Crystal Unclear

A behind-the-scenes look at how researchers solved the high-resolution crystal structure of the nucleosome core particle raises the age-old question of assigning credit in science.

Oct 15, 2015
Tracy Vence
Figure 2a: Nucleosome crystals in the form of hexagonal rodsJ.M. HARP ET AL., UNPUBLISHED DATA
Since its discovery in the late 1970s, biologists the world over have sought to better understand chromatin’s basic organizational subunit, the nucleosome—a complex of eight histone proteins around which winds a short length of eukaryotic DNA. Don and Ada Olins are two such biologists. In an effort to supplement their own microscopy-based chromatin research, the husband-and-wife pair in 1977 recruited structural biologist Gerard Bunick from the University of Pennsylvania to join their labs at Oak Ridge National Laboratory in Tennessee and help solve the nucleosome structure. In time, Bunick built up a group of his own; he secured independent funding and set up a lab down the hall.

Working in collaboration with his Oak Ridge colleague Ed Uberbacher, Bunick—who died of cancer in 2007—spent the better part of two decades working on the structure.

“As the years passed, Gerry became more independent and shared fewer details of his progress with us,” the Olinses, who are now at the University of New England, told The Scientist in an email. “Ironically, he was afraid that we would reveal details of his results when we traveled to meetings.”

During the late 1980s and early ’90s, Bunick’s was one of several teams that had published structures of the nucleosome at low resolutions, ranging from 7 Å to 3.1 Å. Like other groups working to solve the same structure, the Bunick lab found its progress toward a more-precise picture of the nucleosome stymied by an apparent resolution barrier. For a while, it seemed no one could make crystals that diffracted to a resolution finer than 3.1 Å—the limit reached by researchers at Johns Hopkins University in 1991. Technical challenges were in part to blame, but it was the way that nucleosomes bind specific regions of DNA that posed the biggest problem. The 145- to 147-base-pair sequences used in many nucleosome crystal preparations yielded fuzzy X-ray images.

Over lunch one day, Loren Hauser, a staff scientist who worked with the Olinses, joined Bunick and Uberbacher in a conversation about the DNA sequences they were using. To overcome the resolution barrier, the structural biologists needed sequences that would wrap symmetrically around the histone octamer, they explained. Hauser just happened to have studied repetitive DNA sequences while he was a PhD student. Maybe alpha-satellite DNA sequence data from his thesis could help Bunick and Uberbacher, he suggested. Did they want to have a look?

Hauser’s thesis data in fact would prove critical to Bunick and Uberbacher’s team.

Little did he and his Oak Ridge colleagues know, Hauser’s thesis data would also be critical to the success of their fiercest competitors.

Three’s a crowd

Within structural biology, crystallography is a notoriously competitive field. It’s also a fairly small one. The number of crystallographers working on chromatin is especially small.

“Chromatin has always been kind of a shark tank—a very, very competitive field,” says Joel Harp, who worked with Bunick and Uberbacher while he was a research associate at Oak Ridge. Now a biochemist at the Vanderbilt University Medical Center in Nashville, Tennessee, Harp says it is not uncommon for competing labs to keep tabs on one another’s progress by investigating who is working on what.

Looking back to his time spent working on the nucleosome at Oak Ridge, Harp recalls an atmosphere of intensity. In the race to solve the high-resolution structure of the nucleosome, three teams had emerged as keen competitors: Evangelos Moudrianakis’s group at Johns Hopkins; Timothy Richmond’s at the Swiss Federal Institute of Technology in Zürich (ETH Zürich); and Bunick’s.

“It was not a great time, because I basically got to focus on just one problem, and we were always kind of in a rush,” Harp says. “We knew we were in a race. We spent three years just trying to figure out how DNA binds to the histone octamer. We went almost five years without a publication.”

“There were only three laboratories,” recalls Moudrianakis. “Unfortunately, during that period . . . depending on who read what and how they interpreted that finding, there came many views and some polarization that, in the end, was not really justified.”

Arnold Stein (left) and Tim Richmond (right) at Cold Spring Harbor Laboratory Symposia on Quantitative Biology: DNA & Chromosomes, LXIV, 1993COURTESY OF COLD SPRING HARBOR LABORATORY LIBRARY & ARCHIVES“The competition was really nasty,” echoes Richmond. He recalls being questioned by attendees at a chromatin-focused Gordon Research Conference about his team’s nucleosome crystals. “I wouldn’t say . . . and they got really upset with me.”

Moudrianakis recalls having exchanged a few heated letters with Aaron Klug, who performed foundational X-ray crystallography research at the Medical Research Council (MRC) Laboratory of Molecular Biology in Cambridge, U.K., and had been Richmond’s postdoc adviser from 1978 to 1980. While a senior staff scientist at MRC, Richmond had coauthored with Klug and their colleagues a 7 Å nucleosome structure published in Nature in 1984. Klug had ceded the nucleosome project to Richmond when his former trainee left the MRC to start his own lab at ETH Zürich in 1987.

Richmond remembers those letters between his former adviser and Moudrianakis too. “It was really highly competitive,” says Richmond. “At the time, the nucleosome or the histone octamer were the structures to have. Those were the number-one things to do, at least in my opinion at the time.”


Aaron Klug discussing the high-resolution crystal structure of the nucleosome: “I interceded and said, [Evangelos] Moudrianakis is wrong, totally wrong, and we can demonstrate that.”WEB OF STORIES

Tensions came to a head in 1993. The year was punctuated by several society meetings, at which members of all three competing labs presented on their progress toward solving the nucleosome structure. Nineteen-ninety-three was also the year that, using the unpublished sequence data Hauser had generated while a PhD student at the University of California, Irvine, the Oak Ridge group made nucleosome crystals that finally broke the 3.1 Å resolution barrier.

Rejection

From Hauser’s alpha-satellite DNA, Uberbacher selected 73 base pairs with which to create a palindromic sequence. “This was viewed as key to making sure that they would line up precisely,” Uberbacher tells The Scientist. To produce crystals that diffracted X-rays to better than 3.1 Å resolution, he adds, “the molecules had to line up exactly in the correct way all along the lattice.” While assessing repetitive DNA sequences they might use, “we looked at a number of different sequences and tried to come up with mathematical ways to predict where histone octamers would form, and see if they were consistent with experimental data,” Uberbacher continues. It ended up that a stretch of Hauser’s sequence “had the best signal of any of those we looked at,” he says. Hauser adds: “The palindrome was the last piece of the puzzle . . . to analyze the diffraction pattern and make sense of it. Otherwise, the diffraction pattern was too complicated.”

After replicating the results and consulting trusted colleagues in the field, the team prepared a manuscript. Among the colleagues Bunick contacted was one of his former Oak Ridge collaborators, MRC’s Venki Ramakrishnan, who suggested the team submit its results to the Journal of Molecular Biology, where Klug was an editor.

“In those days, there were not many journals that would take a crystallization paper. JMB was one of those journals,” Ramakrishnan tells The Scientist. “I said [to Bunick], ‘Maybe Aaron [Klug] would be interested because he is interested in chromatin.’ And I said, ‘He [Klug] doesn’t work on the nucleosome anymore, because the project has been taken by Tim Richmond at the ETH.’”

“I thought of Klug as a grand old man of the field who would be happy to see progress,” Ramakrishnan continues. “I didn’t think he would care who did the nucleosome. Klug would have had no conflict of interest since he was not competing to solve the structure.”

In a manuscript submitted in August 1993 (PDF) and acknowledged by JMB two months later, Bunick and Uberbacher’s team described the crystallization of a nucleosome core particle containing 146 base pairs of palindromic DNA that, at best, diffracted X-rays at a resolution of 2.6 Å. These crystals, the team believed, were key to solving the nucleosome structure.

In November 1993, Bunick’s team received notification that the paper had been provisionally accepted. “I have consulted a referee whose report is enclosed and I agree with his recommendations,” Klug wrote (PDF).

Figure 2b: Nucleosome crystals in the form of hexagonal rodsJ.M. HARP ET AL., UNPUBLISHED DATA“This work is of interest because it reports an advance in the published resolution of diffraction data for this important nucleoprotein complex. However, it is difficult to understand how the work of Richmond and his colleagues on previous reconstitution and crystallization with defined sequence DNA was overlooked, as well as the crystallographic studies of particles under different hydration conditions by Klug and his coworkers,” the anonymous reviewer wrote. “These references must be cited properly throughout the submitted manuscript before it is acceptable for publication.”

The reviewer went on to suggest that, to make it suitable for publication, a resubmission of the manuscript should also include additional crystal preparation information as well as diffraction images.

“I believe we were suspicious of the comments at the time,” Uberbacher says. The Oak Ridge researchers were put off by some of the reviewer’s requests and worried that their competitors might get wind of their progress. Still, the researchers intended to resubmit; five months later, they did.

The Scripps Research Institute’s Peter Wright, editor in chief of JMB since 1990, tells The Scientist that neither the journal nor its publisher, Elsevier, can identify the reviewer. “I am afraid there are no records of submissions and reviews from the early 1990s, either in my office or at Elsevier,” Wright noted in an email. “The old paper manuscripts and review files from the 1990s, which took up several filing cabinets, were discarded long ago.”

Klug, now 89, no longer gives interviews. Once Richmond took the nucleosome project to ETH Zürich, “Dr. Klug’s own research interests had moved over entirely to studies of the structure and function of zinc-finger proteins involved in regulation of gene expression,” Richard Henderson, who succeeded Klug as president of the MRC Laboratory of Molecular Biology in 1996, wrote in an email to The Scientist. “By 1993, therefore, Dr. Klug would have had no direct conflict of interest in editing a manuscript for J.Mol.Biol. on either chromatin or the nucleosome.”

After receiving a second provisional acceptance from Klug (PDF), the Oak Ridge researchers, feeling as though they’d been given the runaround, withdrew their submission in 1994. An expanded version of their work was eventually published—two years later—in Acta Crystallographica. According to Google Scholar, the paper has been cited 27 times.

“Crowning results”

It was the ETH Zürich group that ultimately won the race to solve the nucleosome structure, in September 1997. In a Nature paper cited 5,987 times to date, according to Google Scholar, Richmond and his colleagues presented a crystal structure of the nucleosome core particle at an unprecedented 2.8 Å resolution.

The crystal structure of the nucleosome core particle consisting of H2A, H2B, H3, H4 core histones, and DNA. RSCB PROTEIN DATA BANK 1KX5; C.A. DAVEY ET AL. (VIA WIKIMEDIA, ZEPHYRIS)“I have been working on this problem nonstop for 18 years,” Richmond told The New York Times one month after his team’s results were published. “We were always worried about competitors. But the main driving force is the desire to see what the thing looks like. Fortunately, I was able to infect my colleagues, and they became as excited about it as me. As I talk to you it begins to sink in that we actually did it.”

“All in all, I spent about eight years on the project,” says study coauthor Karolin Luger, who was a research assistant professor at ETH Zürich when the paper was published and now leads a lab at the University of Colorado Boulder.

Another study coauthor, David Sargent of ETH Zürich, tells The Scientist: “As in all major scientific achievements, there’s a huge amount of spade work: trying out blind alleys, flashes of inspiration, developing new methods, making use of the latest experimental advances . . . that ultimately enables the crowning results. This was certainly the case here.”

Reading the published results, members of the Oak Ridge team were flabbergasted. The sequence reported in their competitors’ article, a sequence which was critical to the success of the crystal structure, differed from the one they submitted to JMB in 1993—and later published in Acta Crystallographica—by a single base pair. How, if not from their March 1996 Acta Crystallographica paper, could Richmond’s team have converged upon a near-identical sequence, Bunick and his colleagues wondered. The Oak Ridge group suspected foul play, but could not show that the similarity between the sequences was anything other than coincidence.

A careful examination of references cited within the 1997 Nature paper reveals an omission. Richmond and his colleagues cited the Oak Ridge team’s 1996 Acta Crystallographica paper as proof of concept: “The potential for structure determination of a mutant of one of the α-satellite repeats has been noted,” the group wrote. Later in the article, the authors referenced a 1982 PNAS paper—in which Hauser’s PhD adviser at UC Irvine, Barbara Hamkalo, and her colleagues described a cloned repetitive DNA sequence—as the source of the 146 base-pair alpha-satellite DNA sequence they used to prepare nucleosome crystals. But that 1982 paper does not contain any sequence data.

In fact, The Scientist has learned, Richmond and his colleagues did not cite the actual source of the all-important sequence data: Hauser’s unpublished thesis.

Covert operation

Richmond tells The Scientist that a member of his lab obtained Hauser’s thesis from UC Irvine directly—without contacting Hauser or Hamkalo.

“One certainly knows that she [Hamkalo] was working on DNA sequences in chromatin, and one wants to cover their bases,” Richmond says. “One way or another it was learned that there was a PhD thesis from an interesting lab.”

“I don’t remember how I came up with the thesis, I think one of my students did,” he continues. “We just wrote to the school . . . and they sent us a hard copy of the thesis we could read.”

According to a spokesperson from the office of Special Collections and Archives at UC Irvine Libraries, the lone original hard copy of Hauser’s 1985 thesis, “Structural analysis of members of a repeated DNA family in primates,” was circulated three times, and last returned to the university in 1995—before the Oak Ridge team’s Acta Crystallographica paper, which cited the thesis, was published.

One way or another it was learned that there was a PhD thesis from an interesting lab.—Tim Richmond,
ETH Zürich

Asked why his team did not contact Hauser or his former adviser to request the data, Richmond demurs, recalling acts of calculated secrecy. “Look, I’m telling you, it was very competitive. We like to keep our tracks covered for that kind of thing,” he says. “We don’t want to give away information about what we were doing then—what we were trying to do.”

Following the exchange of several emails, Hamkalo declined to be interviewed for this story.

Luger says she does not clearly recall how her team first learned of Hauser’s thesis. Sargent says he did not take part in that portion of the project. The Scientist received no response from emails attempting to contact study coauthor Armin Mäder and was unable to locate the study’s fifth author, Robin Richmond (who was married to Tim when the paper was published).

Tim Richmond says he and his colleagues did not cite Hauser’s thesis as the source of the sequence data they used because of an oversight. (Luger tells The Scientist she did not at the time realize the citation was left out.) He denies having reviewed—or had anything to do with—the Oak Ridge group’s 1993 submission to JMB. While the Zürich group did circumvent the traditional avenue for requesting data (asking the author), Richmond maintains that he and his colleagues played fair. “We did what we had to do in terms of investigating,” he tells The Scientist. “We looked at a lot of different sources [of sequence data], and that’s the one we found. We were able to take information from that thesis and use it to our advantage.”

Until The Scientist shared its reporting, Hauser did not know that the ETH Zürich team had used his thesis data. “If they went to the trouble of citing [the 1982 PNAS] paper, then they could have gone to the trouble of citing something more accurate,” Hauser tells The Scientist. “I consider it poor professionalism.”

Where credit is due

According to Ramakrishnan, had the ETH Zürich researchers used another DNA sequence, they might well have still been first to solve the nucleosome structure. “If they [the Oak Ridge team] thought this sequence was this magic bullet, and they were worried about it, they should have waited until they had more data—actually made progress toward the structure,” Ramakrishnan says. “Maybe, in hindsight, they should have sent it [the 1993 manuscript] somewhere else. But I could hardly predict that it would have been rejected,” he adds. “Had it been accepted, I’m not sure it would have changed the course of events much.”

“I got along very well with Bunick, liked him personally, and was sorry to see him scooped,” says Ramakrishnan.

Given a chance to rewrite the past, Harp says, he would not agree to sending his team’s manuscript to the former adviser of a competitor. Uberbacher, on the other hand, is not so sure he’d change anything. “I think we did everything correctly and submitted it to a journal we trusted, and we had every reason to believe our data would be handled correctly,” he says. “We were a very small lab and we didn’t have a lot of clout. . . . When you’re in science, you have to trust people.”

At some point, most every working scientist will feel slighted by a peer who, to his mind, hasn’t properly acknowledged his contributions or sufficiently cited his work in an article. Most often, scientists fail to cite or otherwise acknowledge the achievements of others because they are unaware of them. Sometimes they intentionally disregard the work of others. Publishing practices, such as journal citation limits, are part of the problem. Politics undoubtedly plays a role in some cases.

Research institutions, funding organizations, scholarly associations, and journal publishers all set their own rules as to who should be credited for scientific achievements. According to the US National Academies’ Committee on Science, Engineering, and Public Policy, anyone who makes an intellectual contribution to a project should somehow be credited for it. “When a beginning researcher makes an intellectual contribution to a project, that contribution deserves to be recognized, including when the work is undertaken independently of the laboratory’s principal investigator,” the committee wrote in its 2009 book, On Being a Scientist: A Guide to Responsible Conduct in Research: Third Edition.

Markus Röthlisberger of the Swiss National Science Foundation (SNSF) tells The Scientist in an email that his organization treats the assigning of credit in publications as though it were evaluating a grant proposal: “When submitting an application to the SNSF, applicants have to confirm that . . . ‘Earlier work of the applicants and third parties is declared as such and publications of the applicants and of third parties are correctly cited.’”

When a beginning researcher makes an intellectual contribution to a project, that contribution deserves to be recognized, including when the work is undertaken independently of the laboratory’s principal investigator.—US National Academies’ Committee on Science, Engineering, and Public Policy

That anyone who contributed to the success of a project should be credited for their contributions is largely accepted as a truism in the scientific community, says Melissa Anderson, a professor of higher education and bioethics at the University of Minnesota who studies scientific integrity. So there’s no question that the ETH Zürich team should have cited Hauser’s thesis, published or not. “If [a] dissertation is available, it perfectly well can be used, as long as it’s cited,” says Anderson. “The fact that they used it without citation, that sounds really suspect to me. . . . Nobody should use somebody’s work without citing it. That’s totally unacceptable.”

Citation snubbing can negatively impact a scientist’s career, Anderson adds. “Credit for your intellectual achievement is the coin of the realm. That’s what everything is based on—your ability to get a position, your ability to advance, to get tenure, to get salary raises, to get grants, to get in national and international associations,” she explains. “When that’s compromised, that’s definitely causing harm to your career.”

Hauser agrees his contribution to the winning publication should have been credited. While he would like to see the record corrected, he doubts a citation will have any impact now, nearly two decades later. This summer, he began the latest phase of his career—teaching high-school chemistry.

Richmond says there’s no need for a correction. “There has never been an objection registered with me about the lack of a citation to the Hauser thesis by the editors of Nature, by the reviewers of our paper, or by anyone else in the subsequent 18 years after the publication of our article,” he wrote in an email.

Correction (October 15): This story has been updated from an earlier version that incorrectly stated Moudrianakis and his colleagues published an erroneous description of the surface of the nucleosome’s histone octamer in PNAS in 1993. An incorrect structure was in fact published in Science in 1984. Moudrianakis and his colleagues later corrected the record. The Scientist regrets the error.

Update (October 16): Links to the websites of Don and Ada Olins have been updated from those included in an earlier version of this article.

Update (October 16): Following publication of this article, Ramakrishnan wrote to emphasize the contributions of Moudrianakis and his colleagues to deciphering the high-resolution crystal structure of the core histone octamer of the nucleosome. “As a result of Moudrianakis’s 1991 paper, people had a general idea of what the histone octamer looked like,” he noted in an email to The Scientist. “A knowledge of the general fold of each histone and the architecture of the octamer would have been useful for interpreting later structures like the Luger et al. work that had the complete complex including the DNA.”

Moudrianakis pointed to Richmond and his colleagues’ 1993 presentation, “Studies of nucleosome structure,” at the Cold Spring Harbor Laboratory Symposia on Quantitative Biology: DNA & Chromosomes, LXIV, in which the authors noted the contributions of Moudrianakis and his colleagues (PDF, emphasis Moudrianakis’s).
 
Editor’s note (October 27): Editors who reviewed this story prior to publication did not alter wording that described Harp et al.’s submissions to the Journal of Molecular Biology as provisionally rejected. Upon further examination, because Klug indicated he would consider the manuscripts for acceptance once the requested revisions were made, the papers were in fact provisionally accepted. This article has been updated accordingly.