Identification of chaperones as essential components of the tumor rejection moieties of cancers

During my graduate work (1979–81) at the Center for Cellular and Molecular Biology in Hyderabad, India, and far away from the epicenter of cancer immunology, I had identified a tumor-derived protein of 100,000 daltons (d) that could elicit tumor resistance in rats immunized with it (1). I had moved on to postdoctoral work in developmental biology at Yale University. Since my graduate work was done pretty much by myself in an environment where there was not much knowledge of immunology, I was unsure if my excitement about it was justified. To put some context to it, I sent the then-unpublished manuscript to Lloyd Old, whose name I knew only from the literature. I had read with intense delight many of his papers (mainly with Albert DeLeo) attempting to define the unique antigens of mouse sarcomas [see (2) for review], and I reasoned that he would be an ideal person to comment on my fledgling work. I did not hear from Old for about two months, and my repeated calls to his office were rebuffed by his secretary whom I would later come to know and love as Charlotte Daub. Frustrated by my inability to reach Old, I contacted the New York Telephone Operator and acquired his home telephone number.

On a Saturday morning at about 8 a.m., I called him. I was prepared for any response but for the one that I received. Old acknowledged that he had indeed received my manuscript and had read it carefully. He then went on to summarize my findings in extraordinary detail right down to the name of the unusual tumor that I had used, to the low titers of the antibodies that I had made, to the antibodies being IgM and not switching to IgG, and to the size of the antigen recognized by them. He commended me profusely and thanked me for sending the manuscript to him. I was flattered that this distinguished scientist would find my first piece of work so absorbing. I was also puzzled. I asked him why he did not respond to my letter if he had read my paper so carefully and had enjoyed it so much. He simply told me to come see him. I took the train from New Haven, Connecticut, and made my first trip to New York. I am not quite sure what I expected from our meeting, but I recall being very surprised: Old was full of humor, and immensely and intensely curious. He wanted to know intricate details of my work, my career plans, my likes and dislikes, my personal background. His interests seemed to cover a wide territory, and yet all roads did lead to cancer. He seemed infinitely interested in cancer and was filled with boundless belief that immunotherapy was the way to treat it. Charlotte had given me 30 minutes with him. I left after about two hours with a spring in my step: my work done in isolation more than 7,000 miles away was fascinating to Lloyd Old!

Eventually, I did join his laboratory as a postdoctoral fellow of the Cancer Research Institute (CRI). His laboratory at the time was actually a vast empire consisting of perhaps over a hundred people across two campuses, taking up many floors at each. Having made several major discoveries using the mouse system over the preceding two decades, Old was a towering presence in the field, and was now focusing his attention mostly on the immunology of human cancers. His laboratory was busy defining the cell surface antigens of human cancers. I had no interest in human work at the time, and I was drawn to his laboratory entirely because of his previous work with the mouse system. Fortunately for me, Old was content with having me continue my previous work on defining the tumor rejection antigens of experimentally induced cancers. Thus, I came to be on the 10^th floor of Memorial Sloan-Kettering Cancer Center’s Howard Building, working in a small unit with Albert DeLeo and Elisabeth Stockert. The other postdocs in the Old lab, working on the human cancers on the 9^th floor of the neighboring Schwartz Building, included Alan Houghton, Tony Albino, Paco Real, Tim Thompson, Wolfgang Rettig, and others.

I continued to fractionate the mouse tumor extracts by column chromatography and to test the fractions by tumor rejection, in the manner in which I had done it in Hyderabad. Much to my surprise, the active antigen identified in the mouse fibrosarcomas turned out to be apparently the same ∼100,000 d protein that I had identified as the tumor rejection antigen from a rat hepatocarcinoma. After a more precise size calculation, I named it gp96.

My interactions with Old were episodic during my first year as a fellow. He would often call Albert DeLeo to speak on the phone, and would occasionally ask to speak with me, as well. I could not but notice that he always seemed irrationally exuberant (although Alan Greenspan had not yet coined the term). At some point, Albert suggested that I put together my data and have a formal meeting with him. I gathered about 50 or so figures and tables, and the three of us had the meeting in his office, now on the 10^th floor of Schwartz. We met for about two hours, amid several admonishments from Charlotte. Old seemed absorbed in the data, with many, many comments and allusions, a number of which I only dimly understood. Our conversation about data was interwoven in a seamless tapestry with discussions about our ancestries (Swiss-German, Indian, and Italian), the nature of New York City, the merits and demerits of various cuisines, and Mozart. He seemed incredulous, almost hurt, that I did not find the music of Mozart to be the most exalted sound that I had ever heard, and worse, that I had never heard of Robert Schumann! (About 25 years later, we would bridge that cultural divide when, to Old’s great delight, my son Vasishth Srivastava, born in New York City and raised in Connecticut, would discuss the music of Mozart with him.)

Soon after my first meeting with Old, we established a pattern of communication that we would maintain for the next 25 years. He learned that I spent most Sunday afternoons in the laboratory, and he took to calling me in the lab at that time. If, for some reason, I was not in the lab, I would call him from my apartment or a pay phone. We changed the timings and venues of these calls, but for the next 25 or so years, we spoke almost every weekend (with no more than 5–10 exceptions) for anywhere from 20 to as long as 120 minutes.

I wanted to purify large quantities of gp96 but my arcane chromatography system was not up to the task. I brought this to Old’s attention; he gave me a telephone number and asked me to call a woman named Betsy Carswell. I tried to ask him more about this number and Betsy Carswell, but he did not elaborate. I made the requisite call, and was told by Betsy to visit her in Rye, New York. All of this had a bit of Alice-in-Wonderland quality for me. I did not know who Betsy was, nor what Rye was. I found myself one day carried by a shuttle from Sloan-Kettering’s NYC campus to the Rye campus and was ushered into the sharp, warm, and brilliant presence of Betsy Carswell, who then introduced me to the wonders of Fast Performance Liquid Chromatography, and to Barbara Williamson (aka Miss Doodles), Jay Prendergast, and others at “the country club.” Through many visits to Rye, I purified enough gp96 to submit it to the protein microsequencing facility at The Rockefeller University, and listened in awe several weeks later as Diane Atherton of that facility read off the amino terminal sequence of gp96 to me (the sound still rings in my ears: E V D V D G T V…) through the earpiece of a black telephone.

Old was thrilled that I had identified a tumor rejection antigen from chemically induced mouse sarcomas, and was even more thrilled that it showed individually specific immunogenicity, strictly dependent on the source from which it was isolated. He communicated this finding to the Proceedings of the National Academy of Sciences (3). Writing this manuscript with him was a major experience for me. Although this was my first paper with him, I had written about six first-author papers before. I had written all these papers in English mostly by myself and had them published in good peer reviewed journals. I had a certain confidence in my writing abilities. I submitted the first draft of the PNAS manuscript to him, expecting minor alterations and a final draft soon to follow. We went through 26 drafts, with my discontent brewing closer and closer to an exploding point with each revision. He sensed my discontent, but never acknowledged it. I can summarize his “teachings” on writing a manuscript in the following five bullet points:

1. Writing a paper was like extracting orange juice—one had to extract every last drop of juice from the orange, every last bit of meaning from the data.

2. A well-written paper was such that a single syllable added to it would seem superfluous, and a single syllable deleted would deplete it of meaning; writing had to be crystalline. As someone who deeply admired the writing of V. S. Naipaul, I found this lesson appealing, even in those days.

3. I had shown that while a few micrograms of gp96 could immunize mice against the tumor, the use of five times more gp96 (which was still a very tiny amount of protein) failed to elicit tumor immunity. When the manuscript was well advanced, he asked me why I had not included those data in the paper. I told him that the results were bizarre, and I simply could not understand them, and so I had decided not to include them for now. He asked how many times I had performed the experiment, to which I responded three times. He then asked if there was something untrustworthy about the data; I told him there was not—the controls worked fine, the tumor curves were close together in each group, and so on. He then told me to go ahead and include the data in the paper, and simply say that their basis was difficult to explain. He insisted that the paper tell the whole story, not simply the story that I could explain. (In our current publishing environment, that admirable lesson seems quaint in hindsight.) Many years later, Rajiv Chandawarkar and I (and others) would dissect the mechanistic basis of dose restriction of the immunogenicity of gp96 (4–6).

4. In writing a legend to a Table, I had copied verbatim a part of a sentence from another paper, which had described the same method. I told Old that the other paper had written it more succinctly than I had in my draft; hence I chose the words from it. He taught me that that could be reasonably construed as plagiarism.

5. One day, after the 26^th draft, he told me that we were done, and that I should get everything ready for submission. I was elated that we had completed a masterpiece of scientific writing, and I told him so. He informed me that in all likelihood we had not, but that we were simply tired of the paper and it was honorable to call it quits at some arbitrary point.

I had sequenced the N terminus of gp96 from two antigenically distinct tumors, and the sequences were the same. I had generated a polyclonal rabbit serum against the protein in a rabbit called ADAM, and the antiserum detected an identical-sized protein in all tumors and every tissue. In spite of this uniformity, the protein from each tumor elicited immunity only against the tumor from which it was isolated. Gp96 from normal tissues did not elicit tumor immunity against any tumor tested. These findings exhilarated me. I thought that I may have identified a polymorphic gene, perhaps like the MHC or the immunoglobulin family, which was rearranged in tumors and might be responsible for the individually distinct antigenicity of chemically induced mouse sarcomas. After extensive efforts at cloning and sequencing the gp96 cDNA (with the help of Yao-Tseng Chen) (7), and doing intensive Southern blots of genomic DNA from normal tissues and many tumors, there was no hint of polymorphisms or rearrangements. During this analysis, a paper was published that appeared to show a polymorphic protein, consisting of two disulfide-linked subunits, expressed on a number of lymphomas (8). Old had brought this paper to my attention and told me to remember it. He was very admiring of it, and also a bit chagrined that he had not made the observation in his extensive study of mouse leukemias. This protein was easy to identify by a clever method developed by the authors: if one surface-labeled the cells, made extracts in non-reducing buffers, ran a gel, and then ran a second dimension of the lane in a reducing environment, one saw on an autoradiogram a diagonal, from which the two sub-units “fell off” (since they were held together in the non-reducing environment of the first gel but dissociated in the reducing environment of the second dimension). I thought of testing the tumors by this method and, to my delight, found exactly the same pattern! The tumors expressed this highly polymorphic protein! The excitement turned to disappointment when I realized that I saw the pattern only when I used tumor cells isolated from a tumor in vivo, but not when I used tumor cells cultured in vitro. Clearly, the signal I had seen before was derived from the T lymphocytes infiltrating the tumor. I was too embarrassed to tell Old what I had been up to, but he eventually found out, and commended me for trying.

All the more obvious mechanisms to explain the basis of the specificity of immunogenicity of gp96 preparations had by now been exhausted. I was a bit exhausted too, as I began the painful task of trying to test if any contaminating proteins, lipids, carbohydrates (free or on the protein), DNA, or RNA could explain the data. It took over a year to test these possibilities, and I effectively ruled them out. However, I still did not have any satisfactory explanation of the data. I kept myself productively occupied and distracted by doing lateral studies, such as mapping the gene encoding gp96 (9), identifying the human homologue of gp96 (10), and biochemically characterizing (with Dirk Schadendorf) a gp60 antigen recognized by one of the many monoclonal antibodies generated in the lab (11).

Parenthetically, I was by now the resident biochemist of the Old lab. In that capacity, I was occasionally asked to do minor things, as a good citizen. I recall once receiving two samples, labeled “mystery juice #1” and “mystery juice #2” from Betsy Carswell. My job was to analyze them for purity by gel electrophoresis. The samples showed intense bands of 18,000 d with a weaker band about twice the size. I would later learn that these were samples of tumor necrosis factor (TNF) headed for possible microsequencing! TNF was, of course, Old’s seminal discovery and arguably his greatest discovery (see Carswell and Williamson’s reflection in this issue).

At some point during all this, CRI awarded me a fellowship named in honor of Old’s parents. I became the first John Hans and Edna Alice Old Fellow of the CRI. I had known Old’s sister, Constance, for quite some time by then, and through the agency of this new award, I came to know more of her, and of their parents and grandparents. I was always touched by Old’s devotion to his sister and her family, and to the memory of his parents, both of whom had succumbed to cancer.

Old and I fell into the habit of walking out at lunchtime, talking about data and everything else. Walking with him was always a great adventure. He would stop and greet many people, and he would have complete conversations with each of them every time. He would be distracted by any music that he heard on the street, any headlines that he saw on the newspapers on newsstands, the hairstyle of a passerby, the fruits he saw on display in the stores, and just about anything else. Each of these stimuli would elicit a Wikipedia-esque dissertation on the cause of the stimulation. From beginning to end, our walks would therefore take about two hours on the average. During these walks, I learned much about the murine leukemia viruses, the lipopolysaccharides, Schwartzmann reaction, L-asparaginase, the reticuloendothelial system, opsonins, the virtues of mixed hemadsorption, and other such subjects. His lunch menu was quite predictable, and mine only a bit less so. Every day for almost three years, he insisted on paying for my lunch, telling me that I could pay for my own lunch, and his as well, when I became a professor. (Some years later, when we did lunch together, and I had become a professor, he moved the goalpost—I could pay for our lunches when I became a member of the National Academy.) During these walks, I would occasionally give him my unsolicited thoughts on how he was not running the lab as I thought he should, and I would often resist his efforts in trying to rope me into helping some clinical fellow or other with experiments in which I had no interest. He would talk of his excitement of what some of the antibodies could do clinically, and in those days, my eyes glazed at any discussion of such medical possibilities.

Somewhere along the line, I had applied for the First Independent Research Support and Transition (FIRST) award of the National Institutes of Health (NIH), and to my surprise, I was awarded the grant on first attempt. This generated considerable excitement in the lab, and no one was more proud than Old. He remained exuberant and unconditionally happy whenever I discussed my results with him, and this was often. However, our approaches to the problem were beginning to diverge. He advised me to keep looking for genetic polymorphisms in gp96 genes for as long as it would take to find them; I was convinced that more efforts in the same direction would be fruitless. By now, we knew from the sequencing efforts that gp96 belonged to the hsp90 family, and I was well aware that this family of proteins was known for chaperoning other molecules. Of course, my biochemical efforts could not find any other molecules in the gp96 preparations. I had become aware of and thrilled by the then-recent work of Emil Unanue (12) and Alain Townsend (13), demonstrating that the MHC molecules could associate with antigenic peptides. The idea that gp96 may be doing something similar was beginning to cross my mind; however, because of my lack of formal training in immunology, I only vaguely understood the papers of Unanue and Townsend, and this lack of clarity inhibited me from thinking more aggressively along those lines. I discussed these thoughts with Old. He was uncomfortable with the MHC-peptide interaction paradigm at the time and therefore discouraged me from pursuing the possibility that gp96 was binding peptides. Differences in our scientific styles were beginning to emerge and becoming clear to us both. I was unhesitant to hypothesize outside the bounds of my experience; Old was loath to do so. I was keen to begin to explore if immunization with gp96 was eliciting T cell responses, although I knew nothing about how to go about doing so. Old, on the other hand, was intent that I pursue my sequencing studies, until the answer became evident. It seemed to me already evident that the answer was not to be found in the primary structure of gp96. I spent several months in this uncomfortable ambiguity, without any clarity on why the gp96 preparations were immunologically distinct.

External events were beginning to impinge upon the laboratory by this time. Old was increasingly uncertain that he could continue to keep his laboratory funded through the industry-sponsored grants, as he had done for many years. He began reducing his commitments by attrition of personnel, and this took a toll on him. Although Dr. Old was never one to discuss his anxiety openly, I was nonetheless keenly aware of what was happening to him. Realizing that my work was a considerable expense for him, I told him that I would start looking for a faculty position, preferably within the city. He urged me to stay with him, but I felt that I could be of more help by leaving and remaining close by than by remaining part of his laboratory. Fortunately, a position soon became available at the Mount Sinai School of Medicine, and I moved there as an assistant professor in the fall of 1988. Bob Maki was then doing his Ph.D. work (as part of an M.D./Ph.D. program) with me and was cloning the human gp96 genes and transcripts. He moved to Mount Sinai with me, and in 1990, Old and I (with Bob Maki as first author) published our last paper together (14).

While at Mount Sinai, I discovered that not only gp96 but also other chaperones of the hsp70 and hsp90 families elicited tumor-specific immunity, dependent upon the source from where they were purified (15–17). Finally, I fully developed the hypothesis that the immunogenicity of these chaperone preparations was derived from chaperoned peptides, and that such chaperone-peptide complexes were essential for direct presentation and cross-priming (18). The role of chaperone-peptide complexes in direct presentation is now firmly established (19–21), and although evidence for their essential role in cross-priming exists (22–24), evidence to the contrary exists as well (25). [See (26) for a review.] Old and I discussed my work during our weekend conversations, and he remained deeply engaged with my findings. He was unfailingly curious, rigorously critical, and unconditionally supportive. By this time, he had assumed the directorship of the Ludwig Institute for Cancer Research and had brought his own research into full swing again. During the early 1990s, he invited me several times to join the Ludwig Institute; the idea had its obvious attractions, including the fact that I had fond memories of interacting with Old on a daily basis. I decided against it, mainly because I was concerned that the move would, in ways big and small, necessitate a change in the direction of my scientific work.

At some point during my academic journey, I co-founded (with Garo Armen) a biotechnology company, Antigenics (currently Agenus Inc.), in order to explore the clinical activity of the tumor-derived chaperone-peptide complexes. Through this effort, a series of phase I, II, and III trials was conducted, and autologous tumor-derived gp96-peptide complexes became the first therapeutic cancer vaccine to be approved anywhere in the world (27). (They were approved in Russia, where they are in regular clinical use; the U.S. approval requires a positive confirmatory trial, which is yet to be done.) Old followed my efforts in the clinical arena with intense interest. He would often remind me playfully of the time when any discussion of matters of clinical interest would send me dreaming, or worse. He deeply celebrated my successes, and he comforted me in my disappointments.

Within the bounds of our closeness, Old and I disagreed more than we agreed. An interesting area of our disagreement was the importance of the individually distinct antigens of tumors from a therapeutic perspective. I have long believed that the mutated unique antigens of tumors are going to be far more potent than the common shared tumor antigens in mediating clinically significant anti-tumor immune responses (28, 29). Our disagreement on this subject was ironic, since my belief in favor of the unique antigens was grounded in Old’s own considerable writings on the subject [see (2)]. I would remind him that he had strayed from those beliefs and the data on which they were based. He did not disagree, but pointed out to me what an interesting and immunogenic antigen NY-ESO-1 was. I did not disagree. Our differences on the question of unique versus shared antigens narrowed with time, and he became increasingly convinced of the importance of unique antigens. We had many other differences of opinion, scientific and otherwise, but they were accommodated, and even appreciated, within our ever-deepening friendship. Our conversations, even on the areas of our differences, always left me nourished. I have reason to believe that he felt the same.

Old left me a sizable inheritance of sizzling curiosity and enthusiasm for knowledge, love of the method of scientific inquiry, and an unconditional generosity toward mentees. As a friend, I treasure the memory of his natural elegance, and of his profound gentleness and humanity.