Functional Genomics and Medicine


What I Talk about When I Talk about the Discovery of PD-1 (Assoc.Prof. Ishida)


I was born on August 8, 1961 in the city of Nagoya, Aichi Prefecture, Japan.

In 1982, when I was an undergrad student of Nagoya University School of Medicine, I became interested in the phenomenon of graft rejection after organ transplantation. I wondered why the same human tissues are rejected by our immune system. I strongly wished to understand the molecular mechanisms with which our T lymphocytes (T cells, the conductor of the orchestra of immunity) can distinguish 'self' from 'nonself'. Even now, more than 30 years later, the 'self-nonself' discrimination remains to be at the top of my list of great scientific questions.

In the summer of 1983, I initiated my bench work at Aichi Cancer Center Research Institute,raising rat monoclonal antibodies against mouse teratocarcinoma cell lines under the supervision of Drs. Noriyuki Hamajima, Ryuzo Ueda, and Toshitada Takahashi. Ryuzo and Toshitada are hematologists/tumor immunologists who had been trained as postdocs in Dr. Lloyd J. Old's laboratory at Memorial Sloan Kettering Cancer Center in New York City. From my residence at that time (located in the eastern suburb of the city of Nagoya), Aichi Cancer Center was much closer (~10 min drive) than Nagoya University Hospital (~1.5 hour train and bus rides), and I also preferred being engaged in immunology experiments in the laboratory to attending (boring) lectures at the medical school.

On March 8, 1984, I read, in the Aichi Cancer Center library, three Nature Articles published by the groups of Drs. Tak W. Mak (University of Toronto) and Mark M. Davis (Stanford). There, they reported the discovery of the T-cell antigen receptor genes by using the extremely elegant subtractive hybridization technique in molecular biology. I was deeply impressed with their great achievements and strongly wished to perform similar subtractive hybridization experiments in my own career in the future.

In March 1986, I graduated from Nagoya University School of Medicine.

In May 1986, I passed the national examination for medical doctors (I became an M.D.).

In June 1986, I started to be trained as a resident of internal medicine at the Aichi Cancer Center hospital. I learned how difficult it was at that time to treat leukemia/lymphoma or lung cancer patients by chemotherapy.

In April 1987, I became a graduate student in Tasuku Honjo's laboratory at Kyoto University Faculty of Medicine. At that time, only a handful of bioscience researches in Japan (including Tasuku Honjo, Tadatsugu Taniguchi, Shigetada Nakanishi, and Shosaku Numa) were able to perform difficult & complex molecular biology experiments.

When I joined Tasuku’s lab in 1987, I was seriously thinking about two things.
(1) I wished to elucidate the molecular mechanisms involved in ‘self-nonself’ discrimination.
(2) I also wished to perform the subtractive hybridization experiments.

(1) was an extremely big goal, and I did not know how to initiate my research for it because the picture was too huge, vague, and ambiguous.
(2) was a very clear thing, but I did not know the best application target of subtractive hybridization.

Although I wished to start subtractive hybridization experiments immediately, the initial project that Tasuku gave me was to analyze the immunological abnormalities found in human interleukin-2 (IL-2)/IL-2 receptor α-chain transgenic mice.

Developing immature T cells are strictly 'educated' in the thymus ('T' of T cells came from thymus) and learn what is the self, and what is not the self. Ellis L. Reinherz (Dana-Farber Cancer Institute, Harvard Medical School) just published an interesting paper showing that presence-or-absence of the IL-2 signal in the thymus determines the fates of developing T cells (Ramarli, D. et al. PNAS 84, 8598-8602, 1987). After reading this paper, I persuaded myself that, although the transgenic-mouse experiments were not associated with the sophisticated subtractive-hybridization technique, I might be able to find an important clue for the question of 'self-nonself' discrimination if I work hard, analyzing the curious phenotypes of the human IL-2/IL-2R transgenic mice.

I was very lucky to be able to publish a couple of papers from the transgenic-mouse works (Nishi, M, Ishida, Y., and Honjo, T. Nature 331, 267-269, 1988; Ishida, Y. et al. Int. Immunol. 1, 113-120, 1989; Ishida, Y. et al. J. Exp. Med. 170, 1103-1115, 1989). Since these papers were more than enough for my Ph.D. thesis, Tasuku gave me a certain degree of freedom for the rest of my graduate-study period in his lab.

At that time, bioscience researchers were beginning to realize an attractive and novel notion of apoptosis and/or programmed cell death. In 1989, a UK team published a Nature paper (Smith, C. A. et al. Nature 337, 181-184, 1989), showing that, when strongly stimulated through their T-cell antigen receptors (TCRs), well-educated developing immature T cells in the thymus immediately realize that they are harmfully reacting against 'self' components and commit suicide by undergoing apoptosis/programmed death. Other immunologists also showed that this type of apoptosis/programmed death of immature T cells depends upon de novo gene expression.

When I read this interesting Nature paper in 1989, I instantly realized that I might be able to achieve both of my two main targets simultaneously: if I am able to isolate the genes strongly associated with the apoptosis/programmed death of self-reactive (stimulated) immature T cells by using the subtractive hybridization technique, they must become the very important clues for me to initiate my investigation on the molecular mechanisms involved in ‘self-nonself’ discrimination.

In other words, the Nature paper inspired me and shook me at the unprecedented level because the data there strongly suggested that, if I isolate mRNAs whose transcription is induced immediately after immature T cells get stimulated and decide to commit suicide, I might be able to elucidate (even a part of) the long-standing question about the 'self-nonself' discrimination.

In March 1991, I obtained my Ph.D. from Kyoto University.

In April 1991, I became a special postdoc of JSPS (Japan Society of Promotion of Science) and continued to stay in Tasuku Honjo's lab.

In July 1991, I initiated my subtractive hybridization experiments (see below for details).

On September 1, 1991, I became an assistant professor of Kyoto University Faculty of Medicine and continued my subtractive hybridization experiments in Tasuku Honjo's lab. (see below for details).

By the end of September 1991, I was able to isolate four independent cDNA clones that I initially believed to have been derived from four different candidate genes. Very interestingly, however, the reality was that all of the four cDNA clones were originated from the same single gene (!), that I later named programmed death-1 (PD-1) in the middle of October 1991, hoping that the novel gene could have something to do with apoptosis/programmed death of immature T cells.

As I said, I was able to isolate four candidate cDNA clones in my subtraction experiments. Usually, it is difficult for researchers to reduce the number of candidate genes only by subtraction. People usually combine the subtraction screening with some other experiments in order to be focused on a realistic number of candidate genes. In my case, however, only four candidate clones remained after the large-scale subtraction screening. More surprisingly, the four cDNA clones turned out to be derived from the same single gene! (that I later named PD-1)

In 1992, I published our first PD-1 paper (Ishida, Y. et al. EMBO J. 11, 3887-3895, 1992).

Discovery of PD-1

(this is the detailed record of my subtractive hybridization experiments in 1991. Please skip this part if you are not a molecular biologist)

I performed the subtractive hybridization experiments shown below based on the assumption that identification of the novel genes strongly associated with programmed death of immature T lymphocytes would give us some important clues for the understanding of molecular mechanisms involved in the 'self-nonself' discrimination by the immune system.

  1. Construction of the subtracted cDNA library
    • First strand cDNA synthesis using mRNAs extracted from 2B4.11 (mouse T-cell hybridoma) cells stimulated with ionomycin and PMA
      July 3-4, 1991
    • Subtraction [(stimulated 2B4.11) – (IL3-supplemented LyD9)] using hydroxyapatite column chromatography (LyD9: mouse hematopoietic progenitor cell line)
      July 4-5, 1991
    • Second-strand cDNA synthesis for the subtracted cDNAs
      July 5-8, 1991
    • Construction of the cDNA library using the subtracted cDNAs and the λgt10 arms
      July 8-14, 1991
    • Evaluation of the quality, titer, and specificity of the constructed cDNA library
      July 14-August 1, 1991
    • Construction of the subtracted cDNA library completed
      August 1, 1991
  2. Subtracted cDNA probe for the first round of library screening
    • First strand cDNA synthesis using mRNAs extracted from IL3-depleted LyD9 cells
      August 13-15, 1991
    • Subtraction #1 [(IL3-depleted LyD9) – (IL3-supplemented LyD9)] using hydroxyapatite column chromatography
      August 15-16, 1991
    • Subtraction #2 [(IL3-depleted LyD9) – (IL3-supplemented LyD9)] using hydroxyapatite column chromatography
      August 17-18, 1991
    • Two successive subtractions were carried out in order to increase the specificity of the subtracted cDNA probe.
    • Superhot labeling of the subtracted cDNAs by random priming
      August 19, 1991
    • Preparation of the subtracted cDNA probe completed
      August 19, 1991
  3. First round of screening of the subtracted cDNA library
    • Filter hybridization, washing, and autoradiography
      August 19-21, 1991
    • 120 candidate phage clones picked up
      August 21, 1991
    • Candidate phage clones re-plated for the second round of screening
      August 24-26, 1991
  4. Subtracted cDNA probe for the second round of library screening
    • First strand cDNA synthesis using mRNAs extracted from IL3-depleted LyD9 cells
      August 28, 1991
    • Subtraction #1 [(IL3-depleted LyD9) – (IL3-supplemented LyD9)] using hydroxyapatite column chromatography
      August 29-30, 1991
    • Subtraction #2 [(IL3-depleted LyD9) – (IL3-supplemented LyD9)] using hydroxyapatite column chromatography
      August 31-September 1, 1991
    • Two successive subtractions were carried out in order to increase the specificity of the subtracted cDNA probe.
    • Superhot labeling of the subtracted cDNAs by random priming
      September 2, 1991
    • Preparation of the subtracted cDNA probe completed
      September 2, 1991
  5. Second round of screening of the subtracted cDNA library (120 candidate clones in λgt10)
    • Filter hybridization, washing, and autoradiography
      September 2-6, 1991
    • Four candidate λgt10 clones picked up
      September 8, 1991
  6. Characterization of the identified cDNA clones
    • Four candidate cDNA inserts in the λgt10 phage subcloned into the BSSK plasmid vector
      September 20, 1991
    • Induced mRNA expression in stimulated 2B4.11 cells and IL3-deprived LyD9 cells confirmed for all of the four candidate cDNAs (identical induction patterns and band sizes of mRNAs observed in the Northern-blotting experiment for all of the candidate cDNAs)
      September 24, 1991
    • Restriction mapping patterns of the four cDNA inserts completely overlapped
      September 26, 1991
    • Nucleotide sequences of the cDNA inserts determined and shown to be identical (completely overlapped)
      October 13, 1991
    • The discovered gene named programmed death-1 (PD-1)
      October 13, 1991

In October 1993, I, with my wife and daughter (3 y.o. at that time), moved to the greater Boston area to conduct my postdoctoral research in Philip Leder's laboratory at Harvard Medical School. Phil, formerly working with Marshall W. Nirenberg at NIH, USA, is very famous for his enormous contribution to the cracking of the genetic code in the late 1960s. Phil had been the chairman of Department of Genetics, Harvard Medical School, for more than 20 years (between 1983 and 2006, although I might not be very accurate in the year counting).

When I moved from Kyoto to Boston in 1993, I left all of my PD-1-related works for my successors (talented graduate students) in Tasuku's lab, and I initiated a new gene-trapping project in Phil's lab. Since then, I had devoted myself into the mouse-genomics research, that had nothing to do with PD-1, until 2015.

After my departure, very smart graduate students in Tasuku's lab at Kyoto University elucidated the basic function of PD-1: the PD-1 trans-membrane protein, transiently expressed on the cell surface of activated T cells, is a receptor that receives the negative regulatory signals from the surrounding cells and sends such signals into the cytoplasm of activated T cells (Agata, Y. et al. Int. Immunol. 8, 765-772, 1996; Nishimura, H. et al. Immunity 11, 141-151, 1999; Nishimura, H. et al. Science 291, 319-322, 2001; Okazaki, T. et al. PNAS 98, 13866-13871, 2001).

In other words, while I was absent from Tasuku's lab, my successors (graduate students in Tasuku's lab) showed that:
(1) From the functional point of view, PD-1 is not directly involved in the induction of apoptosis/programmed death of immature T cells.
(2) Instead, PD-1 is a negative regulator of immune responses of T cells.

Because of this, some people (including Tasuku) say that PD-1 was discovered coincidentally. I agree at least partially because PD-1 was not an obvious cell death-inducing gene that I was initially looking for.

But, please remember what my final goal was. My ultimate goal was to elucidate the molecular mechanisms involved in the 'self-nonself' discrimination by our immune system. Nowadays, no one believes that PD-1 is an apoptosis-inducing gene. At the same time, however, no one doubts that PD-1 plays some kinds of pivotal roles in 'self-nonself' discrimination by T cells.

Some of the graduate students in Tasuku's lab also discovered the potential usefulness of the blockade of the PD-1 signaling pathway in cancer treatment (Iwai, Y. et al. PNAS 99, 12293-12297, 2002). An ally of a group in Ono pharmaceutical company (Osaka, Japan) led by Shiro Shibayama, who also had a training in Tasuku's lab in the early 1990s, and the other in Medarex/Bristol-Myers Squibb (USA) had succeeded in the development of a completely humanized anti-human PD-1 monoclonal blocking antibody of the IgG4 subclass (the product name: Opdivo; the general name: nivolumab).

Recent Developments

Recently, as you might know, the immunotherapies using anti-CTLA-4 and/or anti-PD-1 monoclonal blocking antibodies have revolutionized the field of human cancer treatment (please find hundreds of free review articles on the web using the keywords: cancer, immunotherapy, CTLA-4, PD-1, etc.). For instance, Opdivo has been approved by FDA of the United States as an effective drug for the treatment of malignant melanoma, non-small cell lung cancer (NSCLC), and renal cell carcinoma (RCC), and Hodgkin lymphoma (as of June 2016).

Consequently, James P. Allison (Texas, USA) and Tasuku Honjo have been frequently honored with distinguished prizes and awards, including the Tang Prize (Taiwan), the William B. Coley Award (USA), and the Lasker Award (USA), for their significant contribution to cancer immunotherapy.

As I talked earlier, I started my career as a clinical doctor of internal medicine at the Aichi Cancer Center hospital. There, I learned how difficult it was at that time (in 1986) to treat leukemia/lymphoma or lung-cancer patients with standard chemotherapy. Therefore, I feel truly glad and honored that the PD-1 research that I initiated in 1991 has positively transformed the field of cancer therapy, saving the lives of a lot of cancer patients, although I was not expecting at all such wonderful development when I discovered PD-1.

In the summer of 2015, my research group at NAIST launched the PD-1-related basic investigation because I believe that people in the world still really do not understand the 'real' physiological functions of PD-1 that could better explain the reason(s) why the PD-1 blockade by antibodies is so effective in cancer immunotherapy.

Both CTLA-4 and PD-1 are the negative regulators of immune responses, belonging to the same CD28 family. Also, both CTLA-4 and PD-1 are currently being used as the targets of antibody-mediated functional blockade in human cancer immunotherapy.

Historically speaking, James P. Allison first showed that the CTLA-4 blockade is effective in the murine model of cancer immunotherapy (Science 271, 1734-1736, 1996). Tasuku Honjo and Nagahiro Minato were able to prove the similar (but significantly different) thing for PD-1 six years later (PNAS 99, 12293-12297, 2002). This is why many people think that Jim Allison was the person who first opened the door of current cancer immunotherapy.

The reality of cancer immunotherapy at present in clinics all over the world, however, is a slightly different story. Side effects of cancer immunotherapy using the CTLA-4 antibody are much more serious and grave than those using the PD-1 antibody. Although the FDA approval was earlier for CTLA-4 than for PD-1, the majority of clinicians right now believe that the CTLA-4 blockade would not be able to survive as a mono-therapy: it would survive only in combination with some other things including the PD-1 blockade. On the other hand, the list of cancer types that can be treated through the PD-1 blockade is still constantly expanding. Basically, the PD-1 blockade is much safer and more effective than the CTLA-4 blockade.

Do you know why the effects of cancer therapies using the CTLA-4 and PD-1 blocking antibodies are so much different? Here, you have to think about the physiological functions of CTLA-4 and PD-1.

First of all, let us compare the phenotypes of the knockout mice for these two molecules.

All of the CTLA-4 knockout mice die of the extremely severe autoimmune diseases within three weeks after birth (Immunity 3, 541-547, 1995; Science 270, 985-988, 1995). In contrast, the PD-1 knockout mice (especially on the C57BL/6 genetic background) are perfectly healthy for about one year after birth (Immunity 11, 141-151, 1999). Sporadic and mild autoimmune diseases gradually begin to arise in the aged PD-1 knockouts, only about one year after birth.

This is because, I believe, CTLA-4 is negatively regulating virtually all immune reactions while PD-1 is suppressing only the limited types of immune responses like those against cancer cells. In the CTLA-4 knockout mice, all immune reactions become unleashed and uncontrolled while, in the PD-1 knockout mice, only the limited types of immune responses are gradually re-activated.

After blocking the functions of CTLA-4 or PD-1 by using antibodies, the same patterns of things happen in cancer patients: serious and grave side effects with the CTLA-4 antibody; almost nothing with the PD-1 antibody.

A recent finding also suggest that ipilimumab (the FDA-approved anti-human CTLA-4 monoclonal antibody of the IgG1 subclass) might not be acting simply as the checkpoint blocker. Instead, the antibody could also be unleashing a broad spectrum of immune reactions by killing regulatory T cells (Tregs), which are well-known to constitutively express the CTLA-4 molecules on their cell surface, through its antibody-dependent cellular cytotoxicity (ADCC) activity (PNAS 112, 6140-6145, 2015). In general, the ADCC activity of IgG1 (e.g., ipilimumab) is relatively high while that of IgG4 (e.g., nivolumab/Opdivo) is negligible.

In my opinion, Jim Allison was really great because he first opened the door of current cancer immunotherapy. However, I am not certain if he has created the best (or most appropriate) strategy for this purpose.

Remaining Questions about the 'Real' Physiological Functions of PD-1

As I stated above, the PD-1 knockout mice (especially on the C57BL/6 genetic background) are perfectly healthy, at least for about one year after birth (Immunity 11, 141-151, 1999). Then, what is PD-1 doing in young mice? Also, why does the mild and sporadic autoimmunity arise only in aged mice?

Upon blockade of the PD-1's function(s) in our body by using the inhibitory antibodies, the immune system begins to attack cancer cells. Then, what is PD-1 protecting? In the normal condition (i.e., without injection of the inhibitory antibodies), PD-1 appears to be protecting cancer cells from the attack by the immune system. Is PD-1 on our side? or on the side of cancer cells?

Why do almost all of the 'bad guys' [cancer cells, chronically virus-infected cells (Nature 439, 682-687, 2006; Nature 443, 350-354, 2006; J. Virol. 80, 11398-11403, 2006), and the other pathogens that cause chronic infections in humans (Nat. Immunol. 13, 188-195, 2012; PNAS 108, 9196-9201, 2011; PLoS Pathog. 5, e1000431, 2009)] target PD-1 for their long-term survival in our body?

These are the remaining unanswered questions.

I have a hypothesis about the 'real' physiological function(s) of PD-1, with which I believe I would be able to answer most of the above questions quite well. With the final proof of my hypothesis, people would be more firmly convinced that PD-1 plays the extremely important role(s) in 'self-nonself' discrimination. As the discoverer of PD-1, I think I am obliged to carry out this project because elucidation of the molecular mechanisms involved in 'self-nonself' discrimination by the immune system was the final goal for me from the very beginning of my career as a scientist.
(June 28, 2016)

page top