Journal of Microscopy, Vol. 201, Pt 1, January 2001, pp. 70±76. Received 23 June 2000; accepted 25 September 2000
Techniques for reducing the interfering effects of autofluorescence in fluorescence microscopy: improved detection of sulphorhodamine B-labelled albumin in arterial tissue T. J. STAUGHTON, C. J. MCGILLICUDDY & P. D. WEINBERG School of Animal and Microbial Sciences, University of Reading, Whiteknights P O Box 228, Reading RG6 6AJ, U.K.
Key words. Autofluorescence, filters, fixatives, fluorescence microscopy, sulphorhodamine B, temperature.
Summary
Introduction
Measurements of the transport of circulating sulphorhodamine B-labelled albumin into the arterial wall, made by applying digital imaging fluorescence microscopy to sections of arteries fixed in situ, are limited in sensitivity by the low levels of tracer fluorescence and high levels of autofluorescence emitted from the tissue. Three attempts to improve these ratios are described. In the first, spectra of the tracer in solution and of arterial autofluorescence were used to design novel microscope filters for rhodamine-like dyes. By exciting with the rarely used yellow lines of the mercury arc lamp and detecting a narrow band of emission with Stokes shifts as small as 15 nm, the ratio of tracer fluorescence to autofluorescence was tripled. In the second, effects of different fixatives were investigated. Using a model system, it was confirmed that Karnovsky's fixative gives good tracer immobilization but elevates autofluorescence, whereas fixative-free buffer solutions give low autofluorescence but do not retain the tracer. It was further found that simple formaldehyde-based fixatives, hitherto considered to be poor fixatives of albumin, immobilized the tracer as well as the glutaraldehyde-based fixative, whilst giving autofluorescence levels comparable to those seen with buffer alone; they therefore give excellent tracer fluorescence to autofluorescence ratios. In the third, lowering specimen temperature by 50 8C was found to increase the intensity of tracer fluorescence by 30% whilst autofluorescence was unaffected. These data may have relevance to microscopical studies using other tissues and fluorescent tracers.
Investigations in this laboratory are concerned with the transport of plasma proteins into the arterial wall (Sebkhi & Weinberg, 1994, 1996; Forster & Weinberg, 1997). Transport is studied in animals by labelling serum albumin with the fluorescent dye sulphorhodamine B, administering it intravenously, and subsequently measuring its concentration in sections of arterial wall by using digital imaging fluorescence microscopy (Weinberg et al., 1994). Uptake of tracer by the wall is low, and arterial tissue is highly autofluorescent; the ratio of tracer fluorescence to autofluorescence is therefore of critical importance. Here we describe three attempts to improve this ratio; the methods employed may have more general application. The first investigation concerned the microscope filters. Standard filter sets can be far from optimal for many fluorescent dyes, yet their improvement is problematic and rarely attempted because excitation and emission spectra of fluorophores are usually measured in solution and may be shifted in microscopical preparations. Furthermore, autofluorescence spectra are difficult to determine and usually unavailable. An element of trial and error is therefore involved. Standard filter sets for rhodamine excite with the green 546 nm mercury line and employ long pass barrier filters with maximum transmittance at $ 600 nm. However, sulphorhodamine B conjugates in aqueous solution have an absorbance maximum at 575 nm, a Stokes shift of 20 nm and a narrower emission bandwidth than arterial autofluorescence (Fothergill, 1964; Sartori et al., 1988). Consequently, we assessed custom filters that excite with the yellow mercury lines and pass a narrow emission bandwidth, cutting on as close to the excitation wavelengths as practicable. The second investigation concerned fixatives, which have
Correspondence: Dr P. D. Weinberg. Tel.: 144 (0)118 9875123 ext 7053; fax: 144 (0)118 9310180; e-mail:
[email protected]
70
q 2001 The Royal Microscopical Society
T E C H N IQ U E S F O R R E D U C I N G AU TO FL U O R E S C E N C E
been shown in a few previous studies to profoundly influence tracer fluorescence to autofluorescence ratios. Glutaraldehyde is a good protein fixative but induces intense autofluorescence in arterial tissue (Weinberg, 1989b). Formaldehyde gives low autofluorescence, but Flitney (1965) has reported that it does not fix albumin well. Our previous studies therefore employed formaldehyde and mercuric chloride (formal sublimate), which gives low autofluorescence and good fixation, although quenching of fluorescence after prolonged fixation has been reported (Eastham, 1968). Recently, however, we found that mercury vapour was evolved during in situ perfusion with this fixative, making it too hazardous for routine use. Consequently, we have re-investigated the suitability of aldehyde fixatives. Effects of both formalin and paraformaldehyde solutions on tracer mobility and autofluorescence were studied, and the influence of pH, reported to affect the efficiency of fixation, was also examined. The third investigation concerned specimen temperature. Temperature is known to affect the quantum efficiency of fluorescence (Rost, 1992). We investigated the possibility that tracer fluorescence and autofluorescence are differentially quenched, hence allowing the ratio of their intensities to be improved by heating or cooling. We hypothesized that the rigid, planar sulphorhodamine B molecule would be less affected by temperature than the species responsible for autofluorescence.
Methods General Bovine serum albumin (BSA, Sigma or First Link UK Ltd) was labelled with sulphorhodamine B (Lissamine rhodamine B, CI 45100, Sigma) as previously described (Weinberg et al., 1994). To manufacture fluorescent standards (Weinberg, 1989a), the tracer was diluted 1 : 100 in molten 12.5% wt/ vol. gelatin (Type B, Sigma). In some experiments, tracer was omitted in order to determine levels of autofluorescence from the gelatin alone. The gelatin was cooled to 4 8C in 200 mL drops which, when set, were placed in the fixative solutions described below, dehydrated in ethanol, embedded in epoxy resin (TAAB 812 epon substitute, TAAB Laboratories Equipment, Aldermaston, U.K.) (Glauert, 1991) and sectioned at 2 mm. Sections were mounted on coverslips (to reduce autofluorescence from glass) and digital images of their emission were obtained using an epifluorescence microscope (Zeiss Axioplan) and the various objective lenses, filters and cameras described below. Digital images were processed using the custom software previously described (Weinberg et al., 1994) or slight modifications of it, unless otherwise stated. Briefly, D.C. offsets, stray light and autofluorescence from glass components were subtracted, and spatial biases in the microscope and detection system were removed by a flatfield correction. q 2001 The Royal Microscopical Society, Journal of Microscopy, 201, 70±76
71
Adjustments were also made for camera gain settings (where appropriate), fluctuations in the exciting illumination, and photobleaching. An average pixel intensity was then calculated for each image.
Effect of changing microscope filters Autofluorescence and tracer fluorescence were measured with a conventional rhodamine filter set (546 ^ 6 nm bandpass excitation filter, FT 580 nm dichroic mirror, 590 nm long pass barrier filter, Zeiss), typical of those supplied with microscopes, and with a novel set manufactured to a detailed specification by Omega Optical (Brattleboro, VT, U.S.A.). This consisted of a 577 ^ 7 nm bandpass excitation filter (transmission . 50% for most of its bandwidth), a 610 ^ 10 nm bandpass barrier filter (transmission . 80% for most of its bandwidth) and a 590 nm dichroic mirror (transmission . 90% for most of the barrier filter bandwidth). The use of six-cavity interference filters with sharp cut-ons and cut-offs permitted a separation of only 15 nm between excitation and emission wavelengths at 50% of maximum transmission. Tracer fluorescence was assessed by using the tracercontaining gels described above. (The tracer concentration was sufficiently high for the influence of gelatin autofluorescence to be ignored.) As in all our previous studies, the gels were fixed for 20 min in formal sublimate. Autofluorescence was assessed by using rabbit thoracic aorta which had been fixed in situ with the same fixative, as previously described (Sebkhi & Weinberg, 1994, 1996), and then embedded and sectioned as for the gels. A detection system based on an intensified CCD camera (Darkstar, Photonic Science, Robertsbridge, U.K.), described in detail by Weinberg et al. (1994), and a 40 oil objective were used. Pairwise comparisons were made, each specimen being imaged with both the conventional and the new filter sets. Eight autofluorescence and eight tracer specimens were examined in this way, eight images being averaged for each specimen with each filter set.
Effect of different fixative solutions The ability of paraformaldehyde and formalin solutions to fix tracer in gels and their effects on autofluorescence were investigated at various pHs. Karnovsky's fixative, known to immobilize proteins well but to give high levels of autofluorescence, was used as a positive control, and buffer solutions (no fixation and low autoflorescence) as negative controls. In this experiment and the one described below, precise spectral characteristics of the fluorescence were not important. Consequently, autofluorescence was studied by using tracer-free gelatin rather than artery. Samples were placed in 5 mL of fixative solution or pHmatched buffer alone. The fixatives used were 4% wt/vol.
72
T. J. S TAU G H TO N E T A L .
paraformaldehyde, 10% neutral buffered formalin, and Karnovsky's fixative (4% wt/vol. glutaraldehyde and 5% vol/vol. formaldehyde; Karnovsky, 1965). Sodium phosphate buffer (80 mm, pH 6.4, 8 or 11) was used to dissolve the paraformaldehyde. Formalin was supplied in an identical buffer at pH 7.4, and was also examined at pH 11 by adding NaOH. Three experiments were conducted with each solution. The loss of labelled albumin from the gelatin after 24 h of fixation was measured by two methods. First, the concentration of tracer that had leached into the fixative solution was assessed from the absorbance of the solution at 575 nm (lmax for the tracer). Results were obtained in triplicate and expressed as the percentage loss of tracer from the gelatin. Secondly, the concentration of tracer remaining in the gels was determined by applying digital imaging fluorescence microscopy to sections cut from them. Six sections from each gel were examined, using a 40 oil objective and the novel filter set described in the preceding section. The intensified CCD was replaced with a cooled CCD (AX-2 Viper, Axiom Research, with Kodak KAF 1600 chip), which had better stability and uniformity, and lower noise. The previously described software (Weinberg et al., 1994) was modified for the absence of gain controls and the replacement of frame averaging by prolonged exposures (10 s at 27 8C).
Effect of specimen temperature A custom specimen holder was constructed in order to modify the temperature of sections. It consisted of a brass slab (70 55 6 mm) in which a serpentine channel was bored so that a cooling or heating fluid could circulate through it. The holder was thermally insulated from the microscope stage by expanded polystyrene foam; a hole drilled through the holder and foam allowed exciting and emitted light to pass to the substage. The fluids used were water and, for the sub-zero temperatures, ethanol. Equilibrium temperatures were measured by using thermal indictor strips (RS Components Ltd, Corby, U.K., 0.1 8C accuracy) mounted on coverslips and placed in the same position as the specimens. The strips did not extend to the lowest specimen temperatures, which were therefore estimated as the fluid temperature plus the difference observed between fluid temperature and specimen temperature in the other experiments. The error would have been a few degrees at most, which would not have changed any conclusions. Sections of formalin-fixed gels were used to assess both tracer fluorescence and autofluorescence. They were imaged by using the novel filter set and the cooled CCD. Because the effects of extreme temperatures on the refractive index of immersion oils and on glass optics were unknown, a 25 dry objective was used. Each section was repeatedly imaged without being moved, so high accuracy could be obtained by
quantifying the intensity from a small region and observing its fractional change with temperature. Flatfield and other corrections were therefore not required (although offsets still needed to be subtracted), and commercial software (Mira A/P, Axiom Research, Tucson, AZ) could be employed. For each section (n 3±9 at each temperature), three readings were taken at room temperature, three after the temperature had been changed, and three when it was returned to room temperature. The middle value obtained at the altered temperature was expressed as a percentage of the initial intensity. To obtain an equivalent percentage for room temperature, the values obtained just before and just after the temperature change were similarly expressed as a percentage of the initial temperature. The average room temperature value did not equal 100% of the initial intensity, despite the absence of temperature change, because the repetitive imaging caused significant photobleaching. The correction factor required to bring this average to 100% was calculated and applied to all values in order to correct for photobleaching. Separate corrections were calculated for tracer fluorescence and autofluorescence, as they bleached at different rates.
Results Filters Mean intensities of tracer fluorescence and arterial autofluorescence are shown for the conventional and novel filters in Figs 1(a) and (b), respectively. Pairwise comparisons of each specimen examined with each filter set showed that the novel filters reduced autofluorescence 17.4-fold (SD 2.84, n 8) and dye fluorescence 5.4-fold (SD 0.54, n 8). Thus, at the cost of a five-fold decrease in intensity, the ratio of tracer fluorescence to autofluorescence was tripled.
Fixatives The effects of different fixatives on tracer fluorescence and gelatin autofluorescence are shown in Figs 2(a) and (b), respectively. The positive and negative controls behaved as expected: there was a high level of tracer fluorescence and an exceptionally high level of autofluorescence from gels fixed with Karnovsky's fixative, and low levels of tracer fluorescence and autofluorescence when sodium phosphate buffers were used. Unexpectedly, however, paraformaldehyde and formalin at their optimum pHs gave intensities of tracer fluorescence that at least equalled those obtained with Karnovsky's fixative (Fig. 2a) and autofluorescence levels similar to those obtained with buffer alone (Fig. 2b). (The striking difference in autofluorescence levels produced by formalin and glutaraldehyde-containing fixatives has previously been observed in sections of artery [Weinberg, 1989b], supporting q 2001 The Royal Microscopical Society, Journal of Microscopy, 201, 70±76
T E C H N IQ U E S F O R R E D U C I N G AU TO FL U O R E S C E N C E
73
Fig. 1. Effects of different microscope filters on the intensity (in arbitrary units) of fluorescence emitted from sections of (a) gels containing tracer, and (b) rabbit aorta (autofluorescence). Specimens of each type were imaged with (1) a conventional filter set for rhodamine (BP546 ^ 6 nm excitation filter, FT 580 dichroic mirror, LP 590 barrier filter, Zeiss) and (2) a novel filter set (577 ^ 7 nm excitation filter, 590 nm dichroic mirror and 610 ^ 10 nm barrier filter). Bars show mean ^ SEM, n 8.
Fig. 2. Effects of fixative and buffer solutions on the intensity (in arbitrary units) of fluorescence emitted from gels (a) containing tracer and (b) not containing tracer (autofluorescence). The gels had been immersed in the indicated solutions for 24 h before being embedded and sectioned. Each bar shows the mean (^ SEM) of three experiments in which intensities from six sections were measured. q 2001 The Royal Microscopical Society, Journal of Microscopy, 201, 70±76
74
T. J. S TAU G H TO N E T A L .
Fig. 3. Loss of tracer from gels immersed for 24 h in fixative or buffer solutions. Losses, expressed as a percentage of the tracer initially present in the gels, were calculated from the absorbance of the fixative and buffer solutions at 575 nm. Absorbances of comparable solutions not exposed to gels were subtracted. Each bar shows the mean (^ SEM) of three experiments in which absorbances were measured in triplicate.
the view that the model system employed in the present study gives results of relevance to biological specimens.) Overall, paraformaldehyde did not perform better than formalin. For formalin, tracer fluorescence was higher at pH 7.4 than at pH 11, although this difference did not reach significance (t 1.58, P . 0.05). For paraformaldehyde, tracer fluorescence was approximately one-third higher at pH 11 than at the lower pHs (t 2.70, P , 0.05). Figure 3 shows absorbances of the solutions after gels containing tracer had been immersed in them for 24 h; a failure to immobilize the tracer would have resulted in a higher absorbance. In general, there was good agreement between this data and that in Fig. 2(a): low fluorescence was observed from those gels where a high percentage of the tracer leached into the fixative solution and, conversely, intensities were high when tracer loss was low. Up to 70% of the tracer leached out of the gels over a 24-h period when buffer but no fixative was present. Additional tracer could have leached out during dehydration and embedding, so this is a minimum estimate of the total loss from the gel. The best fixatives retained 80±90% of the tracer, although there may again have been additional (presumably smaller) losses at later stages. In these data, which were less variable than those obtained by sectioning the gels, the difference in the efficiency of formalin fixation at pH 7.4 and pH 11 was significant (t 5.98, P , 0.01). There were two exceptions to the generally good inverse relation between the absorbance of the fixative solution and the intensity of fluorescence from the corresponding gels. First, the absorbance of paraformaldehyde solutions at low pH seemed too high to be compatible with the section data; it was comparable with that seen for buffer alone. However, a substantial absorbance was observed for paraformaldehyde at low pH even when the solution was not exposed to gels containing tracer; it was caused by the fixative solution becoming turbid. Although such control absorbances were
subtracted from experimental values, this correction may have been insufficient: for example, turbidity might have increased when tracer leached from the gels. The second exception was that Karnovsky's fixative gave a somewhat lower fluorescence intensity than expected from the corresponding absorbance of the solution, the latter being the lowest observed. This fixative may partly quench tracer fluorescence, perhaps by forming the extensively crosslinked structures that have been implicated in its influence on autofluorescence (Weinberg, 1989b).
Temperature Effects of temperature on individual sections were consistent and reversible. Mean effects at each temperature are shown
Fig. 4. Effect of temperature on the intensity of tracer fluorescence and autofluorescence. Intensities from sections of gels containing or not containing tracer were measured at room temperature, and at raised or lowered temperatures. Data were expressed as a percentage of the intensity at room temperature, and corrections were made for effects of photobleaching. Mean ^ SD, n 3±9 at each temperature. q 2001 The Royal Microscopical Society, Journal of Microscopy, 201, 70±76
T E C H N IQ U E S F O R R E D U C I N G AU TO FL U O R E S C E N C E
in Fig. 4. There was a 30% increase in tracer fluorescence as specimen temperature was decreased by 50 8C from room temperature. The relation between temperature and intensity was linear. Raising the temperature to 45 8C produced a 12% drop in tracer intensity, consistent with the slope of 0.6% per degree seen when temperatures were reduced, but above 45 8C the change was negligible. Over the same temperature range, there was no discernible change in the intensity of autofluorescence (Fig. 4). Thus, contrary to expectation, it was tracer fluorescence, not autofluorescence, that increased when temperature was lowered. Fluorescence from the coverslips on which the specimens were mounted, which was subtracted from tracer fluorescence and autofluorescence values, was similarly unaffected (data not shown).
Discussion Filters The novel microscope filters tripled the ratio of tracer fluorescence to autofluorescence. Part of the improvement is likely to have been caused by the < 30 nm increase in excitation wavelength. This will have resulted in the tracer being excited more efficiently (by . 50%, if the absorption spectra in aqueous solution are applicable); it may also have decreased autofluorescence, which is generally less well excited by longer wavelengths, although the benefit cannot be estimated because the excitation spectrum of arterial autofluorescence has not been reported. The remaining part of the improvement will have resulted from the decrease in the cut-off wavelength of the barrier filter, because the intensity of tracer fluorescence decays to 20% of its maximum by 620 nm (Fothergill, 1964), whereas arterial autofluorescence decays to only 60% of its maximum (Sartori et al., 1988). The new filters also caused a five-fold reduction in overall intensity. This may, in part, reflect the 20 nm bandwidth of the barrier filter; the long-pass filter in the conventional set would transmit more tracer emission. Furthermore, although all the yellow lines of the mercury lamp were used for excitation, even their combined intensity may be less than that of the bright 546 nm line conventionally used for rhodamine dyes. The five-fold decrease in intensity is not a serious drawback; with the high sensitivity of modern cameras, detection is more often determined by tracer fluorescence to autofluorescence ratios than by tracer fluorescence alone. Indeed, even with the new filters we can use a neutral density filter in the excitation lightpath to reduce photobleaching. Thus, the filters provide a substantial net benefit. The results obtained for tracer in gelatin standards are expected to apply to tracer in arterial specimens because sulphorhodamine B is relatively insensitive to environmental q 2001 The Royal Microscopical Society, Journal of Microscopy, 201, 70±76
75
factors (such as pH; Fothergill & Ward, 1976); the resin embedding medium will in any case dominate the environment in both cases. The results may also have relevance to other tracers and tissues. Many long-wavelength dyes could best be excited with the yellow mercury lines. These include not only rhodamine-based dyes such as rhodamine-Red X but also newer fluorophores such as Alexa 568 and 594, Bodipy 581/591 and Phycoerythrin. Furthermore, autofluorescence tends to derive from several tissue components, and thus it generally has a broader emission spectrum than single fluorochromes. Consequently, restricting emission frequencies to a narrow band near the maximum for the fluorochrome is likely to reduce effects of autofluorescence in non-arterial tissues.
Fixatives Formalin at neutral pH gave intensities of tracer fluorescence as high as those obtained with Karnovsky's fixative, but with autofluorescence levels not exceeding those seen when using buffer alone. Hence, this cheap and readily available fixative gives an excellent tracer fluorescence to autofluorescence ratio, at least as good as the hazardous fixatives based on mercuric chloride which we previously employed. The high level of tracer fluorescence was surprising given previous evidence that formalin is poor at immobilizing bovine serum albumin in gelatin (Flitney, 1965). The discrepancy may reflect small but significant methodological differences. Flitney examined fixation in thin sections of gelatin; furthermore, the gelatin had been frozen and thawed prior to the experiment, probably leading to disruption by ice crystals. The diffusion of tracer out of the gelatin in the absence of fixative was consequently much more rapid (50% in 1.3 min) than would be expected for its diffusion out of the arterial wall and other tissues. Additionally, fixation was restricted to 2 h, whereas complete fixation by formaldehyde takes considerably longer (Kiernan, 1990; Bacallao et al., 1995), and the high tracer concentration (5% w/w) would have favoured cross-linking between tracer molecules themselves rather than between tracer molecules and the stationary phase. Preliminary data from in vivo experiments employing formalin fixation (Staughton & Weinberg, 1999) show good tracer retention, and support the view that the current models have greater relevance to physiological studies. Overall, paraformaldehyde gave fluorescence intensities similar to those obtained with formalin solutions, but the dependence on pH differed. The better fixation by paraformaldehyde at pH 11 than at neutral or acidic pHs probably reflects the fact that the rate at which formaldehyde can cross-link proteins is limited by deprotonation of amino groups (Kiernan, 1990; Bacallao et al., 1995), a step favoured by an alkaline pH. It is not clear why formalin
76
T. J. S TAU G H TO N E T A L .
solutions behaved differently. Formaldehyde reacts with water to form methylene glycol in both paraformaldehyde and formalin solutions (Kiernan, 1990). Furthermore, although soluble polymers of formaldehyde are present in formalin, depolymerization tends to occur when it is diluted (Kiernan, 1990). The composition of the fixative in the two solutions should thus have been similar. Differences in the effects of pH may have arisen as a result of the additional presence in the formalin of methanol (added as a stabiliser to inhibit polymerization) or formic acid (formed by the Cannizarro reaction during prolonged storage) (Fox et al., 1985; Kiernan, 1990).
Temperature The experiments in which specimen temperature was altered are, to our knowledge, the first to investigate whether this technique might be a useful way of improving tracer fuorescence to autofluorescence ratios. The intensity of tracer fluorescence was increased by lowering specimen temperature, whereas autofluorescence was unaffected. Although this result is interesting, the technique is unlikely to provide a useful benefit. The 30% improvement in tracer intensity at 224 8C does not compensate for the fact that immersion lenses cannot be used at such low temperatures: the loss of resolution and intensity is too great. However, cryogenic temperatures might be worth investigating if problems such as condensation could be overcome. Extrapolation of our results suggests that cooling with liquid nitrogen could increase tracer fluorescence by 130%. It is also possible that other fluorochromes might show a greater fractional increase in emission on cooling.
Acknowledgements This study was funded by the British Heart Foundation and the Royal Society. We thank Dr A. Sebkhi for providing tissue and Mr C. Grilli for building the thermal stage.
References Bacallao, R., Kiai, K. & Jesaitis, L. (1995) Guiding principles of specimen preservation for confocal fluorescence microscopy. Handbook of Biological Confocal Microscopy (ed. by J. B. Pawley), pp. 311±325. Plenum Press, New York. Eastham, W.N. (1968) Fluorescence in paraffin sections of
dye-labelled protein administered in vivo; effect of HgCl2 fixation. Stain Technol. 43, 59±61. Flitney, F.W. (1965) The time course of the fixation of albumin by formaldehyde, glutaraldehyde, acrolein and other higher aldehydes. J. Roy. Microsc. Soc. 85, 353±364. Forster, B.A. & Weinberg, P.D. (1997) Changes with age in the influence of endogenous nitric oxide on transport properties of the rabbit aortic wall near branches. Arterioscler. Thromb. Vasc. Biol. 17, 1361±1368. Fothergill, J.E. (1964) Properties of conjugated proteins. Fluorescent Protein Tracing 2nd edn (ed. by R. C. Nairn), pp. 34±59. E and S Livingstone, Edinburgh. Fox, C.H., Johnson, F.B., Whiting, J. & Roller, P.P. (1985) Formaldehyde fixation. J. Histochem. Cytochem. 33, 845±853. Glauert, A.M. (1991) Epoxy resins: an update on their selection and use. Microsc. Anal. September, 15±20. Karnovsky, M.J. (1965) A formaldehyde-glutaraldehyde fixative of high osmolarity for use in electron microscopy. J. Cell Biol. 27, 137A. Kiernan, J.A. (1990) Histological and Histochemical Methods. Theory and Practice. Pergamon Press, Oxford, pp. 10±31. Rost, F.W.D. (1992) Fluorescence Microscopy, Vol. I. Cambridge University Press, Cambridge, pp. 33±34. Sartori, M., Weilbaecher, D., Valderrama, G.L., Kubodera, S., Chin, R.C., Berry, M.J., Tittel, F.K., Saurbrey, R. & Henry, P.D. (1988) Laser-induced autofluorescence of human arteries. Circ. Res. 63, 1053±1059. Sebkhi, A. & Weinberg, P.D. (1994) Age-related variations in transport properties of the rabbit arterial wall near branches. Atherosclerosis, 106, 1±8. Sebkhi, A. & Weinberg, P.D. (1996) Effect of age on the pattern of short-term albumin uptake by the rabbit aortic wall near intercostal branch ostia. Arterioscler. Thromb. Vasc. Biol. 16, 317±327. Staughton, T.J. & Weinberg, P.D. (1999) Effect of rabbit strain on age-related changes in the pattern of arterial wall permeability. Atherosclerosis, 147, 210 (Abstract). Weinberg, P.D. (1989a) Densitometry of photomicrographic negatives for the determination of fluorophores in sections of tissue. Anal. Chimica Acta, 227, 235±241. Weinberg, P.D. (1989b) Effect of tissue processing techniques on the sensitivity of microfluorimetric determinations of tracers in tissue sections. Anal. Chimica Acta, 227, 243±249. Weinberg, P.D., Winlove, C.P. & Parker, K.H. (1994) Measurement of absolute tracer concentrations in tissue sections by using digital imaging fluorescence microscopy. Application to the study of plasma protein uptake by the arterial wall. J. Microsc. 173, 127±141.
q 2001 The Royal Microscopical Society, Journal of Microscopy, 201, 70±76
