Mass transport between blood and the arterial wall, and within the wall itself, plays important roles in the maintenance of the wall tissue, which can be up to thick but is largely avascular. For example, smooth muscle cells within the wall require delivery of small solutes such as oxygen, ions, glucose, and amino acids, and macromolecules such as growth factors and lipoproteins. Imbalances in the transport of such molecules, and particularly in the transport of lipoproteins, have been implicated in the development of atherosclerosis, the underlying cause of most heart attacks and strokes, which is characterized in its early stages by the deposition of lipids within the inner layers of the wall. Supporting this view, areas of the wall prone to atherosclerosis are characterized by an enhanced uptake of macromolecules from plasma.1 Mass transport is also key to the delivery of drugs to the arterial media, including the antiproliferative drugs released by drug-eluting stents and drugs to control vascular diameter and stiffness through an influence on smooth muscle tone. Consequently, the transport of plasma-borne molecules into and within the arterial wall has been intensively studied, both experimentally and theoretically, over the past half-century.2 Studying the diffusion and distribution of macromolecules has generally involved the use of radioactive or fluorescent labels. Much attention has focused on the transport of low density lipoproteins (LDL),3, 4, 5, 6, 7 serum albumin,8, 9 and horseradish peroxidase.10, 11 Similar methods have been used to study the transport and distribution of drugs in the wall.12, 13, 14, 15, 16, 17
In the present study, we show that arterial wall mass transport can be studied ex vivo by Fourier transform infrared (FTIR) spectroscopic imaging using an attenuated total reflection (ATR) approach. No dyes or labels are required since the FTIR spectrum contains the signatures of vibrational modes involving specific functionalities of a molecule. A focal plane array (FPA) detector is combined with an FTIR spectrometer to obtain FTIR spectroscopic images using different ATR accessories.18, 19, 20, 21 Thousands of infrared absorption spectra are acquired simultaneously from different locations along a specimen surface; the sample depth probed by ATR–FTIR imaging is restricted to a few micrometers from the surface of the specimen in contact with the ATR crystal. The distribution of different constituents within a sample can be obtained by analyzing the acquired spectral dataset for different bands in the mid-IR region: qualitative analysis of the FTIR spectrum gives information about the state and environment of the molecule, whereas quantitative analysis through the Beer-Lambert law can be performed by assessing the integrated absorbance of the bands. Studying samples containing water or in contact with aqueous solutions can be challenging because of the absorption of infrared radiation by water. However, we have recently demonstrated the applicability of ATR–FTIR imaging to the study of pharmaceutical and biomedical samples in contact with water and imaged the distribution of water within atherosclerotic rabbit abdominal aorta.20
Chemical specificity is the major advantage of this analytical tool, allowing a broad range of applications in different fields.22, 23, 24, 25, 26, 27, 28, 29, 30, 31 The high speed of acquisition offered by the FPA detector is also valuable, enabling kinetic (e.g., pharmacokinetic) properties of different materials to be determined and allowing the study of permeation, diffusion, and dissolution.32, 33, 34, 35, 36 The use of ATR–FTIR imaging to study dynamic properties of biomedical systems (e.g., drug release and sorption) is a rapidly developing area with applications ranging from organs, namely skin,24 hair,37 to tablet dissolution.32, 35, 38
In our previous work, micro ATR–FTIR spectroscopic imaging was applied to histological sections of the thoracic aorta of cholesterol-fed rabbit in order to map the chemical composition and location in the wall of experimental atherosclerosis and, in particular, to examine the effects of age and dietary L-arginine on lesions around branch ostia.39, 40 Here, we demonstrate that imaging such sections while applying model drugs or albumin to their endothelial surface allows a temporal series of concentration profiles across the wall to be obtained, without the need for radioactive or fluorescent labels. Moreover, spectroscopic imaging in conjunction with multivariate analysis allows simultaneous observation of the distribution, and hence transport, of at least two biomolecules simultaneously.
Materials and Methods
Benzyl nicotinate ( pure) was obtained from Sigma-Aldrich, ibuprofen from Whitehall International, and bovine serum albumin (fatty acid free) from First Link.
Benzyl nicotinate, poorly soluble in aqueous media, was dissolved in Ringer’s solution (composition in g/L: NaCl, 9.0; KCl, 0.2; , 0.2; , 0.1) at the saturation limit, and the resulting solution was used. Albumin 6% (w/w) was dissolved in phosphate buffered saline (PBS; 0.15M, pH 7.4). Ibuprofen, also sparingly soluble in water, was added to the albumin solution; the supernatant of the drug-saturated solution was used.
Animal procedures complied with the Animals (Scientific Procedures) Act 1986. Immature ( old) male New Zealand white rabbits (HSDIF strain, Harlan) were used in this study. They were housed at under a light cycle and fed per day of normal chow (9603 TRB Rabbit; Harlan Teklad, Bicester, UK) supplemented for with 1% (w/w) cholesterol (Sigma). The cholesterol was added in ether (inhibitor-free spectrophotometer grade, Sigma), which was then evaporated.
Each rabbit was sacrificed with an overdose of pentobarbital (Euthatal, Rhône Mérieux, iv), and the descending aorta was excised after fixation in situ at physiological pressure for with 10% neutral buffered formalin. A segment of thoracic aorta containing the third pair of intercostal branches was post-fixed for at least in further formalin solution and then opened along the ventral wall. The segment was immersed in a sucrose solution overnight to inhibit the subsequent formation of ice crystals, before being embedded in Bright cryo-m-bed (Bright Instruments, Huntingdon, UK) and frozen. Longitudinal sections thick were cut in a cryostat and mounted on standard microscope slides.
ATR–FTIR images were acquired using a rapid scan Varian system comprising an FTIR spectrometer (Bio-Rad FTS-60A) coupled to a large sample compartment containing a macro-ATR accessory with diamond crystal (Specac Ltd.) and a liquid nitrogen–cooled FPA detector. Each image corresponded to a area of the specimen. Spatial resolution with this experimental setup is about .41
Sections of rabbit aorta were soaked with Ringer’s solution and gently detached from the microscope slide before being transferred onto the ATR crystal for imaging. A piece of glass slide was placed on top of the section to seal and create a diffusion compartment. Only a low contact pressure was applied, through an anvil, to avoid damaging the tissue; this generally ensured a good contact between specimen and crystal and prevented the leakage of diffusing solution underneath the tissue (see schematic diagram of the experiment in Fig. 1 ). The evanescent wave from the crystal has a depth of penetration of approximately at a wavenumber of .
Before each experiment, the tissue sections were rehydrated by applying Ringer’s solution and allowing them to equilibrate on the ATR crystal. (Generally was sufficient.) Rehydration will have further improved the seal between the section and the glass or crystal. It was monitored by measuring spectroscopic images until no change was observed in the absorbance of water bands between successive snapshots. Solutions of drug or protein were then applied to the endothelial surface of the section, between the glass slide and crystal. Experiments were conducted at least in duplicate.
Varian Resolution Pro 4.0 software was used for acquisition and manipulation of the data. An IR absorption spectrum was obtained for each detector pixel by co-adding 32 interferograms. Spectra were obtained over the range at resolution with a zero filling factor of 2, achieving a collection time of approximately two minutes per image. For all image measurements, a background was acquired in the absence of a sample.
Each spectroscopic image is a 3-D dataset of absorbance values at various combinations of wavenumber and position in the image plane. The fingerprint region of the mid-infrared spectrum was analyzed for each detector pixel, and chemical images were generated by integration of characteristic bands contained therein. To calculate the absorbance of each band, a linear baseline was drawn through the troughs either side of the peak, and the absorption profile above this line was integrated with respect to wavenumber.
Multivariate analysis was also performed on an FTIR image of ibuprofen diffusion using ISys 4.0 software. The image measured at the beginning of the experiment, before the application of the drug, was subtracted from the image collected at the end of the experiment to remove the spectral contribution due to the tissue. Factor analysis was employed in the fingerprint region of the resulting dataset; a linear baseline was drawn through the extremes of this spectral range to avoid spurious baseline effects. Selection of relevant factors was based on visual inspection of the score images and loading spectra, as previously described.40
Diffusion of Benzyl Nicotinate
Benzyl nicotinate in its liquid state readily diffuses across sections of thoracic aorta from cholesterol-fed rabbits (data not shown); however, this process has no in vivo relevance. To reproduce a more realistic situation, a saturated solution of benzyl nicotinate in Ringer’s solution was applied to the sections. Note that the concentration of benzyl nicotinate at the saturation limit was too low for the drug to be detectable in the FTIR spectrum of the solution.
Figure 2 shows ATR–FTIR images from a representative experiment to assess diffusion across a longitudinal section of rabbit thoracic aorta at a location upstream of the intercostal branch ostium. No atherosclerotic lesion was visible in this region. The images showing the tissue at different times during the experiment are derived from the amide II band , corresponding mainly to N–H bending of the peptide group [Fig. 2a]. The benzyl nicotinate solution was applied to the endothelial edge of the arterial section. The spectroscopic images showing the distribution of the drug at different times during the experiment were obtained using the asymmetric stretching band of the benzyl nicotinate [Fig. 2b]. The presence of benzyl nicotinate in the tissue was detected a few minutes after the start of the experiment. It penetrated the endothelial surface of the tissue in contact with the drug solution and progressively accumulated within the intima, while the interior of the wall (media) was almost spared. The concentration of drug within the inner wall appears higher than that in the solution in contact with the tissue [see Fig. 2b]; hence a partitioning effect of benzyl nicotinate, possibly into lipids that accumulated in the intima during the cholesterol-feeding, seems to have occurred. The increase of band absorbance in the adventitia may reflect seepage of the solution around the section and subsequent entry from the adventitial side. A more likely explanation, however, is that it also results from partitioning, since the adventitia is rich in lipids.
Figure 3a shows the evolution of the FTIR spectrum as a function of the time at a pixel within the intima, indicated by the asterisk in Fig. 2. The spectrum of pure benzyl nicotinate is also presented for comparison. The accumulation of the drug into this area of the tissue is indicated by an increase in the asymmetric stretching band. The graph in Fig. 3b displays the integrated absorbance of this band for this location as a function of time after the start of the experiment; the drug concentration increases rapidly at the beginning of the experiment and then continues to increase at a less rapid rate.
The diffusion of benzyl nicotinate was also studied across specimens of immature rabbit aorta containing an atherosclerotic lesion (at the downstream margin of the branch ostium). The ATR–FTIR images for this experiment are shown in Fig. 4 . Concentration of benzyl nicotinate was observed in the lesion [see Fig. 4c], while no permeation of the medial layers of the aortic wall was detected, supporting the idea of partitioning. Note the different scales required in Fig. 2b and Fig. 4c, indicating that the concentration of drug in the lesion was more pronounced than that in the inner healthy wall at comparable times.
The spectra corresponding to a single-pixel location within the lesion (asterisk in Fig. 4) show an increase of absorbance of the benzyl nicotinate bands with time [Fig. 5a ]. The concentration of benzyl nicotinate, expressed by the integrated absorbance of the band in the range , appeared to increase linearly during the course of the experiment [Fig. 5b]. Comparing the plots in Figs. 5b and 3b, confirms that the diffusion and accumulation of benzyl nicotinate within the lesion is higher than in the normal aortic intima.
Diffusion of Ibuprofen-Saturated Albumin
At saturation, the concentration of ibuprofen in aqueous solution was too low to produce detectable levels of the drug in tissue sections unless albumin was added as a carrier. A solution of ibuprofen and 6% BSA in PBS was used for the representative experiment reported here. As before, the drug solution was applied to the endothelial edge of the aortic tissue. Figure 6 shows spectroscopic images of the diffusion of ibuprofen-saturated albumin across a specimen of rabbit aorta without visible atherosclerotic lesions. The images of the tissue at different times during the experiment were obtained using the amide II band absorbance, in the range [Fig. 6a]. The images showing the distribution of the drug as a function of the time were derived from integration of the ibuprofen band in the range ( stretching (COOH) and bending) [Fig. 6b].
Figure 7 shows the evolution of FTIR spectra with time for two locations in the tissue, one in the intima (indicated by an asterisk in Fig. 6) and the other in the media (circle in Fig. 6). For both sites, the fingerprint profile gradually increases in intensity as the diffusion time increases; in particular, the amide I and II band absorbance increases due to the penetration of albumin within the tissue. Note that the spectral evolution is greater in the case of the intima [Fig. 7a], which is the part of the tissue directly in contact with the solution, than in the medial layer [Fig. 7b]. The plots in Figure 8 show the evolution of the FTIR spectra at the same locations [Fig. 8a: intima; Fig. 8b: media] as a function of time after the ibuprofen–albumin solution was removed from the sample compartment. These results demonstrate the reversibility of the transport process.
Analysis of albumin diffusion
Spectra were extracted from the ATR–FTIR images described earlier along a horizontal line of the detector array corresponding to a line across the aortic wall [see Fig. 9b ]. Integration of the spectral region between 1800 and , which corresponds to the amide I and II bands, was performed according to the method in Fig. 9a. Figure 9c shows the transmural concentration profiles for albumin across the artery wall at different times, obtained in this way. (The corresponding absorbance at the start of the experiment was subtracted so that no contribution arose from the tissue itself.) The transmural concentration plots show that the penetration of albumin occurs first in the inner layers of the tissue in contact with the solution ( thickness). An increase in the albumin content is observed with time, especially within the intima/inner media. Low permeation of the outermost media/adventitia is detected, although an increase in protein levels is observed towards the end of the experiment.
Multivariate image analysis
Factor analysis was applied to the dataset collected at the end of the experiment , after the dataset for the beginning of the experiment had been subtracted. Two factors were selected, based on the spectral features of the loading spectra and on visual inspection of the score images. The first factor yielded the distribution shown in Fig. 10a . The corresponding loading spectrum showed the signature of albumin.42 The image reveals a greater concentration of albumin outside the tissue than inside it, as expected.
The second factor [Fig. 10b] showed a distribution that we attribute to the ibuprofen–albumin complex; the loading spectrum contained signals due to both components [see pure ibuprofen spectrum in Fig. 10c]. The undissolved ibuprofen observed by univariate analysis [FTIR images in Fig. 6b] is here depicted at the right-hand edge of the score image [see Fig. 10b]. The complex is clearly detected within the tissue [see oval in Fig. 10b]; this result could not have been obtained by integration over the original dataset because the absorbance of the ibuprofen bands in the spectra of the tissue was too low.
Figure 10d is the transmural concentration plot of the ibuprofen–albumin factor across a horizontal line of the array [see Fig. 10b]. The graph shows the presence of the complex in approximately half the overall thickness of the wall.
We previously used micro ATR–FTIR imaging to examine the location and chemical composition of atherosclerotic lesions in sections of the aortic wall of rabbits that had been fed a cholesterol-enhanced diet.39, 40 The approach, dubbed “chemical photography,”39, 40 gave good discrimination between the composition of the normal and diseased wall and revealed effects of age and dietary L-arginine supplements on the distribution and extent of atherosclerosis. Here, we have demonstrated using sections of the same tissue that ATR–FTIR imaging can also be used to assess transport of small and large solutes across the aortic wall. We studied albumin as an inert model of the macromolecules that play critical roles in the development of atherosclerosis; its concentration profile across the aortic tissue was followed as a function of time. We also studied two model drugs: benzyl nicotinate (relative molecular mass ) and ibuprofen . Benzyl nicotinate is the benzyl ester form of niacin. It is more lipophilic than the parent compound (which lowers circulating lipid levels, is a potent vasodilator, and is frequently used in clinical practice43, 44) and was employed in the present study because its partitioning into lipid-rich lesions provides a good test of the ability of ATR–FTIR imaging to discriminate between uptake by overtly healthy and diseased tissue. Ibuprofen is a cyclo-oxygenase inhibitor with potent anti-inflammatory effects; recent studies have investigated its influence on experimental atherosclerosis in animals.45, 46 It was used in the present study because the facilitation of its transport into the wall by albumin provides a good test of the ability of multivariate analysis to increase the discriminatory power of ATR–FTIR imaging.
The image area was and the spatial resolution approximately . The sample depth probed is restricted to few micrometers by the depth of penetration of the evanescent wave into the specimen. This compares favorably with the large number of radiotracer-based studies in which solutes have been detected by counting radioactivity in whole sections of the wall, with typical dimensions on the order of . Although higher resolutions can theoretically be obtained with radiotracers by using autoradiography, very few such studies have been published; the low uptake of plasma molecules into the wall means that impracticably high concentrations and long exposures are required. Since both these radiotracer detection methods destroy the sample, they can be used only at one time point, whereas the ATR–FTIR-based method can follow transport continuously.
Methods employing fluorescence microscopy in conjunction with dye-labeled solutes have somewhat greater resolution than we obtained in the present study. Furthermore, it is possible to make frequent measurements on the same sample47 so long as care is taken to limit photobleaching of the dye during the repeated exposures to the fluorescence-exciting illumination. However, the need to bind large dye molecules (typically, ) to the solute is a substantial disadvantage; even proteins may be significantly modified by such conjugation,48 while the size of the label may equal or even exceed that of smaller solutes. ATR–FTIR imaging avoids this problem since its chemical specificity is alone sufficient to discriminate between the solute and tissue sample. An additional advantage arising from this chemical specificity is that mixtures of solutes can be examined and separated. In the present study, for example, we were able to demonstrate that albumin acts as drug carrier49 for ibuprofen, which was otherwise not observed to enter the aortic wall; factor analysis revealed the presence of an ibuprofen–albumin complex within the artery wall that was undetected by univariate analysis.
A disadvantage of the method used for assessing arterial mass transport in the present study is that no pressure gradient is applied across wall, unlike in vivo, where there is a higher pressure inside the vessel lumen than outside. Additionally, the tissue was dead and had been fixed, ruling out the possibility of active or vesicular transport within the endothelial layer. These disadvantages reflect limitations of the specimen handling and mounting procedures, and would be surmountable with suitable apparatus; they do not alter conclusions about the fundamental utility of ATR–FTIR imaging for such studies. Transport of small molecules across the wall in vivo is predominantly diffusive and independent of active or vesicular processes within endothelial cells, but the significance of pressure-driven convection and vesicular transport increases for larger solutes. Nevertheless, diffusional, endothelium-independent properties are of interest even for arterial macromolecular transport (and have been examined with radiotracer techniques9), while for vessels not exposed to high luminal pressures (such as veins) and for nonvascular tissues, they are of paramount importance. The in situ fixation of the tissue at physiological pressure prevented it from collapsing when excised but will have altered the chemistry of wall components. We previously showed40 that it has only minor influences on absorbances in the spectral regions examined here; others have shown an absence of effects of fixation on transport properties such as hydraulic conductance of the endothelium50 and permselectivity of intact vascular beds51 which may explain why our results are consistent with earlier work employing dead but unfixed tissue.
The fact that ATR–FTIR imaging allows repetitive measurement of the same sample is important because the evolution of concentration profiles across the wall with time provides additional information about transport processes. After short times, when concentrations of solute in the wall are low, uptake depends primarily on the resistance to entry. After several hours, however, an equilibrium is reached, and concentrations then depend on the space available for solutes within the wall. Consistent with previous results,9 we found that at these longer times, concentrations of macromolecules were higher in the inner and outer layers of the wall than in the middle; this phenomenon is thought to reflect a looser packing of the ground substance of the wall in these regions. The packing density is in turn likely to influence the resistance of the different layers to transport across the wall.
We have demonstrated that ATR–FTIR imaging allows measurement of the diffusion of small solutes, such as drugs, and macromolecules into the arterial wall in vitro. The model drug benzyl nicotinate showed higher affinity for atherosclerotic plaque than for healthy tissue, while ibuprofen required albumin to act as a drug carrier; it was otherwise undetected within the tissue. The presence and distribution of ibuprofen in the artery wall, transported by albumin was demonstrated by multivariate subtraction image analysis. The results were obtained by a chemically specific method, as a function of time with the same specimen, and without labeling the solutes, a unique combination of advantages.
This project was funded by EPSRC Grant No. EP/E003281. We thank Dr. Stephanie G. Cremers for the preparation of the tissues used in this study.