The place was Ann Arbor, University of Michigan, in the spring of 1994. Three years earlier, I had switched from studying extracellular matrix proteins to investigating how death receptors induced apoptosis. At the time, only two death receptors were known: tumour necrosis factor receptor 1 (TNFR1) and Fas/CD95. It was a risky move as all my grant support was for research on matrix biology, but it felt like I was just dotting the i's and crossing the t's in this field. I wanted a larger canvas to paint on, and the opportunity to make my mark in an emerging field. Despite knowing that I needed a bigger challenge, I was filled with insecurity and uncertain whether the gamble would pay off. The dire consequences of failure were never far from my mind, especially as an untenured assistant professor. I did not want to be the one demonstrating no progress to the funding agencies, having frittered away their precious resources on some harebrained idea!

So, since I was hell-bent on changing fields, what was I going to work on? Cell cycle? Exciting, but way too crowded. How about the opposite process? Cell death. Now that was intriguing to a pathologist with a morbid streak! In the early nineties, little was known. The morphological description of apoptosis and the beginnings of a pathway were being worked out in the worm Caenorhabditis elegans. I had never seen these worms, so that was not the route for me. What I needed was a cell culture system that was amenable to the induction of apoptosis. Various noxious chemicals caused cells to undergo apoptotic demise, but there was no information on their targets. Perusing the literature, I was captured by observations that one could induce apoptosis in certain cultured cells either with TNF, the cognate ligand for the TNFR, or with an agonist antibody to the Fas receptor. Hurriedly, I wrote to Genentech to obtain recombinant TNF and they supplied me with a veritable ton of material. Dr Yonehara in Japan also kindly provided the agonist anti-Fas antibody.

The first experiments with these 'death ligands' and MCF7 cultured cells were amazing. Initially, the cells did just fine, but after a few hours they underwent a catastrophic 'dance of death'. There was violent blebbing of the cell membrane, dramatic condensation of the nucleus, and the cytoplasm became vacuolated as if on a high boil. Wow! Now here was a pathway worthy of study. What were the death-inducing components engaged by death receptors? There was a body of literature implicating kinases, ion channels, phospholipases, and phosphatases, but I favoured something new. The contribution of the known suspects seemed partial, as if they were peripheral to the central process. So what could it be? An MD/PhD graduate student in the laboratory, Muneesh Tewari, was unleashed on the project. Bright, ambitious, and with a desire to accomplish something important, he tried everything under the sun to inhibit the pathway. Overexpression of cDNA libraries and application of kinase, phosphatase, and lipase inhibitors all yielded nothing. The inhibition, when observed, was always partial.

We were also running out of time. Muneesh needed to graduate and I needed to renew my grants. Something had to give or we were going to be in a whole heap of trouble. Fortunately, a vital clue emerged from the world of C. elegans. The worm death gene, ced3, was discovered to encode a protein homologous to a mammalian cysteine protease called ICE (interleukin-1 converting enzyme). Could ICE or an ICE-like protease operate in the death receptor pathway? I realized this notion could be tested because there was a cowpox virus inhibitor of ICE called CrmA. All we needed to do was transfect CrmA into MCF7 cells and see if we conferred resistance to TNF or the agonist anti-Fas antibody. A quick note to David Pickup at Duke University, North Carolina, garnered us the CrmA expression construct. I was working late, probably on a grant, when Muneesh came running into my office. His face was filled with incredulity, as though he had just unearthed the most astonishing treasure. He pulled me to the microscope and showed me vector-transfected MCF7 cells and CrmA-expressing cells, both exposed to TNF. The vector-transfected cells were dead as door nails but, lo and behold, the CrmA-expressing cells looked the image of health. They were totally oblivious to the added TNF. Indeed, they were proliferating, indicating that CrmA had totally eliminated the death signal. The inescapable conclusion was that a component of the death receptor pathway was ICE or a related protease.

In short order, work from a number of laboratories, including our own, defined a number of mammalian ICE-like proteases, renamed caspases, which were able to induce apoptosis. The question that remained was, how do death receptors activate caspases? At that time it was thought that receptors either served as ion channels or altered phosphorylation and dephosphorylation events. Death receptors, it was assumed, must signal in a similar fashion. Here came the second major surprise: postdoctoral fellow Marta Muzio and graduate student Arul Chinnaiyan were able to show conclusively that death domain-containing receptors signalled through an entirely new mechanism, by generating a protease as the 'second messenger'! Processing of downstream caspase zymogens caused precipitous cleavage of cellular substrates and a rapid apoptotic demise of the cell. The rest is history.