4. The end of a beginning

It is customary fate of new truths

to begin as heresies

and to end as superstitions.

- Thomas H. Huxley

The manuscript was first sent to Nature in March, 1993 with a title of 'The chromosomal DNA of Streptomyces lividans 66 is linear' with Yi-Shing, Helen, and David as my co-authors. About a month later, it was returned without even being sent to referees for review.

We were disappointed, and opted hopefully for the second journal - Science. David had told our story to Richard Losick at Harvard, who studied differentiation in Streptomyces (and Bacilli). Losick was on the editorial board of Science, and he was interested. We rewrote and sent the revised manuscript to Science on 25, May. A piece of good news came on a postcard about a month later, saying that the manuscript was under review by the referees. This meant that the paper had passed the initial screening and was under consideration for publication. On 9, July the comments by the referees came by fax. The situation appeared very cozy. One referee agreed strongly for publication in Science ('It is therefore of general interest and appropriate for Science'), and raised a small point in one of the presentations. The second referee considered 'the thesis that the chromosome of S. lividans is linear is very exciting and looks to be correct'. He/she raised a few questions, which were easy to respond to. So, we quickly sent in the replies to the referees' comments by fax, and waited again hopefully. We were getting very anxious because the paper by the Kaiserslautern group had already come out in Journal of Bacteriology in June. Ours was still up in the air!

We waited for about a week, and I became impatient, so I called up Science to inquire about the outcome, and was told that it would be decided in a forthcoming editorial meeting. The editorial meeting eventually turned the paper down. Yi-Shing and I was really upset, so were David and Helen when I told them the bad news. I was particularly upset by being rejected at an editorial meeting two months after our submission. Now sixty precious days had been wasted, and we faced a new revision again.

Stan Cohen of Stanford University came to Taiwan on one of his several sabbatical visits. As usual, he inquired about the progress of our story. He was sympathetic, but not surprised. He had warned me about the editorial politics of these journals like Nature and Science. A comforting and assertive remark by Stan did me a lot of good: ' An important discovery is important no matter where it is published'.

Internal replication origin on the linear plasmids

In 1992 Dov Shiffman, a postdoctoral fellow in Stan Cohen's lab at Stanford Medical School, made an interesting observation on a 12-kb linear plasmid pSCL from Streptomyces clavuligerus. They subcloned the internal portion of the pSCL DNA on an E. coli vector. This circular equivalent of pSCL replicated in S. lividans as a circular molecule. This was amazing and unexpected. Now, removal of the ends eliminated the possibility of replication from the ends, necessitating some sort of replication from an internal origin. Thus, pSCL appeared to have dual modes of replication, one starting from both termini, and the other starting internally. This was in contrast to the well-studied cases of f29 and adenoviruses.

The work of Shiffman and Cohen became particularly significant when we were convinced of the linearity of the Streptomyces chromosomes. Two groups of scientists led by Frank Schmidt at Saint Louis, Missouri and Hildgund Schrempf at Osnabrk had isolated the oriC sequences on the chromosomes of S. coelicolor and S. lividans. Colin Smith, Steven Oliver and co-workers at University of Manchester in U. K. sent us their experimental evidence showing the oriC of S. coelicolor to be functional and replication initiated there appeared to cover most of the chromosome. It, therefore, seemed that the Streptomyces chromosomes might also have a dual modes of replication, just like their plasmid counterparts.

Spontaneous circularization of the chromosome

The existence of an internal origin for replication on the linear chromosomes and Shiffman and Cohen's demonstration of the replication of circularized pSCL prompted me into thinking of the circularization of the linear Streptomyces chromosomes. The first hint came from the combined map of Matthias' and ourselves.

The fact that the telomeres fell in the unstable region of the chromosome told us that the telomeres were unstable. Looking at the maps of deletions in this region sent by Matthias, I began to see that in some of the mutants both telomeres appeared to have been removed. The simultaneous deletions of two telomeres might have been two independent events, or it might have been resulted from a single recombination that joined the arms. In the latter case, the chromosome would have been circularized (Fig. 9). The more I studied the maps, the more I suspected circularization was definitely possible, at least in some mutants.

I told Yi-Shing the idea in December 1992, and she quickly saw the important implications. I contacted Matthias, and he sent five of the deletion mutants in January 1993. Yi-Shing's first PFGE result said that, yes, the chromosomes in all these mutants appeared to be circular - they did not move into the gel. This was then confirmed by hybridization studies. (Matthias also expressed later in his paper a suspicion that these chromosomes were circular).

Fig. 10. Spontaneous circularization of Streptomyces chromosome mediated by recombination between the unstable sequences (gray) on the two arms. The terminal sequences, together with TP (solid circles), are deleted in the process.

This was exciting! We were witnessing the circularization of a linear chromosome, and the circularization happened spontaneously at a relatively high frequency of about one in a thousand! More importantly, these cultures with the circularized chromosomes appeared to grow as happily as the normal cultures.

Artificial circularization of the chromosome

This brought out a naughty notion from me: could we try to circularize the S. lividans chromosome ourselves by design? I thought of a scheme to circularize the linear chromosome using the gene targeting procedure. We constructed a bridge that would eventually connect the two telomeres with a kanamycin resistance gene (aphII). To the left of aphII, we placed a piece of DNA from one end of the chromosome, and to its right another piece from the other end. The whole constructed bridge was placed on an E. coli vector that could not replicate in Streptomyces. Another drug resistance gene (tsr for thiostrepton resistance) was placed between the cassette and the E. coli vector sequence (Fig. 10). Firstly, we transformed S. lividans with the recombinant plasmid and screened for thiostrepton and kanamycin resistant transformants, in which the plasmid was expected to have inserted into the chromosome by recombination. Yi-Shing grew up these transformants in the absence of thiostrepton and kanamycin to allow for a second recombination at the other side of aphII. This would join the two arms of the chromosome, and replace the two telomeres with a continuous piece of DNA, the aphII gene (Fig. 10).

Fig. 11. Artificial circularization by targeted recombination. The suicide vector containing the kanamycin resistance gene (aph) flanked by two sequences from the two arms of the ZX7 chromosome was used to transform ZX7. Kanamycin resistant transformants, on further segregation, gave rise to circular chromosomes, in which the two arms of the chromosomes were joined by the aph gene to form a circle.

It took Yi-Shing only a little time to construct the suicide vector, which she immediately used to circularize the S. lividans chromosome. It was relatively easy to obtain the first thiostrepton resistant transformants, and even more easy to obtain thiostrepton sensitive derivative(s). About one third of the thiostrepton sensitive spores had circularized chromosomes. This was a pleasant surprise, as we were ready to face failure. A circularized chromosome like this might be lethal to a Streptomyces cell for all kinds of reasons.

The ease of generating circular chromosomes from linear chromosomes told us that the circularization was readily acceptable to the cells, although some of the cultures with circularized chromosomes grew more slowly and sporulated more poorly. Perhaps the Streptomyces cells were not good at dealing with the complication of replicating circular chromosomes. The important point is that the linear chromosome could be circularized at all. This brought up an interesting question of the evolution of the topology of bacterial chromosomes.

Linear chromosomes at the Keystone symposium

From the 18 to 25 April, 1993, a Symposium on Bacterial Chromosomes was held in Keystone, Colorado. I took the opportunity to visit, John Spizizen and Jun Ito, now at the University of Arizona at Tucson, before the symposium. John was semi-retired, but still very active and energetic. Jun had switched his object of study from f29 to an E. coli bacteriophage PRD1, whose DNA was also an invertron. He was very interested in our story about the linear chromosomes of Streptomyces, but remained somehow skeptical. So was the audience in the seminar that I gave there. Perhaps I did not state the case well. Perhaps my English was getting rusty. Perhaps this new idea could not sink in easily.

The Symposium on Bacterial Chromosome was almost all about E. coli chromosomes. I put on my poster one afternoon. It received little attention, particularly from the younger persons. Howard Nash dropped by to renew our acquaintance started in my days of quadruplex DNA. I also met and had a lengthy discussion with Monica Riley, an expert on bacterial chromosomes from Marine Biological Laboratory (Woods Hole, Massachusetts). Kit Tilly from National Institute of Health Rocky Mountain Laboratory (Hamilton, Montana) had been working on the linear plasmids and linear chromosome of Borrelia burgdorferi, and she was writing a review with a colleague. She learned that we had submitted our paper to Science, and asked me to send her a copy. Later, we exchanged manuscripts and galley proofs, and eventually found our papers published next to each other.

A special issue of Journal of Cellular Biochemistry published concurrently with the Symposium contained the abstract of our presentation with the title 'The chromosome of Streptomyces lividans 66 appears to be linear'. The three words 'appears to be' reflected the caution when we submitted the abstract months ago. People who visited our poster that afternoon, however, would notice an unusual sight - the words 'appears to be' in the title had been stricken out and replaced with a big 'is'. We were already sure!

Accepted by Molecular Microbiology

The final rejection by Science was a demoralizing. We were also loosing time fast. Pierre Leblond's paper had appeared in the June issue of Journal of Bacteriology, and Redenbach's was coming out soon. In the paper that Pierre co-authored with John and Matthias, they broke up the original circular map of the ZX7 chromosome at the point between the H1- and the A-fragments (Gone was C-54). They attributed their inability to obtain a linking clone that would close up the circle to the discontinuity (making a reference to our discovery). The inability to clone a certain DNA by itself was not unusual and certainly did not constitute as evidence for the absence of that DNA. Thus their data was only consistent with the linearity of the ZX7 chromosome. Nonetheless, theirs was the first formal paper to mention a linear chromosome of a Streptomyces species. Our Keystone Symposium abstract published earlier said that the S. lividans chromosome 'appears to be linear'. David and I shared a strong anxiety that we ought to get our paper published very soon.

It was then August, 1993, and I was going to attend the International Congress of Genetics at Birmingham, UK. David was Chairman of the Programme Committee and had suggested to Keith Chater, a coordinator of the Workshop Novel Genetic Phenomena in Prokaryotes, to invite me to organize and chair that session.

On the way to UK, I visited John Cullum, his lab, and his family, at Kaiserslautern, Germany. After spending two nights in his house, I went on to John Innes Institute. Arriving at Norwich by train from London, I was picked up by Tobias at the station. That evening I met Matthias Redenbach for the first time at dinner in the Kissers' house. He was spending a few months there working on the S. coelicolor chromosome before going on to Kinashi's lab in Japan. We talked about current affairs and future plans. Somehow neither Matthias nor I brought up our previous exciting exchanges over the facsimile (They had to wait when we met again in Hiroshima a year later).

David and I were down to two choices: EMBO Journal or Molecular Microbiology. Stan Cohen had told me back in Taiwan that some journals (like Molecular Microbiology) had a policy for quick publication of papers that had been rejected by Nature or Science based on editorial, not scientific, considerations. It turned out that the Chief Editor of Molecular Microbiology, Chris Higgins, was chairing a symposium session at the Genetic Congress, and we could talk to him. We got hold of Higgins at the end of his session and told him our story. He said to send him the revised manuscript, and the Science referees' comments, and he would see what he could do. It was then we finally decided to do just that. We also decided to add our circularization experiments to the revised manuscript, which I originally planned to make another story and publish. David thought that since we were late, we might as well do it with a 'bigger bang'. I went along. Later when the paper came out, I was rather happy about the decision. It was a very good paper.

The workshop Novel Genetic Phenomena in Prokaryotes was well attended and went rather well. I gave a talk on our linear chromosomes - the first time this story was reported orally in a major international meeting. Questions were raised and points discussed. I believed that the most people in the audience were convinced and stirred.

After the meeting we drove back to Norwich. David put me up in a lovely country inn near John Innes Institute, and the spent the next two days with me revising and expanding the manuscript for Molecular Microbiology. I brought the revision back to Taipei, modified some figures, and added a couple more. I sent the revised manuscript together with our communications with Science to Chris Higgins by Federal Express. The manuscript reached Molecular Microbiology on the 7, August, and was accepted for publication two days later without further review. While our paper is 'in press', Kit Tilly sent from Montana a copy of galley proof of their review (co-authored with Joe Hinnebusch) also scheduled to appear in Molecular Microbiology. Their review 'Linear plasmids and chromosomes in bacteria' ended up timely in the same issue next to our paper, citing our discovery among others.

The face values for the 'circular' chromosomes

A lesson from our discovery, old but nevertheless important, is the caution we must take to escape from our conceptual blocks. Taking things at face value and taking analogy too far are two major pitfalls that are difficult to escape from, particularly in biology.

Let us first examine the issue of taking things for granted. The extremely complex and delicate structures of a cell make it tremendously difficult to study them in vivo. A biologist trying to isolate and analyze individual cellular components in vitro is always trapped in the same kind of dilemma as a quantum physicists facing the uncertain principle. The component may or may not be the same once you take it out of the cell, or it simply cannot be isolated intact.

Take the case of E. coli chromosome for example. Although its general size (4.7 Mb) and topology (circle) have been established and thousands of genes located on it, the evidence for its circularity has always been indirect and circumstantial until recently. One of the earliest and probably most powerful piece of experimental evidence has been John Cairns' famous autoradiogram of the E. coli chromosome. However, those circular images of the chromosome only demonstrated the circular disposition of the tritium-labeled nucleoside. They did not show the chromosomal DNA as a continuous circular molecule. The chromosomal DNA could have been linear with its ends held together by some non-covalent interactions. If we were to examine the linear chromosomes of Streptomyces the same way, we might also see circular images, for the same reason. The DNA molecules of adenoviruses and f29 DNA have been shown to form circular shapes like this, connected by cohesive terminal proteins.

The circular genetic map of E. coli, though consistent with a circular chromosome, is not a sufficient proof. We have gone over earlier the possibilities of getting a circular genetic map from a linear genome. The circular genetic maps of Streptomyces is now another case.

Recently, the E. coli chromosomes have been assembled from various kinds of restriction fragments into a circle using restriction and hybridization mapping. Still, there are potential pitfalls here. We have shown, in the case of Streptomyces chromosome, how inverted repeat at the ends of a linear chromosome may misguide the alignment. Although detailed restriction mapping will uncover the misalignment, it will be helpless against linear genomes with terminal direct repeats. The direct repeat at the two ends will give rise to overlapping cosmids that appear to connect the terminal DNA (into a circle). Linear genomes with circularly permuted sequences do not require terminal direct repeat to confuse us. They cannot be distinguished from circular ones by classical genetic or physical mapping. The trickiest genomes are those with both circular permutation and terminal redundancy, like phage T4. Therefore, critically speaking, the circles assembled with restriction fragments or cosmids are supporting but insufficient evidence for a circular chromosome.

In my opinion, the best procedure currently available for investigating the topology of a chromosome is a careful examination using PFGE. Unlike linear DNA, circular bacterial chromosomes (in Mb's) cannot move in PFGE, unless they are broken. Careful studies with proper controls can make clear distinctions. Yi-Shing has used PFGE to check out the chromosomes of some E. coli K12 strains. The chromosomes from exponentially growing bacteria stayed immobile, although cultures in stationary phase did give some linear DNA (of about 4.5 Mb), presumably breakdown products. Thus, the chromosome of K12 appeared to be circular. That was a relief.

So, if the best case of a circular chromosome - that of E. coli - was so difficult to prove, what about the other bacteria? I am afraid that a lot of the claims for circular chromosomes have not been as vigorously tested as should be. Many of them rely on the preconception of circular bacterial chromosomes. Many pieces of evidence that supported a circular chromosome did not rule out a linear chromosome with special terminal structures. In other words, they are circumstantial. Then, as Holmes said in 'The Boscombe Valley Mystery':

Circumstantial evidence is a very tricky thing. It may seem to point very straight to one thing, but if you shift your own point of view a little, you may find it pointing in a equally uncompromising manner to something entirely different.

The structural and functional homology shared by various subjects of studies a unique feature that distinguishes biology from other hard sciences. It has been frequent used (sometimes overused) by the biologists and the biochemists to make predictions and conclusions. Most of the time, it works fairly successfully. Occasionally a homology was carried too far and a mistake is made. Nature is not to be guessed or outsmarted. Not yet at least.

The pre-conception of a circular chromosome must have influenced some people, when they were constructing the physical map of a bacterial chromosome. One can hardly blame them. Two years ago, I certainly would have tried to assemble a Streptomyces chromosome map into a circle, if I had to do it. This mental block is compounded by blind spots, and the terminal inverted repeats, a main feature of the linear chromosomes of Streptomyces, is exactly the kind of blind spots that plays tricks on the map constructors.

Grace for all seasons

Throughout the ordeal, David bore a great deal of burden and dilemma. He was caught between their work which had just been published and our collaborative study which would soon prove it flawed. He was also caught between the results conveyed by fellow scientists at Kaiserslautern, and the results of Yi-Shing, Helen and me, which again was to prove the former wrong. The first dilemma was probably less complicated than the second one.

I had watched David's and Helen's feelings and reactions from the very beginning. After all, here was someone coming from the other side of the globe to tell you that you might have a major flaw in an important paper you just published. And he was soliciting your help to prove his case. It took the best of a scientist to accommodate a situation like this. Yet, during the course of our studies, David and Helen had never lost their objectivity, and helped me more than any visiting scientist could expect. Even later when mistakes were indeed discovered in their physical map of S. coelicolor, they never lost their good natures, and admitted them openly in our Molecular Microbiology paper.

David, during his life-long distinguished career, has received a string of awards and honors. In June, 1994 David was knighted by the queen for his outstanding scientific contribution and leadership. So, he is now Sir David. 'I regard this as a tribute to the team and the scientific community at Norwich than a personal honor', said he to the press. We knew he really deserved it, and because of that, he received jokes and teases from friends who loved him.

In May, 1994, Helen flew to Taiwan to visit us. Using her own money, she purchased a British Airways ticket with a special discount that she redeemed with a pile of food bills of over six hundred pounds (money, not weight) at the Sainsbury grocers. 'Will Yi-Shing be there?' she asked before the trip. Intimate collaborators for two years, the two had never met or talked on the phone.

The excitement of the historical meeting of Helen and Yi-Shing in the corridor outside our lab was only detectable by the sparkles in their eyes. I think only the uncertainty about Taiwanese custom restrained Helen from embracing Yi-Shing. Helen stayed in this part of the world for almost three weeks, plus a one-week trip to Japan to see Matthias. By the time she left, she became a good friend to everyone in the lab.

As to the second dilemma, I think David, Helen, and I have also dealt with it fairly. In retrospect, I think there were two reasons why I did not hesitate to help Matthias and his colleagues out. Firstly, they had communicated experimental results to us. Although we had started our quest for the linear chromosome of S. lividans, before we were aware of their work, their results, despite its mistake, did help speed up our progress. We did not base our final conclusion on their data. On the contrary, we made a point to be able to sustain our final conclusions on our own data. Their findings, nevertheless, helped. The second reason was that somehow from the beginning I had liked Matthias Redenbach, the vulnerable and frank student. I decided to help him out, and that turned out to be more than an act of good will. In the process of helping him I found the clue to the whereabouts and the structure of the other telomere.

Sometime I cannot help wondering how things would be, if we had not accidentally encountered and discovered the linearity of the Streptomyces chromosomes. Leblond and his colleagues would probably go on and publish their circular map of the S. lividans chromosome, further strengthening the circularity of the Streptomyces chromosomes. It would make future challenges much more difficult, although some day someone would certainly found the mistakes. Perhaps Matthias would eventually find his way to the ends of the ZX7 chromosome, if he continued his pursuit perseveringly enough. Just how long would we be kept from the truth? It's hard to say.

Cat out of the bag

We are at the end of the story, but scientifically we are at the beginning of a new era - the era of linear bacterial chromosomes.

The linearity of the chromosome in Streptomyces now appear to be a general rule. Beside the nine cases we discovered, people studying different species of Streptomyces in other parts of the world (such as John Cullum, Pierre Leblond, and Haruyasu Kinashi) are finding their chromosomes linear. I expect that we will soon find linear chromosomes in more and more bacteria with our new powerful techniques. And we know so little about these linear chromosomes! What has been described in the textbook about the metabolic activities of bacterial chromosomes were modeled on circular chromosomes. Some of these models will certainly not be applicable to linear chromosomes, and require re-thinking and re-investigation.. Besides these physiological concerns, the discovery of linear bacterial chromosomes has thrown a rock in the pool of eukaryotic evolution, raising important evolutionary questions that beg for serious attentions.

In the summer of 1994, I was invited to give a talk at the Annual Meeting of the Society of Actinomycetes (SAJ) of Japan at Hiroshima, and two talks at the International Symposium on the Genetics of Industrial Organisms (GIM) at Montreal. For the GIM meeting I asked Yi-Shing to give one for me, giving her an opportunity to practice and experience. The two meetings were close together, so I made both in the same trip. I first attended SAJ, in which all the talks were in Japanese, except mine and Matthias'. I spent an evening at the Kinashi's house, meeting his charming wife and daughters. Haruyasu Kinashi's lab had already put together a linear map of the Streptomyces griseus chromosome.

At the SAJ meeting, I was told by Kunimoto Hotta, a professor at the National Institute of Health in Tokyo and the president of SAJ, that I should be proud because I was the first speaker SAJ had ever invited from abroad. I was very proud indeed. So were my family. Among them, I think, this honor meant most to my ailing father, who was born in Taiwan during the Japanese occupation and received his university education in Tokyo.

Matthias, much more relaxed now, talked to me about the old times. He confirmed the story about how he had almost given up science when he discovered his mistake. He said that he was so afraid that there might be other mistakes as well. It was John Cullum who brought him back to finish his thesis. He felt guilty that some people mistook them as the first discoverers of the linear chromosomes of Streptomyces. He took all the opportunities he could to tell people that they would never have got their story right without our help, and that the credit for the discovery of the linear chromosomes of Streptomyces should rightfully go to us.

At the GIM meeting in Montreal, Yi-Shing and I spoke to an international audience. Despite her nervousness and inexperience, Yi-Shing did a good job delivering her speech. Both of our talks appeared to be well received. To many people this was their first time to hear about the linearity of the Streptomyces chromosomes formally. We added some exciting new discoveries since our publication in Molecular Microbiology. Many interesting questions and comments were raised during and after the talk. New paths had been opened, inviting us to pursue into new lands. New angles had been revealed for old puzzles. It was exciting.

A Streptomyces friend, Eric Cundliffe from University of Leicester, after a lengthy discussion and reflection by the exhibition booth of Panlabs, made this departing remark,

'Carton, you have let the cat out of the bag'.