DeOlmos Protocol
Methods for the selective staining of neurons by bath-immersion of tissue sections have been notoriously capricious, ever since the days of Retzius, Cajal, and Golgi. That is why they were largely superceded in the 60's and 70's by dye-injection techniques, in the 80's and 90's by immunocytochemistry, and more and more today by molecular biological techniques for expression of fluorescent proteins in selected neurons (refs). Nevertheless, the need remains to selectively stain neurons in defined physiological states, independent of neuronal cell type or gene-expression profiles.

One clear-cut state needing recognition is incipient or ongoing cell death, such as occurs naturally during development, or occurs agonally during various chemical or physical insults to the nervous system. Since we have long been interested in monitoring cell death during glutamate-induced and other forms of neuronal excitotoxicity (refs), we have always required a consistent and reproducible method of staining dying cells in our experimental animals. The DeOlmos silver-impregnation method has most closely matched our requirements, but when applied to rodents and especially to baby rodents or late-term fetal rodents, our silver-impregnations have required extensive modification to be fully dependable and reproducible.

In this report we spell out these modifications, confident that if they are followed accurately, they will yield consistent and useful results in other investigators' hands as well. Our impetus to publish these details of our methods has come from the many requests we have received to teach the technique first-hand to other investigators in other laboratories, plus the many sad tales we have heard of other investigators learning less from their experiments on neuronal death and survival than they might have, simply due to failures or inconsistencies in their staining of dying neurons.

We should stress that the exact physiological state of the nerves that stain positively to our variation of the DeOlmos technique (or to any other similar technique) is not precisely known, but it correlates very strongly with cells that also react positively to "tunnel" staining (immunocytochemical staining of active caspase 2, a marker of incipient and inevitable neuronal apoptosis). However, our positively-impregnated cells typically include more than just "tunnel"- positive cells, and these we consider to be cells slightly further advanced in apoptosis and/or cells undergoing other forms of cell death. Important as these distinctions are for understanding the relationship between any particular experimental insult to the nervous system and the nerve cell response, they are issues of far broader scope than can be discussed here. This report will be simply a practical guide for successful silver-impregnation of very recently dead and/or dying neurons in young rodents, regardless of the cause of neuronal cell death.


As with any histological technique, the final success of silver-impregnation is determined almost from the start, just by the quality of the initial fixation of the tissue. A poor start will not only produce unsatisfactory tissue architecture, but in practical terms, will result in nervous tissue that is sticky, friable, difficult to vibratome section (a critical step detailed below), and prone to excessively high background-deposition of silver in inappropriate loci, such as in cell nuclei and especially nucleoli. Top-notch fixation requires transcardiac perfusion, first, of carefully prepared buffers that will wash the blood entirely out of the brain, and soon thereafter, of freshly prepared formaldehyde in the same buffers.

As always, perfusion remains an art as much as a science, but we have found the following details to be important for success. First, the volume of buffer passed through the vasculature before introduction of fixative needs to be surprisingly large, perhaps because the vessels of the brain are particularly reactive. Nearly 0.2 liters is required for a good-sized rat (some 10 times its blood volume). Perhaps also because of the reactivity of CNS vessels, this buffer must be passed under as high a pressure and at as high a rate as can be tolerated by the heart/cannula seal that we establish. Temperature is also an important factor here, since cold buffers constrict CNS vasculature and prevent adequate expulsion of erythrocytes from brain capillaries, while buffers warmed to 37˚ yield the most consistently successful perfusions. (For simplicity, however, we routinely profuse room temperature buffers and fixatives.)

Despite the importance of displacing all erythrocytes from the brain and establishing the complete patency of CNS vasculature for uniform fixation, our goal is of course not to study the agonal changes of nervous tissue subjected to prolonged anoxia; hence, the above perfusion of buffer must be completed as rapidly as possible, and we must proceed promptly to perfusion of the fixative per se. The exact physical composition of our fixative is not unusual, and could just as well be used for high quality electron microscopy; nevertheless, it must be rigidly controlled to maintain consistency, and is adapted for the eventual peculiarities that our tissues will experience during silver-impregnation. Thus, phosphate buffers must be totally avoided, as DeOlmos originally recognized, or else grossly excessive background staining will result.


  • if blood is left in the capillaries they stain positive and make the background worse,
  • the fixative buffer needs to be cacodylate (why?),
  • formaldehyde can be 4-10%,
  • it takes ~180 ml of fixative,
  • after perfusion, the brain must be left immersed in the 4% formaldehyde and cacodylate for 2 days before vibratome sectioning,
  • after tissue sectioning, it's even okay to leave sections in formaldehyde in water,
  • the sections can be left in fixative for 3 days to 6 months,
  • if left in the 4% formaldehyde and cacodylate for longer than 6 mos, background increases (especially the nucleolar staining),
  • in discussing poor sectioning, note that it will invariably lead to poor mounting, for obvious reasons.