Modulation of Th1/Th2 subsets by granulocyte-colony stimulating factor

The article by Sloand et al1-1 that was published in the April 1 issue of Blood suggests a direct regulatory role of granulocyte-colony stimulating factor (G-CSF) over T cells. The authors show a shift towards Th2, or interleukin 4 (IL-4) production, of CD4⁺ T lymphocytes from peripheral blood mononuclear cells (PBMCs) obtained from stem-cell donors treated with G-CSF (G-PBMC) on the basis of intracellular detection of IL-4 and interferon (IFN)-γ by flow cytometry. Detection of intracellular lymphokines by flow cytometry requires controls usually not required for detection of surface molecules1-2 such as competition of the specific staining with the unlabeled antibody, since isotype control monoclonal antibody (mAb) does not perform well as a negative control for intracellular staining as it does for surface staining.1-2 

To address the same question raised by our colleagues, we recently performed studies on G-PBMC T cells. Using the same approach as the one used by Sloand et al,1-1 our results were initially very similar to those published by Sloand's group (Figure 1-1A, left panels). However, we were surprised by the fact that the frequency of IL-4⁺ cells plus IFN-γ⁺ cells added up to almost 100%, a result very different from that obtained using healthy volunteers.1,2In fact, data in the literature estimate the mean relative number of IL-4–producing cells to be about 4% to 5% and IFN-γ– producing cells to be about 25% to 30% of T cells. In addition, the staining pattern we obtained was not repetitive and did not allow us to set, with confidence, quadrants for statistical analysis.

In order to understand the reasons for these discrepancies, we used 2 different techniques to stain PBMCs. When the cells are fixed with paraformaldehyde (PF) for 30 minutes at 4°C,1-6 the whole cell population can be considered positive for either lymphokine, and the staining is only marginally inhibited by a 25-fold excess of unlabeled antibody (Figure1-1A, left panels). On the other hand, if the cells are fixed with PF for 5 minutes at 37°C, the percentage of positive cells agrees with the literature (Figure 1-1A, right panels), and the staining for both IL-4 and IFN-γ can be specifically inhibited by the matched unlabeled antibodies (Figure 1-1A, right panels). This apparently minor technical difference generates data that can mislead the interpretation of the regulatory effects of G-CSF over the Th1/Th2 phenotype.

As shown in Figure 1-1B, a long fixation time (30 minutes) at 4°C suggests an augmentation of both IL-4– and IFN-γ–producing cells after in vivo G-CSF treatment, whereas a short fixation time (5 minutes) at 37°C suggests exactly the opposite; that is, any inhibition on both cell types. Since the former technique cannot be confirmed with the specific controls (Figure 1-1A), we conclude that only the latter should be used. In the present situation, if an interpretation can be drawn from our preliminary data, it would be that G-CSF plays an inhibitory role over the frequency of IL-4– and IFN-γ–producing cells. More importantly, these results alert us to the significance of showing the specificity controls when performing intracellular staining for cytokines, in order to avoid misinterpretations based on technical artifacts.

In a manuscript recently published inBlood,2-1 we demonstrated that lymphocytes stimulated with either CD3 monoclonal antibody (mAb) or phytohemagglutinin (PHA) in the presence of granulocyte-colony stimulating factor (G-CSF) altered the cytokine expression, resulting in a shift in cytokine response from a Th1 to a Th2 pattern. Lymphocytes obtained from healthy individuals following G-CSF mobilization for the purpose of stem cell donation also showed shifted patterns of cytokine expression following coculture with either CD3 mAb or PHA.

We certainly agree with Vasconelos et al that appropriate controls are of paramount importance when performing intracellular staining. In order to ensure that our staining methods were sensitive and specific, we (1) optimized the amount of antibody that was added so as to decrease nonspecific staining, (2) performed neutralizing studies with unlabeled interferon-γ (IFN-γ) and IFN-γ mAb, (3) checked concordance between intracellular staining using antibodies from different manufacturers and by enzyme-linked immunosorbent assay (ELISA) measurements on duplicate samples, (4) measured IFN-γ and interleukin 4 (IL-4) using 2-color staining to ensure that the cell populations did not overlap extensively, and (5) used appropriate isotypic controls.

We found that the addition of excessive anticytokine antibody to the lymphocytes resulted in significant nonspecific staining, particularly when using IL-4 mAb. Results from titration experiments were used to optimize antibody concentrations. In addition, we performed multiple experiments using stimulated and unstimulated lymphocytes from healthy donors: PHA stimulation of cultured cells resulted in a mean of 35% of cells staining with IFN-γ with IFN-γ–fluorescein isothiocyanate (FITC) with a mean channel fluorescence (MCF) of 5.6, whereas isotype controls resulted in an MCF of less than 0.10. Lymphocytes cultured in the presence of IFN-γ or IL-10—stimuli expected to favor development of the Th1 phenotype—stained primarily with IFN-γ, and lymphocytes cocultivated with IL-4 or IL-12—stimuli that favor development of Th2 cells—stained primarily with IL-4. Preincubation of anti–IFN-γ mAb with unlabeled IFN-γ competitively eliminated the staining, as did preincubation of cells with unlabeled IFN-γ mAb; unlabeled irrelevant mAb had no effect. The failure of Vasconelos et al to observe neutralization with the permeabilization protocol recommended by the manufacturer of the IFN-γ mAb may be related to use of a cold IFN-γ mAb directed at a different epitope than the labeled IFN-γ mAb, or to excess labeled IFN-γ mAb in the sample. To assess the reproducibility of intracellular staining, we also tested another antibody from a different manufacturer (BioSource, Camarillo, CA; Pharmingen, San Diego, CA). Results obtained from comparison of 26 sets of samples showed a highly significant correlation (Pearson correlation coefficient = 0.96). In addition, IL-4 and IFN-γ ELISA measurements of lysed cell samples correlated with the results of intracellular cytokine staining. Measurement of IL-4 and IFN-γ on CD4 cells using 3-color staining showed no overlap between CD4 cells staining with each antibody.

We performed appropriate isotypic controls using labeled antibodies provided by the manufacturer of the cytokine mAbs and as described in Current Protocols in Immunology.2-2Although there are inherent problems in using isotypic controls, they still provide great advantages over unstained cells for negative gating cells. In none of the papers referenced by Vasconelos et al were unstained cells used as controls, and more recent articles by Gurunathan et al2-3 utilize the same isotypic controls as we report in our study.

When we applied the staining method suggested by Vasconcelos et al (a short fixation period at 37°C), we found significant nonspecific uptake of stain in the isotypic control as well as in unstimulated normal cells. When appropriately gated using an isotypic control also prepared in this fashion, there appeared to be no difference in the amount of uptake of stain by the sample when compared to the other fixation method. We believe that the comparison of stained to unstained cells (Figure 1 of the letter by Vasconcelos et al) may therefore not be valid.

With respect to the comments made by Sivakumaran et al, extrapolation of in vitro data for clinical purposes must always be made cautiously. There is no evidence that administration of G-CSF is detrimental in immunosuppressed patients—in fact, quite to the contrary, as multiple trials involving neutropenic HIV-infected patients show that G-CSF consistently decreased the frequency of serious infection.2-4 2-5 Short courses of G-CSF are especially unlikely to influence tumor progression or chronic infections. Nevertheless, as we describe in the “Discussion” segment of our paper, long-term G-CSF therapy may affect immune processes like graft-versus-host disease or autoimmune marrow destruction.