Because Felix knew that my pursuit of a post-doctoral fellowship was driven by an interest in the genetic basis for antibodies, he introduced me to Sheldon Dray, a widely recognized expert in Ig allotypes who was at the University of Illinois in Chicago (Figure 2).

The allotypic markers that Sheldon was working with in the V_H regions of rabbit Ig greatly influenced these discussions. The V_H allotypes were problematic because they were present on almost all Ig molecules and yet were inherited as alleles. I’ll explain more about this later.

The attitude in Sheldon's lab was conducive to thinking about the antibody problem because the environment he created fostered an open exchange of ideas and collegiality among faculty, staff and students. We had open lab meetings and were actively encouraged to share all of our data. In all of my years with Sheldon, I never heard him allude to, much less directly state, a need for any type of secretive privacy- it wouldn't have even occurred to him.

As I was completing my post-doctoral training with Sheldon, he invited me to remain at the University of Illinois as an assistant professor. In setting up my own laboratory I was thinking about the problems we would investigate. While most immunologists had turned to studying the antibody problem in mice, I couldn’t walk away from the rabbit model and what for me was the burning question: How could rabbit V_H allotypes be inherited as alleles?

After working on this problem for a few years, I had the opportunity to work with Benvenuto Pernis at the Basel Institute of Immunology in Basel, Switzerland. Ben taught me to think about problems in an intensely inclusive manner. A typical day would start over coffee with Ben certain of his hypothesis to explain a given phenomenon, and the day would end with Ben just as certain of a different hypothesis to explain the same problem. He had an uncanny ability to bring together pieces of information and to synthesize them in a manner that enabled you to see things in a new way while simultaneously remaining aware of alternative explanations.

Once back in Chicago, Chris Martens, a post-doctoral fellow, joined my laborabory and together we tried several new biochemical and serologic approaches in an attempt to crack the allotype problem. After a year or so we hit a wall and realized we had to do something different. About this time molecular biology emerged as a new field, and Chris and I became excited by imagining the possibilities that this new field and its technology offered for solving the allotype problem. We knew we would have to learn this new field from scratch yet our shared enthusiasm made it easy to take the plunge and retool. This was no small decision because we were both trained as immunochemists, and molecular biology represented a major new, somewhat daunting, challenge.

We needed to find the right guidance to make this next step. I thought of Lee Hood who at that time was already well into molecular biology, and asked for his advice. He replied with spontaneous generosity, inviting us to work in his laboratory at Cal Tech in Pasadena. He said it would take 6 months, but because of time constraints, we knew we’d have to do it in about half the time. Working round-the-clock with the help of Kevin Moore, we accomplished our goal within 3 months.

By the time we were back in Chicago, we had a rabbit cDNA library that we screened to find an IgG heavy chain gene. Yet until we sequenced the gene we couldn’t be certain that we had cloned a rabbit gene, let alone a heavy chain gene. Chris had learned DNA sequence analysis and began to sequence our clone. Of course, at this time there were no DNA sequencers, nor lab computers and certainly no kits to help us. The day we developed the autoradiograph of the sequencing gel, Chris, my other students, post-docs and technicians gathered in my office, anxious to help determine whether we had cloned a rabbit heavy chain gene. We had no idea where in the gene our nucleotide sequences were coming from- which meant that we had to translate each reading frame in each of the two orientations and then compare them to the sequences in Kabat’s book of Ig protein sequences (2). Suddenly we found that one of the reading frames encoded the hinge region of the rabbit IgG heavy chain. Our foray into molecular biology had begun!

With this new tool- molecular biology- we could begin to clone the V_H genes as a means to discover how the V_H allotypes were inherited as alleles. We knew that V_H allotypes a1, a2 and a3 were present on 80% to 90% of Ig molecules and that they were inherited as alleles (3). However, if there were hundreds of V_H genes in the germline and most of them encoded the V_Ha allotypes, then why weren't there crossovers at meiosis so that eventually the V_H allotype repertoire would be the same for rabbits of different V_H allotypes and the allotypes would then no longer be inherited as alleles (1)? To begin to answer this question we decided to analyze the germline V_H genes from rabbits of the different V_H allotypes.

By now, Bob Becker joined my group. As he and I talked long into the evenings and weekends he became very interested in the allotype problem and decided that he wanted to work on it. I was about to find out not only how the allotypes could be inherited as alleles but, more importantly, what can happen when you team up with someone who shares your excitement.

We cloned the 3’-most V_H genes from rabbits of each of the three allotypes, a1, a2 and a3 (4). The nucleotide sequence showed that the 3'-most V_H gene encoded a prototypic V_H allotype molecule; that is, V_H1 from a1 rabbits encoded a1 allotype molecules, V_H1 from a2 rabbits encoded a2 allotype molecules, and V_H1 from a3 rabbits encoded a3 allotype molecules. About that same time we generated transgenic rabbits that developed B cell leukemias, and when Bob sequenced the VDJ genes from these cells, he found that the sequences of the V_H regions were identical to those of the 3'-most germline V_H gene, V_H1 (4,5).

Many of the other germline V_H genes appeared functional, but they did not encode a prototypic V_Ha allotype. Now we began to wonder whether V_H1 was used in most VDJ gene rearrangements. If so, this finding would provide a simple explanation for the allelic inheritance of the V_Ha allotypes- allelic V_Ha allotype-encoding genes were being used in VDJ gene rearrangements in most B cells.

But how could we test this?

A conversation I had with a colleague stimulated a breakthrough idea. During a ski trip to Switzerland I described the mutant rabbit, Alicia (6), to Michael Steinmetz, with whom I had worked in Lee Hood’s laboratory. The Alicia rabbit had a genetic defect such that it had almost no V_Ha-allotype Ig molecules; instead, it had what we refer to as V_Ha-negative molecules, that is, molecules without the V_Ha allotype. Although we had thought that this rabbit must provide a clue to the allotype inheritance problem, until now, we could not imagine what defect could have resulted in the loss of V_Ha allotype molecules, short of the loss of the V_H chromosomal region, or the loss of some regulatory element for V_H gene expression. From that discussion came the idea that if our hypothesis were correct, namely that V_H1, which encodes prototypic V_Ha allotypes, was used in most VDJ gene rearrangements, then Alicia should have a defective V_H1 gene.

As soon as I returned to Chicago, I shared my thoughts with Bob and sure enough, when we cloned the V_H1 region from Alicia rabbits, we found a 10 kb deletion that included V_H1 (4). So Alicia was like a V_H1 knock-out rabbit, and in the absence of V_H1, this rabbit did not produce normal V_Ha allotype molecules. We were then convinced that V_Ha molecules, which represent 80% to 90% of total Ig molecules, were encoded by V_H1 and consequently that V_H1 was preferentially used in the VDJ genes of most B cells. Now finally, we had resolved the allotype problem, the allelic inheritance of the allotypes was due to preferential usage of the allelic V_Ha-allotype encoding gene, V_H1 (7).

Looking back on the process by which the allotype problem was solved, I realized that once Bob and I started working together, the solution came relatively quickly. I attribute this success largely to a match between Bob’s and my motives coupled with his openness and availability. Because of Bob’s motivation and maturity my primary responsibility to our working alliance consisted of giving him the space and freedom to pursue his interests without any particular goal in mind. And in fact, by letting him decide what to focus on next, he made one of the most important discoveries in the history of my laboratory.

We had focused on the allotype problem for so many years that I never thought about whether there would be life after the allotypes. But as luck would have it, the solution to the allotype problem left us with an even bigger question. That was, if only one V_H gene was used in most of the VDJ gene rearrangements, then where was all of the antibody diversity coming from? Clearly it was not being generated by combinatorial joining between multiple V, D and J gene segments. After Bob began to carefully analyze VDJ genes from adult rabbits and compare them to their germline counterpart, he found the answer. Let me explain.

Here (Figure 3) I compare the nucleotide sequence in FR1 of a VDJ gene cloned from mesenteric lymph node with the sequence of the germline counter part, V_H1 (8). Differences in sequence indicate that this gene has been somatically diversified. You notice that the VDJ gene differs from the gene, V_H1, used in the VDJ gene rearrangement, by several nucleotides clustered together. Of particular interest is the insertion of an entire codon at position 2. Such codon changes do not arise by somatic mutation, the process of somatic diversification in mouse and human Ig genes.

The codon insertions reminded us of the elegant work of Reynaud and Weill in Basel. They showed that chickens use only one V_H gene and one Vl gene in VDJ and VJ gene rearrangements, respectively, and that the antibody repertoire develops through somatic gene conversion, a non reciprocal recombination event, that uses the upstream V genes as donors (9,10). Because of the codon insertion we thought that perhaps rabbits also diversify their VDJ genes by gene conversion. If so, we should find upstream V_H genes with sequences identical to the diversified regions of this VDJ gene that could be used as donors in a gene conversion event.

And indeed we did find such donor genes. In this example (Figure 3), we found that V_H6 had a sequence identical to the diversified region of the VDJ gene. It appeared that V_H6, or a gene very similar to it, was used as a donor gene in a gene conversion event. We concluded that rabbit VDJ genes diversify by a somatic gene conversion-like mechanism. This finding was, in many respects surprising because Ig genes of other mammals, especially mouse, had been examined extensively by several investigators, and it appeared that somatic diversification occurred by a hypermutation mechanism (11).

By this time Mary Crane had joined us and she was interested in the timing of somatic gene conversion and whether the somatic diversification process in rabbit was driven by exogenous antigen. We knew that VDJ gene sequences of newborn rabbits were undiversified, and we suspected that adult sequences were diversified extensively. Mary PCR-amplified VDJ genes from peripheral blood leukocytes (PBL) of newborn to 8-week-old rabbits and compared their nucleotide sequences with the germline V_H1 gene (Figure 4) (12). The sequences from newborn rabbits were essentially undiversified. By 4 weeks of age Mary found a few mutations, and by 6 to 8 weeks of age, essentially all of the VDJ genes were extensively diversified.

This observation, along with the observation by Chander Raman in my laboratory, that all rabbit B cells are CD5⁺ (13), started us thinking that maybe B cells develop in rabbit differently than in other mammals, such as mouse and human. In fact, it seemed that rabbits were more like chicken than like other mammals (Table 1). Rabbits and chickens both use a very limited number of V_H genes in VDJ gene rearrangements; they use gene conversion as a means to somatically diversify their VDJ genes; and the somatic diversification occurs in essentially all B cells early in ontogeny. We knew that in chicken the Ig genes diversify in the bursa before hatching and further, that in the rabbit, the appendix is a major lymphoid organ that is histologically similar to the bursa (14).

Stimulated by Rod Langman and Mel Cohn, who organized a Forum in Immunology entitled "The challenges of chicken and rabbits to immunology" (15), Mary, Chander and I proposed a model for B cell development (Figure 5) that included the appendix and other gut-associated lymphoid tissue (GALT) as the site of somatic gene conversion (16).

We hypothesized that soon after birth, B cells migrate from fetal liver and bone marrow to GALT where they encounter microbial antigens. These microbial antigens, possibly superantigens, would stimulate the B cells to proliferate and to begin to somatically diversify their VDJ genes. Further, we suggested that the B-cell repertoire that developed in GALT during the first few weeks of life was maintained by self-renewing B cells and that new B cells were not continuously generated in the bone marrow throughout life.

On the basis of our discussions about the model, Mary wanted to test two main tenets of the model (Table II). The first was whether the B-cell repertoire was maintained by self-renewing B cells. And the second was whether microbial antigens are required for development of the primary antibody repertoire by somatic diversification of the Ig genes.

At this point, Mary had to make a decision. She was equally interested in both of these issues. If the B-cell repertoire was maintained by self-renewing B cells, then she could easily test this by determining whether new B cells were generated throughout life. Although she was certain that this experiment would work, she also had a strong desire to test whether microbial antigens were required for development of the primary antibody repertoire. I encouraged her to tackle both projects even though I knew that to test the need for microbial antigens, she would have to develop germfree rabbits- and that would be technically difficult.

As it turned out, she did both projects and obtained some interesting results. In terms of whether the B-cell repertoire was maintained by self-renewing B cells or whether B cell precursors were being generated in bone marrow throughout the lifetime of the rabbit, she searched for evidence of ongoing VD and DJ gene rearrangements. During VDJ gene rearrangements, the DNA between V_H and D and D and J_H is deleted and forms a circle with signal joints (17). By making oligomers specific for the regions on either side of the signal joint we were able to PCR-amplify across the joints as a way of searching for these circles. So we isolated DNA from bone marrow of young and adult rabbits and PCR-amplified the DNA by using the appropriate oligomers. Analysis of the amplified product on polyacrylamide gels showed intense bands, both VD and DJ, from young rabbits but almost none from DNA of adult rabbits (Figure 6). Thus it appears that very few, if any, VD and DJ gene rearrangements occur in bone marrow of adult rabbits. We concluded that little if any B lymphopoiesis occurs in adult rabbits.

If new B cells are not being generated in adult rabbits, then presumably the B cells must be long-lived and/or self-renewing. Because in mouse, CD5⁺ B cells are self-renewing, the finding that all rabbit B cells are CD5⁺ fits with the possibility that they are self-renewing.

Meanwhile Mary was also trying to determine whether microbial antigens were associated with the somatic diversification that occurred between 4 and 8 weeks of age or whether the somatic diversification was simply developmentally regulated. Mary began to develop germfree rabbits and, as expected, it was a time-consuming and demanding project. We encountered considerable difficulty in raising germfree rabbits that survived to the necessary 6 to 8 weeks of age. However, this project was fruitful because of the control rabbit. This control rabbit was delivered sterilely by cesarian section and was maintained in the laboratory, without being exposed to the microbial flora of our normal rabbit colony. When this rabbit was three months of age, we PCR-amplified the VDJ genes from PBL of this rabbit, and to our surprise, found that they had undergone only limited somatic diversification. This observation indicated that environmental factors, maybe microbial antigens, were important for somatic diversification of the VDJ genes.

So, germfree rabbits did not appear to be necessary after all. Somatic diversification of the VDJ genes is not developmentally regulated; instead environmental factors, presumably microbial flora, are the stimuli for the somatic diversification. We are currently continuing this project to determine which environmental factors stimulate somatic diversification.

The model for B-cell development also predicted that B cells migrate to the GALT and here undergo somatic diversification. Rose Mage and her colleagues at the NIH showed that indeed the VDJ genes undergo gene conversion in the appendix of 6 week old rabbits (18). We wanted to test whether GALT was essential for this diversification. Now I needed to call on the special talents of Dr. Periannan Sethupathi, a gifted surgeon who has worked with me for several years. He developed what we refer to as GALT-less rabbits. He surgically removed the appendix and sacculus rotundus on the day of birth and the Peyer's patches at 3 weeks of age. We then PCR-amplified the VDJ genes from PBL of these GALT-less rabbits and determined their nucleotide sequences (19). By 10 weeks of age the VDJ genes of normal rabbits were extensively diversified (Figure 7); however, the VDJ genes of the GALT-less rabbits underwent much less somatic diversification. Thus we concluded that GALT is essential for generating the early or primary antibody repertoire.

To date, we have tested several aspects of our model for B-cell development. We know that the VDJ genes of all B cells undergo somatic diversification by approximately 8 weeks of age and that GALT is essential for this diversification. It appears that somatic diversification is somehow initiated by or dependent on environmental factors, but we have not yet identified these factors. It also appears that little if any VDJ gene rearrangement occurs in bone marrow of adult rabbits, suggesting that the B cells that develop early in ontogeny are maintained throughout life and that little if any B lymphopoiesis occurs in adults.

Clearly this model of B-cell development is very general, and many questions remain to be answered, including what factors are responsible for the B cells migrating to GALT soon after birth, how are B cells signalled to proliferate and to somatically diversify the VDJ genes, and what is the molecular and cellular basis for somatic gene conversion?

These questions will keep us busy for some years to come and I look forward with pleasure to continuing these studies with my many co-workers.

Looking back over my experience in experimental science I can see how it has been shaped by the various relationships I’ve had. And this has led me to think about what’s ahead for us, especially for our young scientists.

If we look at our field from an overall viewpoint, it is clear that we are witnessing one of those epochal moments in history. Experimental science is undergoing a sweeping change both from within and by external influences that exert change from without. And these changes affect the very nature of how science is done. First, there are more scientists being educated, and with new technologies continuously emerging, nature is revealing her reality at an unprecedented rate. While in many ways this change is very positive, there are potentially harmful aspects that I believe demand attention.

In the past 20 years we have attracted enormous funding, both public and private. Yet with more scientists at work, with more good ideas, combined with a finite national budget for science, that funding is stretched thin. This situation has created a highly competitive environment in which we all compete for the same dollars.

More recently we're seeing the effects of a new player in our field- that is, a second source of funding from private investment of capital funds. These for-profit funds buy ownership interest in the scientific world. As we develop more and more valuable reagents and new technologies, the lure of literally capitalizing on our discoveries by means of venture capital is having a profound impact on the direction that work in our labs takes-- on both the problems we investigate and the manner of the investigation.

In and of themselves the issues associated with funding in an academic setting are neutral- there is nothing essentially good or bad about them. They are, simply, a fact of life. However these funding issues bring with them certain pressures that uniquely have the potential to unknowingly alter or affect how we do science and how we regulate our scientific process. And how we respond to these pressures will have major beneficial or detrimental effects on our science. The pressure to produce big, positive results can lead to premature publication, secrecy and, in the worst case, to falsified data. The pressure related to competition for funding may result in diminished creativity. For example, an investigator may choose to take a safe approach to solving a research problem that will ensure publishable data, but may not advance our basic knowledge of nature, rather than taking a solely academic investigative approach that may lead to a new scientific paradigm.

Another major source of erosion to the scientific process is secrecy. In effect, secrecy disavows all of the ideals of experimental science, including openness, accessibility and sharing. Secrecy can be the result of competition among scientists for grants, jobs, and patent protection of new ideas. In a recent survey of academic life scientists (20), nearly 20% of those who responded indicated that in the last 3 years they had delayed publication of data by more than 6 months to protect a lead or to allow time to develop patent applications and licensing arrangements, among other reasons. Nearly 9% admitted to not sharing data or materials at all; while 34% indicated they had been denied access to research results or products produced by other academic scientists.

Venture capital brings with it an insidious type of corrosive secrecy. As Steve Rosenberg noted (21), "Secrecy about methods and results has become a common and accepted practice." He goes on to say that, while for-profit biotechnology companies have provided new sources of funding, they bring with them the ethical and operational rules of business rather than those of science. Together, the pressures of funding and the influence of commerce threaten to erode the ideals that regulate how experimental science is conducted.

Keep in mind that there are two essential components to the process of science. One is actually doing it and the other is regulating how we do it on the basis of our ideals and values. We are in the midst of profound changes in our working environment, and the pressures related to these changes have the potential to alter the ideals that regulate our work. So the question is, how do we take care of ourselves now and in ways that do not betray the future to which we aspire? That is, how do we continue doing science without changing our time-honored values?

At this point the answer is far from clear- and it could be argued that there may not be an answer. Further, some may argue that there is no need for an answer because there is no problem. This view would assert that any change in values would be the outcome of an adaptive process and that this would even be of positive benefit for science. But I don’t share that view. I am convinced that these changes present a clear and present danger and that if not responded to in an appropriate manner, they will have an inescapably harmful effect on our science.

And who will control our future? Although it may appear that outside forces are the source of the problem, in fact the issue is the choices we make. For example, if I develop a new reagent that will benefit the scientific enterprise, I am the one who will choose whether to share it with the scientific community or to use it for my personal advancement. The choices we as scientists make will regulate how science is performed in the future.

While there is no single antidote to the potential changes threatening our value system, I’d like to propose one part of the solution and it’s one we already possess, namely, the influential reality of scientific relationships. One of the assertions of this address is that the practice of science is made possible by the deep intensive relationships that are the bedrock and contextual web of all experimental science. There are of course several different types of scientific relationships, including collegial relationships with our peers, and mentoring relationships with our students and post-doctoral fellows. I’d like to focus on the mentoring relationship.

All of our training programs are organized by the goal of enabling the student to develop a scientific minda mind that fully participates in the scientific process. That scientific mind is at the core of our identity as scientists; it is how we understand, organize and respond to our world. Certain characteristics of the scientific mind are familiar to all of us. Students will be able to assimilate data, ask meaningful questions, develop hypotheses, design appropriate experiments to test the hypotheses, and interpret the data critically. Another facet, and perhaps the most important one in terms of our future, pertains to the ideals and values that regulate a scientist’s work. By this I mean a thoughtful thoroughness that is founded on a deep commitment to learn nature’s reality, whatever it is, and just as importantly to be regulated by the precept of sharing that knowledge openly and without restriction.

The uniqueness of each student influences this process. There is not a straight forward set of guidelines that will be effective for every student. The primary principle is that the mentoring relationship is a type of caregiving relationship. The mentor relates to the student in a manner that facilitates the student’s choosing and pursuing her/his own motives. We often assume that the motives of the students are the same as ours and we mentor as we would wish ourselves mentored. Instead, I suggest an ideal where the goal is to understand the student’s motives and then respond accordingly to those motives, whether or not they are the same as ones own. For example, within the context of ongoing projects in a laboratory, the mentor will help the student choose to work on a project that the student is interested in and one that will facilitate the student’s growth and this may be different from the project that the mentor would choose for the student.

Students need to have the experience of making their own decisions and learning from them- no matter the results. To facilitate this experience, mentors must be continuously aware of the choices students make and help them understand and cope with problems or setbacks as they arise. Take, for example, a student who reaches a point in her work where her data suggest two different approaches to a problem. Although the mentor may know that scientifically one choice is better than the other, the mentor may decide to let the student make the less than optimal choice. Depending on the point in the student’s development, if the students choice results in a failed experiment, the mentor’s caregiving response can become an invaluable learning opportunity that enables the student to make more informed choices in the future.

The risk of expecting our students to publish frequently is that they become technically proficient but may not develop the level of creativity or knowledge of the scientific process that will allow them to become independent investigators at a level that will continue our scientific tradition. This pressure to publish often denies students the freedom to do what-if experiments based solely on their gut instincts. Instead, students and mentors often think that each and every experiment needs to be designed so as to give an interpretable result, regardless of its significance. The benefit to students in having the freedom to explore seemingly far-fetched ideas includes the satisfaction and excitement of doing science purely for the sake of generating new knowledge. The benefit to science is that some of these highly speculative ideas may lead to dramatic new scientific insights.

Students are getting a strong message that they need to publish often because if they have more publications, theyll get a better job. Those of us who hire scientists need to be careful not to unknowingly support the view that the greater the productivity during the training period, the better the scientist. For example, when we look at CVs, rather than focusing on how many papers they’ve published, why not consider whether they developed their own scientific mind, i.e., have they learned how to do science; and does the way they do science reflect the sound principles that are inbred into a well trained scientific mind; are they creative, and how well do they communicate their science?

By working with students over a period of years you have an opportunity to get to know how their minds work. And for them, it’s an opportunity to know and increasingly trust our minds. Through this mutualizing process, a scientific mind emerges. And it is through this relationship that mentors can ensure development of a scientific mind that is regulated by the highest ideals and values. This mutualizing process also affords the mentor the opportunity to know the motives of the students. By this mechanism we can often identify those students who, under pressure or otherwise, would consider distorting or falsifying data for the purpose of financial rewards or for advancing their careeror who would simply resort to short cuts and to accepting inaccuracies in their haste to publish.