Sirs

The February 2002 issue of Nature Reviews Immunology contained an Opinion article by Yoshimoto Katsura entitled 'Redefinition of lymphoid progenitors'1. Although it is certainly appropriate for Dr Katsura to state his opinions in the field of his expertise, he has provided an incomplete representation of published data from our laboratory that he considers to be inconsistent with his favoured model of lymphopoiesis. We wish to respond to the information contained in this article1 regarding our findings.

All of Katsura's models of haematopoiesis are based on data that have been obtained from an in vitro fetal thymic-organ culture assay that his laboratory has designed to define retroactively the developmental potential of single fetal-liver haematopoietic progenitors2. As he mentioned, the strength of this system is that it is possible to 'read out' T-cell, B-cell and myeloid outcomes from a single input cell. However, it must be noted also that his assay system is not physiological. To obtain multi-lineage readouts from their in vitro assay, Katsura's group cultures the cells at high oxygen concentrations (70%) that are never attained in vivo. In addition to this concern, it is impossible to validate the results that are obtained from this assay system using in vivo reconstitution assays, because the developmental potential of each cell is characterized retroactively.

So far, Katsura's group has not detected a clonogenic T/B-restricted progenitor in the fetal liver, but it has provided evidence of all other possible outcomes1. Considerable care must be taken in defining the full developmental potential of an individual haematopoietic progenitor retroactively. It is well known that progenitors often give read-out patterns that indicate their developmental capacity only partially. So, we caution against making claims about the presence or absence of specific progenitors using this type of negative data. Despite this, Katsura claims that his data conflict directly with our data showing the existence of a common lymphoid progenitor (CLP) in adult mouse bone marrow. To support this claim, Katsura makes the untested assertion that, except for primitive haematopoiesis in early fetuses, the mechanisms of definitive haematopoiesis in fetal liver are similar to those in adult bone marrow. However, we have published data in the Journal of Immunology that show clear differences between fetal-liver CLPs and adult-bone-marrow CLPs. Specifically, both fetal-liver and adult-bone-marrow CLPs give rise to lymphoid cells in vivo, but only fetal-liver CLPs can give rise to macrophages also in vitro3. Surprisingly, Katsura neither discussed nor referenced these published findings. He extrapolates from his data that there are no CLPs in fetal liver to disprove the existence of CLPs in adult mouse bone marrow4.

Katsura also asserts that the conclusions drawn in our Cell paper4 describing the identification and characterization of adult-bone-marrow CLPs are not sound. He claims that the main problem with our paper is that the assay systems we use are designed to detect only T/B-type progenitors1. However, it is clear from this remark that Katsura has chosen to ignore most of the data in this paper4. Not only did we study the developmental potential of CLPs extensively in in vitro differentiation assays that support both myeloid and lymphoid development, but we also characterized their full developmental potential in in vivo reconstitution experiments4, unlike Katsura and his collegues. Therefore, Katsura's assertion that we failed to take into account the possible existence of bipotent myeloid/lymphoid progenitors in our CLP population is not true.

Katsura also stated that recent data that we published in Nature5 support his common myelo-lymphoid progenitor (CMLP) model of haematopoiesis. This brief statement leaves the reader with the erroneous impression that we have found that wild-type CLPs have myeloid developmental potential; that is not true. The Nature paper5 shows that CLPs transfected with cytokine receptors, such as human interleukin-2 receptor-b or human granulocyte–macrophage colony-stimulating factor receptor-ab, and stimulated with the cognate cytokine change their developmental fate from lymphopoiesis to myelomonocytopoiesis; this has nothing to do with their normal developmental fate.

We are surprised and disappointed by Katsura's misrepresentation of our data and his. What is particularly troubling about the discussion of our data in Katsura's article is that, at present, there are no irresolvable contradictions between our collective data.