August 1978–May 1979

Prompted by the Erikson group’s success in using anti-RSV tumor sera, we set out to raise our own anti-RSV tumor sera by inoculating baby rabbits with a mammotropic strain of RSV, obtaining our first active tumor serum in November 1977. We had used this serum to show that an IP of in vitro-translated v-Src had protein kinase activity that phosphorylated the Ig heavy chain and that the kinase activity of in vitro-translated ts mutant v-Src was greatly reduced, providing strong support for the conclusion that this activity is intrinsic to v-Src, as we reported in a paper submitted in November 1978 (5). It was our experience with the RSV IP kinase assay that allowed Mary Anne Hutchinson, Walter Eckhart, and me to quickly set up and test whether IPs made with an anti-Py tumor serum contained a T antigen-associated protein kinase activity. The first experiment done in August 1978 showed that a band the size of mT itself was phosphorylated in vitro. In subsequent experiments, we found that this band was missing when a nontransforming Py mutant was used, providing the first clue that the mT-associated kinase activity might be important for Py transformation. Subsequently, using strains of Py with different sized mT antigens, we showed that the phosphorylated band was indeed mT itself, and not the Ig heavy chain, which is about the same size. All three groups working on Py mT obtained similar evidence for an mT-associated protein kinase activity and met up at the 1979 Cold Spring Harbor (CSH) Symposium on Tumor Viruses, held May 31–June 5, 1979. Each group gave a talk at the meeting and reported that the mT-associated kinase activity had properties consistent with its being involved in Py transformation. After discussions at the bar, we agreed to cosubmit papers to Cell describing these findings as soon as we returned from the meeting at Cold Spring Harbor Laboratories. The papers were submitted in June 1979. Even before we had submitted our paper, however, I knew that one likely reviewer question would be the following: What amino acid was being phosphorylated in mT and was this activity a Thr-specific kinase like v-Src?

June 1979—The First Sighting of Phosphotyrosine

To preempt this question, the day after our paper was submitted on June 11, 1979, I set about analyzing whether mT was phosphorylated on Thr or Ser, which I did by carrying out partial acid hydrolysis of a ³²P-labeled mT isolated from a band identified by gel separation of an IP kinase assay (see Fig. S1 for the page from my lab notebook). On June 14, 1979, I analyzed the hydrolysate using cellulose thin layer electrophoresis (TLE) at pH 1.9. As told elsewhere, the autoradiogram revealed a ³²P-labeled compound running between the pSer and pThr markers located by ninhydrin staining (Fig. 1A). When I repeated this analysis 10 d later and obtained the same result, it was clear that this observation was not an artifact and needed an explanation. Luckily, my biochemical training at the University of Cambridge had taught me that there was a third hydroxyamino acid, namely tyrosine (Tyr), and I guessed that this novel phosphorylated compound might be phosphotyrosine (pTyr). But I needed some marker pTyr to formally establish this was the case. By mixing POCl₃ with Tyr in water, I succeeded in extracting a small amount of authentic pTyr from the ensuing black tar (!), and, when tested using our flat-bed TLE set up on July 2, this marker pTyr proved to migrate between pSer and pThr. As is widely known, it turned out that the reason that pTyr had resolved from pThr was that I had been too lazy to make up fresh pH 1.9 buffer, which we routinely reused. Upon repeated reuse the pH of the buffer dropped from 1.9 to ∼1.7, and this small pH difference was responsible for the resolution of pTyr and pThr, which comigrate at pH 1.9. Indeed, reviewer 1 of our paper did ask what amino acid was phosphorylated, and, in the revised paper [submitted to Cell on September 21, received on September 25, and accepted on September 27 (!)], we included our evidence that the mT-associated kinase phosphorylates Tyr as a new figure (figure 7 in ref. 6). Our paper was published in Cell in December 1979, back to back with papers from the other two groups, who did not include any data, however, on the amino acid target specificity of the mT-associated kinase (6⇓–8).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

First experiments demonstrating polyoma middle T and v-Src tyrosine kinase activity. (A) Autoradiogram (4 d with screen) of the products of a partial acid hydrolysate (2 h at 110 °C in 6 M HCl) of the ³²P-labeled polyoma (Py) middle T antigen band separated by cellulose thin layer electrophoresis toward the anode at pH 1.9 (90 min at 1 kV). The mT band was isolated from an SDS/PAGE gel of the products of an in vitro kinase assay carried out with an immunoprecipitate (IP) made with Py antitumor serum from polyomavirus (Py)-infected mouse 3T6 cells (6). The origin, the cathode and anode, the positions of the pSer and pThr markers (circled) visualized by ninhydrin staining, and free orthophosphate (Pi) are indicated. “X” marks the ³²P-labeled spot, later shown to be pTyr. (B) Autoradiogram (4 d with screen) of ³²P-labeled acid/pronase hydrolysates separated by thin layer electrophoresis at pH 1.9 as in A. From the left: acid hydrolysate of a Py large T antigen (LT) band from ³²P-labeled infected cells; acid hydrolysates of Py LT and middle T antigen (MT) bands from a gel of an in vitro kinase assay carried out on an antitumor serum IP; pronase proteolysate of a Py MT band; acid hydrolysate of an Ig heavy chain band from an in vitro kinase assay carried out on a v-Src immunoprecipitate made with RSV antitumor serum. The origin, the cathode and anode, and the positions of the marker pSer, pThr, pTyr, and free orthophosphate (Pi) are indicated.

• Download figure
• Open in new tab
• Download powerpoint

Fig. S1.

The June 12, 1979 page from Tony Hunter’s experimental notebook, which describes the first experiment that led to the discovery of phosphotyrosine and highlights the surprise at finding that the radioactive spot did not comigrate with either the pSer or pThr marker.

At this point, with a major discovery tantalizingly close, on July 3, 1979, I set off with a group of friends from the Institute for the two-day drive to Idaho with all our rafting gear for a 7-d self-guided rafting trip on the Middle Fork of the Salmon River, which began on July 6. After getting off the river on July 14, I flew directly from Salmon, Idaho by chartered plane to Salt Lake City, and then on to England for the DNA Tumor Virus meeting being held in Cambridge, July 16–20. I talked at the meeting about the mT-associated kinase but did not say anything about the novel tyrosine kinase activity. After the Cambridge meeting, I visited my sister in London and my parents in Kent and then spent a few days giving talks in Hamburg and Freiburg, finally getting back to La Jolla on August 8. In retrospect, it is hard to believe that I did not rush back to the laboratory sooner!

August 1979—Polyoma mT Antigen Is Phosphorylated on Tyrosine in Vitro

To prove unequivocally that mT was being phosphorylated on Tyr, a number of additional experiments were needed. First, we developed a better method for synthesizing pTyr so that we had enough marker for subsequent analyses and also set up a 2D separation on cellulose thin-layer plates, using electrophoresis at pH 1.9 in the first dimension followed by chromatography in a foul-smelling mixture of isobutyric acid and 0.5 M NH₄OH (5:3) to resolve pSer, pThr, and pTyr better. On August 28, I analyzed the acid hydrolysis product of ³²P-labeled mT in two dimensions, and the ³²P compound comigrated precisely with pTyr (6). At this point, we were convinced that authentic pTyr was being generated in this in vitro assay, but, because we were not protein kinase experts, we thought we might have missed a prior report of protein–tyrosine phosphorylation, and so I decided to ask a few experts to see whether anyone had ever found a tyrosine-specific kinase activity. I called Ed Krebs at the University of Washington in Seattle, the acknowledged leader in the protein kinase field, Joli Traugh at the University of California, Riverside, and Gordon Gill at the University of California, San Diego (two other protein kinase mavens), and also George Taborsky at the University of California, Santa Barbara, who had written a classic review on phosphoproteins. None of them had heard of a tyrosine kinase although I did find out that free phosphotyrosine is used as a tyrosine storage compound in insect eggs, due to its very high solubility. At this stage, it looked like we had discovered a novel type of protein kinase.

September 1979—v-Src Is a Tyrosine Kinase

There was still a nagging concern that the appearance of pTyr in mT antigen was somehow an artifact of using acid hydrolysis. To rule this possibility out, I decided to carry out total pronase proteolysis of ³²P-labeled mT to generate free (phospho)amino acids in another way. Together with the mT proteolysate, I decided to spot on the same TLE plate an acid hydrolysate of ³²P-labeled heavy chain phosphorylated by v-Src to serve as an authentic pThr control. This plate was run at pH 1.9 on September 18, and, much to my amazement, the spot from the heavy chain comigrated with pTyr and not pThr (Fig. 1B). Separation of the same heavy chain hydrolysate in two dimensions on September 19 showed that it contained only ³²P-labeled pTyr. On September 23, Bart Sefton and I quickly repeated the experiment and obtained the same result, and carried out a total pronase digest of the ³²P-labeled heavy chain to confirm the presence of pTyr. This unexpected finding meant that Marc Collett and Ray Erikson had unluckily been misled into concluding that v-Src is a Thr kinase because pThr and pTyr comigrate at pH 1.9 on Whatman 3MM paper, which they had used for electrophoresis. On October 3, I analyzed Ig heavy chain that had been phosphorylated in an IP by the c-Src protein, the cellular homolog of v-Src, and found that it too phosphorylated the heavy chain on Tyr—the first cellular tyrosine kinase [figure 2 in Hunter and Sefton (9), reproduced in Fig. 3]. It is interesting to note that Sara Courtneidge showed subsequently that the tyrosine kinase activity associated with Py mT antigen is not an intrinsic activity, but rather due to tight association with the c-Src protein (10), providing the first example of a tumor virus-transforming protein commandeering and activating a cellular signaling protein.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

v-Src phosphorylates tyrosine in the Ig heavy chain. This figure shows an autoradiogram of a 2D separation of a partial acid hydrolysate of the Ig heavy chain band from an in vitro kinase assay. First (horizontal) dimension was pH 1.9 thin layer electrophoresis with the origin on right (small circle) and the anode on left; second (vertical) dimension was chromatography in isobutyric acid/ammonia. Positions of the pSer, pThr, and pTyr markers detected by ninhydrin staining are shown. Reproduced from ref. 9.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

c-Src has tyrosine kinase activity. This figure shows an autoradiogram of ³²P-labeled partial acid hydrolysates of the Ig heavy chain bands isolated from immune complex kinase assays, carried out with anti-RSV tumor serum IPs from the indicated sources, separated by thin layer electrophoresis at pH 1.9 with the anode at the top. From the left: v-Src isolated from infected chick cells (lane A); v-Src synthesized in vitro in an mRNA-dependent rabbit reticulocyte cell free system (lane B); endogenous c-Src isolated from human HBL-100 mammary epithelial cells (lane C); and endogenous c-Src isolated from chicken embryo cells (lane D). The origin (horizontal arrows on left and right), and the positions of the marker pSer, pThr, pTyr, and free orthophosphate (Pi) are indicated. Reproduced from ref. 9.

Although the in vitro data convinced us that both Py mT antigen and RSV v-Src had associated tyrosine kinase activity, they did not tell us whether mT and v-Src also acted as tyrosine kinases in the cell. To begin to test this possibility, Bart Sefton and I analyzed ³²P-labeled v-Src and the associated 80-kDa (later shown to be the Hsp90 chaperone) and 50-kDa (later shown to be the Cdc37 cochaperone) proteins, which coprecipitated with v-Src from ³²P-labeled RSV-infected chicken embryo fibroblasts [figure 3 in Hunter and Sefton (9)]. Two-dimensional separation of the partial acid hydrolysates of the ³²P-labeled v-Src, p80, and p50 samples, carried out on October 1, showed that p80/Hsp90 contained exclusively pSer whereas v-Src contained equal amounts of pSer and pTyr [figure 4 D–F in Hunter and Sefton (9)]. p50/Cdc37 also contained pSer and pTyr, becoming the first cellular protein identified as a target for a tyrosine kinase. Two-dimensional analysis of tryptic digests of the same ³²P-labeled v-Src band [figure 4A in Hunter and Sefton (9)] revealed that v-Src had two main phosphopeptides: Partial acid hydrolysis of the isolated peptides revealed that the peptide derived from the C-terminal half of v-Src contained pTyr whereas the peptide from the N-terminal half contained pSer (Fig. 4; this figure was in the original version of the paper submitted to PNAS but was omitted from the revised version to save space). These results convinced us that pTyr was a bona fide and stable in vivo posttranslational modification and suggested that v-Src might autophosphorylate and also phosphorylate the cellular p50/Cdc37 protein.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

v-Src contains pSer and pTyr in different tryptic peptides. Tryptic peptides were eluted from a 2D cellulose thin layer separation of a tryptic digest of v-Src obtained from ³²P-labeled chicken embryo cells [shown in figure 4A in Hunter and Sefton (9)]. The eluted α (ALPHA) and β (BETA) peptides were subjected to partial acid hydrolysis, and the hydrolysates were resolved in two dimensions exactly as described in Fig. 2. Autoradiograms of the TLE plates are shown. (A) α peptide. (B) β peptide. Subsequent work showed that the α peptide is derived from the N-terminal half of v-Src and contains pSer17 and that the β peptide is derived from the C-terminal half of v-Src, which comprises the catalytic domain, and contains pTyr416, the major autophosphorylation site. Positions of the pSer, pThr, and pTyr markers detected by ninhydrin staining are shown, and the origins are indicated by small circles at the bottom right.

October 1979—v-Src Phosphorylates Tyrosine in Cellular Proteins

If RSV transforms cells by expressing an active tyrosine kinase, then one might expect to find increased levels of pTyr-containing proteins in RSV-infected cells. To test whether v-Src tyrosine kinase activity causes increased tyrosine phosphorylation, I took a ³²P-labeled RSV-infected chick cell lysate prepared by Bart Sefton and extracted the total protein into phenol. Protein was TCA-precipitated from the phenol layer, and ³²P-labeled lipids were then removed from the protein pellet by chloroform:methanol (2:1) extraction, which was then subjected to partial acid hydrolysis. This rather unconventional extraction seemed to work and got rid of the ³²P-labeled RNA and phospholipid contaminants. Separation of this acid hydrolysate on October 2 by TLE at pH 1.9 followed by chromatography showed a trace of pTyr. However, the resolution of pTyr from the massive pThr spot was marginal and was not going to persuade the uninitiated that this minor spot was not a pThr satellite spot. Nevertheless, when we repeated this experiment (on October 10) and compared ³²P-labeled samples from uninfected and RSV-infected chick cells, the pTyr spot was visible only in the RSV-infected sample, suggesting that RSV infection indeed stimulated tyrosine phosphorylation of cellular proteins.

To try to obtain better separation of pThr and pTyr, I tested whether TLE at pH 3.5 would work as a second dimension, based on a suggestion from Jack Rose, a fellow faculty member at the Salk, who knew that this method had been used in David Baltimore’s and Eckhard Wimmer’s groups to resolve the pTyr generated upon hydrolysis of the poliovirus protein linked to genomic RNA from pSer and pThr. Electrophoresis at pH 3.5 worked extremely well, giving a large separation of pTyr from pThr, and it allowed us to establish conditions for a 2D separation of pSer: pThr and pTyr, using TLE at pH 1.9, followed by TLE at pH 3.5 at right angles to the first dimension, instead of having to use a nauseating isobutyric acid containing buffer in the second dimension. The new pH 1.9/pH 3.5 2D TLE separation proved essential for resolving the much less abundant pTyr cleanly from pThr (9%) and pSer (90%) in hydrolysates of total ³²P-labeled proteins from RSV-infected chick cells. In an experiment analyzed between October 14 and 17 using this new 2D separation, we found that normal chick cells or chick cells infected by transformation-defective RSV had low levels of pTyr in proteins (0.05%) and that this level was increased up to 10-fold in WT RSV-transformed cells [table 1 and figure 5 in Hunter and Sefton (9), reproduced here as Fig. 5], consistent with v-Src’s acting as a tyrosine kinase in cells to phosphorylate cellular proteins. This result was the final piece of data we needed for the paper. We also found that the high level of pTyr in cellular proteins in chicken embryo fibroblasts infected with a ts v-Src mutant RSV was rapidly reduced to the level in normal cells after shifting the cells from the permissive growth temperature to the restrictive temperature when the cells are no longer transformed. Together, these experiments provided compelling evidence that v-Src itself is the kinase responsible for phosphorylating cellular proteins on tyrosine in RSV-transformed cells and also implied that there had to be cellular protein phosphatases that could remove phosphate from pTyr. Even though we had obtained these temperature shift data before we submitted the paper to PNAS, and one of the reviewers even asked for this experiment, we did not include this result in the final version due to space constraints. The temperature shift data were included in a subsequent paper published in Cell in September 1980 (see figure 1 in ref. 11).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

v-Src increases pTyr levels in whole cell proteins. This figure shows autoradiograms of a 2D separation of partial acid hydrolysates of total protein isolated from normal uninfected (A) and Schmidt-Ruppin A RSV-transformed (B) chicken embryo cells labeled for 16 h with ³²P-orthophosphate. As shown in the schematic of the separation on the right (C), the first (horizontal) dimension was thin layer electrophoresis at pH 1.9 with the origin (circled) on the right and the anode on the left; the second (vertical) dimension was thin layer electrophoresis at pH 3.5, with the anode at the top. The positions of the marker pSer, pThr, and pTyr are indicated. On panels A and B the origin is indicated with a small vertical arrow, and the position of pTyr with a large arrow. Partially hydrolyzed peptides are seen at the bottom right. Reproduced from ref. 9.

November 1979—The Paper Is Submitted to PNAS

With the final piece of evidence establishing that v-Src acts as a tyrosine kinase in vivo, we were able to write up our findings for submission to PNAS. In those days, PNAS papers had to be communicated by a National Academy member, and, luckily, Bob Holley, a Nobel prizewinner who had switched to studying factors that negatively regulate cell growth after his prizewinning research on tRNA structure and move to the Salk Institute, had his laboratory on our floor and agreed to communicate our paper. We gave four copies of our paper to Bob Holley on November 12, and he sent these copies out to three (local) reviewers. Remarkably, all of the reviews were received within two weeks (see Fig. 6 for reviewer 3′s handwritten and positive comment and Figs. S2 and S3 for the other two reviews). We addressed the issues they raised by rewriting and also significantly shortening the manuscript to fit the 5-page limit for PNAS. The final version was received by PNAS on December 3, 1979 (see Figs. S4 and S5), but the paper was not published until the March issue in 1980 (9).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Reviewer 3′s comments. The handwritten review on the PNAS review form is shown.

• Download figure
• Open in new tab
• Download powerpoint

Fig. S2.

Reviewer 1 comments on author’s duplicate of original PNAS form.

• Download figure
• Open in new tab
• Download powerpoint

Fig. S3.

Reviewer 2 comments on author’s duplicate of original PNAS form.

• Download figure
• Open in new tab
• Download powerpoint

Fig. S4.

Manuscript cover page with PNAS date stamp of December 3, 1979.

• Download figure
• Open in new tab
• Download powerpoint

Fig. S5.

Cover letter to PNAS accompanying revised manuscript.

December 1979—First Public Presentations

After talking about our results at the TVL floor meeting at the beginning of October (with the current Editor-in-Chief of PNAS, Inder Verma, at the time an Assistant Professor on our floor, in attendance), my first public presentation of our work on tyrosine phosphorylation of polyoma mT and v-Src was at a joint meeting between the groups working on tumor viruses at the Salk Institute and the Fred Hutchinson Cancer Research Center held in Seattle on October 25, 1979. Subsequently, I talked about our work at a Hammer Cancer Workshop at the Salk Institute on November 13, 1979, and then again at the University of California, San Diego, Biology Department Noon Seminar on November 20, 1979. The word about our discovery spread through the community very fast and had even reached the other side of the Atlantic, because, at the end of October, I received an invitation to speak about our work at the “Protein Phosphorylation and Bio-Regulation” meeting that Julian Gordon, Ernesto Podesta, and George Thomas were coorganizing at the Friedrich-Miescher Institute (FMI) in Basel, December 10–12, 1979. The invitation had come about because I had told Tim Hunt, with whom I had collaborated when we were graduate students together in the mid 1960s in Asher Korner’s group in the Department of Biochemistry in Cambridge, about the phosphotyrosine story. Tim was going to be a speaker at the meeting and had persuaded the organizers to ask me—I was squeezed in as a speaker in the last session of the meeting on the topic of IFN and tumor viruses. As an indication of how fast word had spread, both Ray Erikson and Alan Smith talked in this session about their very recent experiments on tyrosine phosphorylation with v-Src and polyoma mT, respectively, having heard about our work through the grapevine, mostly through telephone calls. The younger generation may be surprised to learn that important scientific news spread quickly even before the Internet and email! I had taken a copy of our accepted PNAS paper to the meeting, and the FMI Xerox machines were kept busy making multiple copies to hand out to everyone that was interested. We also sent out over 20 preprints by mail in December. By the end of 1979, a large number of people knew that v-Src was a tyrosine kinase, and many groups had started working on tyrosine phosphorylation in their own systems. Indeed, over the next few months, we received many requests for phosphotyrosine because the compound was not commercially available.

Independently, Owen Witte, Asim Dasgupta, and David Baltimore at MIT had been working on the Abelson murine leukemia-transforming protein, v-Abl, and found that it too had an associated kinase activity that resulted in autophosphorylation of the v-Abl protein. They determined that the in vitro phosphorylated v-Abl contained pTyr and submitted a paper at the end of October 1979, concluding that this phosphorylation might represent a phosphoenzyme intermediate in phosphate transfer to another (protein) substrate. Their Nature paper came out at the end of February 1980 (12), shortly before our PNAS paper on the v-Src tyrosine kinase activity.

In keeping with my general philosophy about science communication, we had not attempted to keep our discovery of tyrosine phosphorylation a secret. Indeed, we had presented our work in public and told people in person or on the phone what we had found well before we had submitted our papers on polyoma mT tyrosine phosphorylation and v-Src as a tyrosine kinase. Indeed, Marc Collett, Tony Purchio, and Ray Erikson were able to submit a paper describing their evidence that v-Src is a tyrosine kinase to Nature in December 1979, based on what we had told them, and it was published in May 1980 (13). In addition, Stanley Cohen, who in 1978 was the first to show that the EGF receptor has an associated kinase activity that he reported was specific for threonine (14), quickly reevaluated this conclusion based on what we had done, and reported that the EGF receptor is also a tyrosine kinase in a Journal of Biological Chemistry paper submitted in June 1980 and published in September 1980 (15). In fact, by the end of 1980, we knew that there were at least four distinct tyrosine kinases: v-Src/c-Src (11, 13), v-Abl (12), v-Fps/v-Fes (16, 17), and the EGF receptor (15).

Coda

The totally unexpected functional commonality between the transforming proteins of an RNA tumor virus and a DNA tumor virus that our work had uncovered immediately suggested that tyrosine phosphorylation of cellular proteins might be a universal mechanism of viral transformation. At the time, we did not know whether the Py mT-associated activity was intrinsic or due to an associated enzyme, but, for v-Src we had strong evidence that the tyrosine kinase activity was intrinsic and essential for RSV transformation, and this insight had enabled us to carry out a series of experiments with v-Src in short order between September 18 and October 17. Thus, amazingly, all of the experiments demonstrating that v-Src is a tyrosine kinase and phosphorylates tyrosine in cellular proteins that were included as figures in the PNAS paper were done in a span of just one month—science could move fast even before the days of molecular biology. In retrospect, the key to discovering that v-Src is a tyrosine kinase was that we were working on both types of tumor virus simultaneously.

The PNAS paper “Transforming gene product of Rous sarcoma virus phosphorylates tyrosine” by me and Bart Sefton was published in March 1980 three months after acceptance (immediate online publication was not an option in 1979!). There was no press release and no news and views, and we did not even think about trying to patent the idea that targeting oncogenic tyrosine kinases might be useful therapeutically—it was a different time in the biological sciences. The concept that v-Src transforms cells through phosphorylation of cellular target proteins on tyrosine was immediately accepted as true by the community without any naysayers—the evidence was simply too compelling. Of course, whether this novel mechanism of cell transformation by tumor viruses would be relevant to human cancer we did not know at the time, but it did not take long for the first human oncogenes encoding mutant tyrosine kinases to be reported. This realization that activated tyrosine kinases could play a role in human cancer was the stimulus for the development of tyrosine kinase inhibitor drugs. Starting in 2001, over 20 small molecule tyrosine kinase inhibitor drugs have now been approved for cancer therapy, with more on the horizon. A gratifying outcome of this set of experiments on a simple chicken tumor virus.