| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Published online before print June 16, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Laboratory of Human Bacterial Pathogenesis, Rocky Mountain Laboratories, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Hamilton, Montana; and
Laboratory of Host Defenses, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, Maryland
1Correspondence: Laboratory of Human Bacterial Pathogenesis, Rocky Mountain Laboratories, National Institute of Allergy and Infectious Diseases, National Institutes of Health, 903 S. 4th St., Hamilton, MT 59840. E-mail: fdeLeo{at}niaid.nih.gov
Many of the points made by Robey are intuitive and well understood by those performing mainstream research. First, the presence of a transcript and its abundance may not always correlate with the amount of protein generated from that transcript. Second, the lack of change in transcript level or the magnitude of a given change is not necessarily a measure of the magnitude or physiologic importance of a specific change. Third, there is investigator bias in focusing on specific patterns of transcription changes. However, as long as statistics support the changes as significant, this may not be viewed as bias but rather part of the creative, intuitive process that allows formulation of testable hypotheses. These points are widely known by investigators performing genome-wide expression analyses and thus, are part of the rationale for performing functional studies to complement and/or validate hypotheses developed from transcription data.
Even recognizing these limitations to transcription pattern analyses, it is still possible to identify cellular processes (such as apoptosis), which are at least partly mediated by a broad change in pattern at the transcriptional level [1 ]. In this regard, the process of apoptosis is much more likely to satisfy criteria of a transcriptional differentiation program (i.e., associated with broad changes) than something occurring rapidly at the enzymatic level (e.g., oxidative burst, degranulation, etc.), and this idea is substantiated by the findings of Kobayashi et al. [2 ]. As Robey correctly notes, investigating the functional relevance of the change for each individual transcript is not feasible for a single study. However, it is possible to test the functional significance of selected gene-regulated pathways, as was done in our studies [2 ]. The general concerns raised by Robey (the so-called "pitfalls") are well recognized by investigators who use gene profiling as a tool to gain insight into cellular processes.
Importantly, Robey’s Letter contains factual errors, which undermine the relevance of his criticisms. His indication that we suggested decreased glucose (Glc) phosphorylating capacity is coupled to increased glycolysis is simply not true. There is no mention of decreased Glc phosphorylating capacity nor is it implied. We reported decreased expression of the gene encoding hexokinase-3 (HK3), which clearly did not affect Glc phosphorylating activity needed by human polymorphonuclear leukocytes (PMNs) during apoptosis, as the decrease in HK3 expression was accompanied by glycolysis. We stated clearly that the finding is counterintuitive to our hypothesis, which was put forth based on the expression pattern of other genes involved in glycolysis and the pentose phosphate pathway. Importantly, HK3 was still expressed, albeit at lower abundance. Inasmuch as this finding seemed at variance with the observation that other genes encoding pentose phosphate pathway enzymes were up-regulated and the functional data indicating increased glycolysis, we did indeed examine relative transcript levels for all HK isoforms represented on the Affymetrix Hu95Av2 oligonucleotide microarray. There are three known human HK isoforms represented on the Hu95Av2 array: HK3 (GB U51333), HK2 (GB Z46376), and glucokinase/HK4 (GB M90299). We analyzed each and found that HK3 was expressed most abundantly in human PMNs. HK2 was called "absent" [2 ] in at least one of the three individuals profiled and was expressed at a relatively low level. Even so, when called "present", HK2 was down-regulated 1.9-fold in activated PMNs compared with unstimulated cells, consistent with the findings for HK3. We determined that HK4 was not expressed in human PMNs. HK1 (GB M75126) is represented on the Affymetrix Hu133A GeneChip and was called absent in human PMNs at most of the times post-phagocytosis. As with other genes, which encode glycolytic pathway enzymes that were not up- or down-regulated, we would have indicated approximate transcript levels for HK1 and HK2 if they were on the array and clearly present in human PMNs in our assay [2 ]. Therefore, consistent with Robey’s suggestion, we included the corresponding expression data for all relevant species within the figures.
Although it is true that HK1 predominates in most blood cells, as Robey indicates, it is crucial to realize that granulocytes are the single exception [3
]. Granulocytes contain 70–80% HK3, and the remaining HK is predominantly HK1 (20–30%) [3
]. The report by Rijksen et al. [3
] concurs very well with our gene expression data, which indicate that HK3 is indeed the most highly expressed isoform in human PMNs. Thus, based on the inaccurate assumption that HK isoform expression is similar in all leukocytes, Robey proposed that "functionally redundant isoforms" might explain our findings. The hypothesis is interesting but unlikely, given that HK3 is the predominant HK isoform in human PMNs and that genes encoding HK1 and HK2 are down-regulated or expressed at very low levels in neutrophils after phagocytosis. In the remainder of the Letter, Robey presents new ideas and speculation, which are also interesting and offer alternative explanations that could very well be true. However, in his concluding paragraph, Robey comments on microarray-based investigation with now-antedated criticisms. Most significantly, Kobayashi et al. [2
] identified gene-regulated pathways that were confirmed by several functional assays (
-glutamyltransferase activity, cell glutathione levels, Glc uptake, lactate production, and effects of Glc on apoptosis). These data are sufficient to support the major functional inferences.
In summary, the patterns of gene transcription and/or expression during PMN apoptosis reported by Kobayashi et al. [2 ] provide very broad insight into the metabolic status of the cell at a crucial point in time. Importantly the studies identified an apoptosis-differentiation program, which is not explained by the abundance or change in abundance of any single transcript but rather the coordinated regulation of gene expression in the cell as a whole. The recent discovery that glycolysis is controlled at the level of transcription in human myeloid cells [4 ] provides strong support for our findings. The new concepts and ideas put forth as a result of functional genomics, such as the PMN apoptosis-differentiation program, are an important step forward in all areas of contemporary biomedical research.
Received May 17, 2003; accepted May 17, 2003.
REFERENCES
Related Article
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
