane (all chemicals from Sigma-Aldrich). T1 was measured in each of the phantoms with a multi-TR rapid acquisition with refocused echoes (RARE) sequence: TEeff = 22 ms; resolution 60 × 30 × 60 μm3; 10 TR values ranging from 65 to 15 000 ms for

I. O. JELESCU ET AL. extraction to the end of the post-challenge imaging was approximately 3 h. This duration is well within the ganglion expected survival time of 48 h (24). The exposure to a more pronounced insult (67% hypotonic ASW and 3 mM ouabain) of half the ganglion population was motivated by the fact that the response to a 33% hypotonic shock or 1 mM ouabain was less marked than literature results on other animal models using similar insults (isolated turtle cerebellum (3) and rat brain (6)). Since the presence of the protective sheath around the ganglia could cause the effective insult experienced by the tissue to be less severe, the levels of hypotonicity and ouabain concentration were therefore increased to test whether the response would be affected. 33% and 67% hypotonic shocks are anticipated to have a large effect on cell volume, with potential for cell lysis. However, previous published data on Aplysia neurons exposed to a 50% hypotonic shock reported an increase in cell cross-sectional area but no cell lysis (25). Nonetheless, in the current study, the integrity of the cell membrane was assessed using optical microscopy, on isolated cells following a 33% hypotonic shock. For cells inside the imaged ganglia, cell lysis would translate into hyperintense areas on T2w images (24): the segmented regions of interest (ROIs) were monitored for such hyperintensities. Anatomical T2w MR imaging of the ganglia Three pairs of buccal ganglia were used to estimate the soma fraction in the region of cell bodies, based on high resolution T2-weighted images. The buccal ganglia were inserted into a 2.0 mm ID glass capillary filled with ASW and imaged using the 2.4 mm ID microcoil. Imaging consisted of a 3D RARE (TEeff = 18.4 ms; TR = 3 s; 25 μm isotropic; TA = 1 h 20 min) for T2 contrast, which allows identification of cell bodies. Single cell optical microscopy imaging Additional neurons isolated from the abdominal and pleural ganglia, approximately 100 μm in diameter, were used to quantify cell and nucleus swelling with fluorescence and absorption microscopy. Each cell was placed inside a cell culture dish filled with ASW. Three neurons were treated for nucleus staining: 2 μL of Hoechst 33342 (Life Technologies, Saint-Aubin, France) were added to the 2 mL of ASW in the culture dish, which was then incubated for 5 min at room temperature, sheltered from light. Cells were then set up under a Zeiss Axio Observer Z1 microscope (Zeiss, Marly-le-Roi, France) for fluorescence imaging. A snapshot image at t = 0 was taken before the cell medium was changed. The new medium was either 33% hypotonic ASW, ASW + 1 mM ouabain, or regular ASW. The acquisition was started immediately, with snapshots acquired every 30 s for 1 h, using Zeiss Axio Vision software. The fluorescence images allowed estimation of cellular and nuclear diameters. In the case of the neuron exposed to ouabain, the staining enabled reliable segmentation of the nuclear but not cellular membrane. An additional neuron was therefore imaged with standard absorption microscopy to determine the cell diameter change with ouabain.

Δφ ¼ ð2π=λÞðnc  nm Þh

[1]

where λ is the wavelength of the lightsource, h the cell thickness, nc the mean IRI (along the path length corresponding to the thickness h) and nm the refractive index of the perfusion solution. To separately obtain both nc and h from the phase shift, a method was developed based on the recording of two holograms and the reconstruction of two quantitative phase images corresponding to two different values of nm. Practically, the cell refractive index nc was determined by sequentially measuring cell phase mapping in two perfusion solutions of different refractive indices but identical osmolarity (26). Specifically, such perfusion solutions have been obtained by adding either 30 mM mannitol (Sigma-Aldrich) or 30 mM Nycodenz (Progen Biotechnik, Heidelberg, Germany), a hydrophilic molecule, to the regular ASW, resulting in two 1120 mM solutions (checked with a Fiske 210 Micro-osmometer) with refractive indices of (1.3372 ± 0.0002) and (1.3403 ± 0.0002) (measured with an Abbe-2WAJ refractometer (GENEQ, Montreal, Canada) at the wavelength of the laser source (680 nm)), respectively. When 33% hypotonic shocks were considered, two 760 mM solutions with refractive indices of (1.3354 ± 0.0002) and (1.3385 ± 0.0002) were used. Buccal ganglion fluorescence microscopy imaging For histological evidence of the composition of the region of cell bodies, one pair of buccal ganglia was fixed using a standard fixation protocol: immersion for 10 min in 4% paraformaldehyde, then in 15% and finally in 30% sucrose. The ganglia, frozen initially at –30 °C and then at 80 °C, were cut into 20 μm thick slices (HM60 Microm Microtech, Francheville, France) which were then placed on glass slips. Immunostaining was performed at room temperature with Neurofilament High (ab4680 primary antibody, revealed with ab96949 secondary antibody coupled to Dylight 594, Abcam, Cambridge, UK) for the cytoskeleton. Slices were mounted with Prolong® (Life Technology, SaintAubin, France) with DAPI (4’,6-diamidino-2-phenylindole, nuclear staining) and then imaged with an Observer Z1 (Zeiss) microscope using Axio Vision software and two successive filters (blue for DAPI and orange for Neurofilament High) at 0.31 × 0.31 μm2 in-plane and 1 μm through-plane resolutions. Data analysis

Digital holographic microscopy

282

As a complement to the fluorescence imaging, DHM was used to monitor the effect of hypotonic shocks at a single cell level. Indeed, DHM gives access to a quantitative phase signal, which allows the measurement of absolute cell volume and IRI: cell swelling and cytoplasmic dilution in response to hypotonic

wileyonlinelibrary.com/journal/nbm

shocks can thus be monitored. The methods for transmission DHM and quantitative phase image reconstruction, used in the present study to estimate the swelling factor induced by a 33% hypotonic shock, have been described previously (26–29). Briefly, mechanically isolated neurons were plated on poly-lysine coated coverslips, subsequently mounted on a custom-made flow chamber and imaged with a DHMT 1000 (Lyncée Tec, PSEEPFL, Lausanne, Switzerland). From a single recorded hologram, quantitative phase images of living cells can be reconstructed by a numerical reconstruction process (28,29). Practically, from the DHM quantitative phase images, the phase shift induced by the observed cell is given by the following equation (29):

ADC change All processing was performed in MATLAB (MathWorks, MA, USA). While non-Gaussian diffusion is manifest in nervous tissue for b-values above 1000 s/mm2 (7), the current study was performed within the limit where the Gaussian phase approximation is still

Copyright © 2014 John Wiley & Sons, Ltd.

NMR Biomed. 2014; 27: 280–290

EVOLUTION OF CELLULAR AND TISSUE WATER DIFFUSION DURING CELL SWELLING valid (bmax = 600 s/mm2). The 3D MR data were displayed as successive slices for segmentation purposes in MATLAB. For the isolated soma, the cell was manually segmented on each slice on which it was present, based on signal contrast with ASW in the b = 10 s/mm2 image (Fig. 1(a)). The voxel count constituted a measurement of cell volume. Signal was averaged over the entire cell 3D ROI for ADC estimation. Whenever the cell was large enough to enable a reliable segmentation of the nucleus, additional analyses of the volumes and ADCs of the nucleus and the cytoplasm were performed separately. For the buccal ganglia, a monolateral ROI encompassing cell bodies that send axonal projections into nerve 2 and nerve 3 was manually segmented on each slice on which this region was present (24,30) (Fig. 1(b), (c)). The size of the resulting 3D ROI was typically around 300 voxels, corresponding to a volume of 38 μL. The signal was averaged over the entire ROI for ADC estimation. The change in ADC induced by the challenge was estimated individually for each cell and ganglion. The significance of the changes was quantified with bilateral paired Student t-tests, performed on the series of ADC values pre- and post-challenge. Because the ganglion response did not scale with levels of hypotonicity or with ouabain concentration, the previous statistical analysis made use of all four ganglia in each insult group, without discriminating between 33% and 67% hypotonic insults, or 1 and 3 mM ouabain concentrations. Soma fraction in the cell body region Each side of the buccal ganglia was treated separately, thus producing six independent estimates of soma volume fraction. On each side, a 3D ROI was outlined corresponding to the region drawn for diffusion measurements (Fig. 1(d)). Histograms of the T2-weighted signal values inside the ROIs displayed two peaks, which matched the signal levels inside and outside cell bodies, respectively. Based on this separation, masks for cell bodies within the ROI were generated automatically (Fig. 1(e)). The soma volume fraction was obtained by comparing the voxel counts of cell body to entire ROI. It should be noted that the obtained value corresponds to a lower limit of soma volume fraction, since cell bodies smaller that the voxel size were likely not accounted for due to partial volume effects (PVEs).

relative change for each challenge group was produced. Bilateral paired t-tests were also performed on the series of pre/post cell volumes for each challenge. In the optical microscopy image series, the cell and nucleus diameters were compared at t = 0 and t = 30 min pre- and postchallenge using Axio Vision software. The changes in diameter were then extrapolated to a change in volume. This constitutes an approximation as the cells and nuclei are not perfectly spherical, nor has their volumetric change been shown to be isotropic. DHM data processing was performed in MATLAB. Barer (31), through his seminal work within the framework of interference microcopy applied to cell imaging, clearly established that phase shift can be related to cell dry mass. In other words, taking into account Equation [1], this means that the IRI can be linearly related to the concentration of the different cell components: nc = n0 + αC, where n0 denotes the solvent refractive index, C the cell dry mass concentration and α the specific refractive increment (31,32). On the other hand, knowing that an osmotic cell swelling is mainly due to an influx of water (dry mass remains constant) inducing a dilution of the intracellular component concentration, the IRI equation can be expressed as a function of the swelling factor β – defined as the ratio of the final-to-initial soma volume – by the following expression: nc(β) = n0 + αC/β. In our case, n0 corresponds to the water refractive index and the value of αC was determined before hypotonic challenge (β = 1) from the measurement of nc . Finally, the swelling factor corresponding to the dilution of the cytoplasm was calculated from nc measured after osmotic shocks.

RESULTS Phantom validation of the 3D DP-FISP sequence The diffusion coefficient measurements obtained with the 3D DP-FISP sequence were within 1% of the values obtained with a standard 3D DW-SE sequence in all three phantoms (Table 1). The DP-FISP sequence implemented thus displayed excellent performance when used in the low b-value range (b ≤ 600 s/mm2) for ADC extraction.

ADC measurements with MRM Cell volume change The cell volume was first assessed with MRI from voxel count V within the cell. The percent relative change in volume of each cell was calculated as (Vpost  Vpre)/Vpre × 100, and an average

The natural logarithm of the diffusion attenuation signal appeared as a linear function of b-value in all cells and ganglia, validating the Gaussian phase approximation approach. The ADC corresponded to the slope of the curve.

NMR Biomed. 2014; 27: 280–290

Copyright © 2014 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/nbm

283

Figure 1. Examples of ROIs for MR data analysis. Shown are representative slices of 3D MR data and corresponding segmentation; cell and ganglion ROIs 2 were segmented on multiple adjacent slices to produce 3D ROIs. (a)–(c) ADC estimation. (a) Single cell image (DP-FISP; b = 10 s/mm ; 25 μm isotropic resolution). The red contour represents the cell ROI on this slice, and the orange contour the nucleus ROI. The cytoplasm is obtained from the difference 2 of the two. The segmentation of the nucleus was not possible on all cells, only on the larger ones. (b) Buccal ganglion image (DP-FISP; b = 10 s/mm ; 50 μm isotropic). The region of cell bodies chosen as the ROI is outlined in red. (c) Schematic diagram of the buccal ganglion, showing the approximate region chosen for ADC estimation. (d), (e) Estimation of soma volumetric fraction. (d) Buccal ganglion T2-weighted image (RARE; 25 μm isotropic); superimposed in light gray is the manually drawn ROI, similar to the ROIs drawn for diffusion measurements. (e) Automatic segmentation of cell bodies (white) within the manually drawn ROI. The RARE image confirms that the automatic segmentation selects cell bodies reliably.

I. O. JELESCU ET AL. Table 1. T1 and ADC estimates (from DW-SE and DP-FISP) in phantoms. The values stem from fits on the average signal in an ROI. The T1 and ADC values of the phantoms are representative of cell survival medium (ASW) and biological samples at 17.2 T Phantom ASW NaCl, CuSO4 Dodecane

T1 (ms)

DW-SE ADC (103 mm2/s)

DP-FISP ADC (103 mm2/s)

Errora (%)

2960 (21)b 1315 (10) 1455 (15)

1.92 (0.01) 2.01 (0.01) 0.76 (0.01)

1.91 (0.01) 1.99 (0.03) 0.76 (0.01)

0.5 1.0

Effects of hypotonic stress and ouabain on the apparent diffusion coefficient of water at cellular and tissue levels in Aplysia.

There is evidence that physiological or pathological cell swelling is associated with a decrease of the apparent diffusion coefficient (ADC) of water ...
2MB Sizes 0 Downloads 0 Views