The shortest fixation time allowing the APO866 in vivo maintenance of intact sections throughout the procedure was 45 min. We tested a battery of antibodies

against various classes of proteins, using tissue routinely fixed by transcardiac perfusion with a 4% paraformaldehyde solution as comparison. Using immunoperoxidase staining, all antibodies tested produced regional immunoreactivity patterns that were at least as well discernible, or better, in sections from immersion-fixed tissue as from perfusion-fixed tissue. Figure 1 depicts comparative immunostaining patterns of CD68, glial fibrillary acidic protein (GFAP), synapsin 1, tyrosine hydroxylase (TH) and serotonin (5-HT) in perfusion-fixed and immersion-fixed tissue. Optimal signal-to-noise ratio, as assessed qualitatively, was obtained in sections from blocks postfixed for 3 h, and this time-point was selected here for illustration. CD68 and GFAP were tested in sections prepared from adult (3 months; perfusion-fixed) and from old mice (19 months; immersion-fixed), but this difference in age had no influence on the quality of the staining. As expected, staining of cytoskeletal proteins (e.g. GFAP) showed little influence from the duration of fixation, Inhibitor Library and a longer post-fixation either had no effect or led to a slight decrease in immunoreactivity (not shown). Abundant transmembrane proteins, such as CD68, myelin-basic

protein or vesicular GABA transporter, likewise showed little dependence on post-fixation duration, and could be detected at high sensitivity

in tissue fixed for 1–6 h. The same result was obtained with transmitter-synthesizing enzymes, for example TH, and with small molecules, such as 5-HT. A pretreatment of sections Pyruvate dehydrogenase with pepsin to better expose fixation-sensitive epitopes yielded similar antigen-retrieving effects in immersion-fixed tissue and in perfusion-fixed tissue (not shown) and did not damage the tissue during handling of free-floating sections, indicating that such procedures are compatible with immersion-fixation of living tissue. In our protocol, there is no blocking step prior to incubation in primary antibodies, and endogenous peroxidase activity is not quenched with H2O2. These two steps were skipped, because they bring no improvement to the quality of immunoperoxidase staining in rodent tissue, when it is adequately fixed. Interanimal variability, reflecting the quality of perfusion, was low and comparable among perfusion-fixed and ACSF-perfused mice (not shown). Immunofluorescence staining and imaging by confocal laser scanning microscopy was performed to assess subcellular distribution of neuronal markers, such as the calcium-binding protein parvalbumin (Fig. 2A) or the GABAAR α2 subunit (Fig. 2B and C), as well as eGFP in transgenic mice expressing GAD67-eGFP (Tamamaki et al., 2003) (Fig. 2D and E) and in adult-born neurons labeled with a retrovirus encoding eGFP (Fig. 2F and G) (Duveau et al.