Download The hematopoietic system has long served as an important model

Cover Letter
Re: A New Method of Mining Developmentally Regulated Genes Identifies a Branchpoint
Between B and T Cell Development
Debashis Sahoo*, Matthew A. Inlay*, Jun Seita, Holger Karsunky, Thomas Serwold,
Sylvia K. Plevritis, Irving L. Weissman, Deepta Bhattacharya, and David L. Dill
*
These authors contributed equally to this work
Dear Editor,
Please find accompanying this letter a copy of our manuscript entitled, "A New
Method of Mining Developmentally Regulated Genes Identifies a Branchpoint Between B
and T Cell Development," that we submit for consideration for publication as an article in
Nature. In this manuscript we present a novel computational method, MiDReG (Mining
Developmentally Regulated Genes), which enables the identification of developmentally
regulated genes using publicly available microarray data. Differentiation from precursor
cells to mature progeny typically proceeds through developmental intermediates, where a
network of gene expression changes gradually establishes the identity of the mature
lineage and extinguishes the potential for alternative lineages. Understanding the
molecular mechanisms of lineage commitment critically depends upon the ability to
identify and isolate cells at intermediate stages of development; however, identification
of these intermediates is challenging. MiDReG enables logical predictions of genes that
might define such intermediates. MiDReG bases its predictions on if-then gene
expression relationships (Boolean implications) that hold consistently over thousands of
microarrays from hundreds of different laboratories, mined using a technique we recently
developed. Here, we demonstrate the utility of MiDReG for identifying new genes
functionally important to B cell biology. In addition, we applied MiDReG to an existing
question in lymphoid development, and discover a surface marker, Ly6d, that identifies
the branchpoint between B cell and T cell development.
The major specific findings of this study are as follows:
1. MiDReG was validated using a well-characterized system, B cell development.
MiDReG predicted 19 genes to be upregulated during the transition from KIT+
hematopoietic stem/progenitors to CD19+ B cells. By quantitative RT-PCR, we
show that 71% of analyzed genes were upregulated during this transition. We
expanded our list of predicted B cell genes to 62 through the combination of
multiple genes. Literature investigation revealed that 36 of these genes have been
knocked out in mouse, of which 19 show specific defects in B cell function and
differentiation. Therefore, MiDReG reliably predicts novel genes that are
functionally related in B cell development.
2. MiDReG identified Ly6d, a surface marker which separates bone marrow-derived
common lymphoid progenitors (CLP) into two distinct fractions. Ly6d- CLP are
unbiased for either B or T cell differentiation, but are lymphoid committed,
whereas Ly6d+ CLP are almost completely B cell committed. Therefore, Ly6d
expression marks the first step in B cell specification, and is the earliest marker
for B cell development identified to date. The Ly6d- CLP lies at a developmental
checkpoint immediately upstream of lineage commitment to the B, T, NK, or DC
cell fates, identifying it as a uniquely important intermediate in hematopoiesis.
We feel that our manuscript will be of great general interest to the readership of Nature.
MiDReG is a powerful tool to identify novel genes and cellular stages in development,
and is applicable to any developmental pathway where some knowledge of the genes
expressed at the beginning and end stages are known. Importantly, in our analysis no
intermediate B cell populations were represented in the microarrays. To identify these
genes using a traditional array based approach would have required careful sorting of
each of the intermediate populations and thousands of dollars in microarrays. MiDReG
uses microarray data that has already been generated and is freely available to predict
genes that change during developmental progressions. Furthermore, we have used
MiDReG to address an important biological question that had not previously been
answered using traditional approaches. As a result, we have isolated the population
making the T vs. B cell fate decision, which will now allow the careful dissection of the
molecular mechanisms involved in that decision.
We are aware that a controversy exists as to the nature of the thymic seeding population,
and that many groups claim that a myeloid/lymphoid progenitor (LMPP) is that
population. Just this month at the AAI conference, nearly every talk that presented a
hematopoietic tree had CLP producing only B cells. Our data clearly demonstrate that
Ly6d- CLP produce T cells in the thymus earlier than the upstream LMPP. While we feel
that the introduction of Ly6d will improve labs’ abilities to isolate this population, and is
therefore an important contribution toward resolving this field, the major point of this
paper is to introduce MiDReG, a method which is applicable to all of developmental
biology. Ideally, this manuscript should be critiqued by reviewers who are familiar with
both computational theory and developmental biology, rather than by those only
interested in this manuscript’s relevance to the CLP/MPP controversy. As potential
reviewers, we recommend Andrea Califano (Columbia University Medical Center), Ihor
Lemischka (Mount Sinai School of Medicine), Hans-Reimer Rodewald (Ulm Institute for
Immunology) and Cornelis Murre (University of California, San Diego). We respectfully
request to exclude David Allman and his collaborators Avinash Bhandoola and Richard
Hardy, due to potential conflicts of interest.
Because this manuscript spans both Computational Biology and Developmental
Immunology, we would like to have two corresponding authors listed: David Dill
(Computational Biology) and Matthew Inlay (Immunology).
Thank you for your consideration,
David L. Dill and Irving L. Weissman