Cyclic neutropenia (CN) and severe congenital neutropenia (SCN) are autosomal dominant conditions characterized by either regular oscillations in absolute neutrophil count (ANC) or sustained neutropenia, respectively (Berliner, et al 2004). Mutations in ELA2, which encodes neutrophil elastase (NE), are found in patients with either disorder (Berliner, et al 2004). Recent studies implicate mis-localized NE as triggering the unfolded protein response (UPR) and accelerated apoptosis (Kollner, et al 2006); however, ELA2 mutations in CN may also produce a negative feedback loop that leads to periodic impairment of neutrophil differentiation and production (the chalone hypothesis). Mutations in other genes have been identified in SCN (Berliner, et al 2004). One such gene is GFI1. Gfi1−/− mice have a hypocellular thymus, and reduced lymphoid progenitors(Karsunky, et al 2002), and both Gfi1−/− and Gfi1N382S mutant mice have defective granulopoiesis and static neutropenia (Zarebski, et al 2008). GFI1-mediated transcriptional repression of ELA2 was discovered in 2 SCN patients each with single mutations in GFI1 (N382S and K403R) whose granulocytes showed increased NE expression (Person, et al 2003); however, there are no reported cases of GFI1 mutations being found in patients with CN. Aberrant expression of serine proteases such as NE can also predispose myeloid cells to attack by CD8 cytotoxic T lymphocytes (CTL). CTL that are specific for the HLA-A2-restricted PR1 peptide, derived from both NE and proteinase 3 (PRTN3), are undetectable (< 1/100,000 CD8+ T cells) by standard peptide/MHC tetramer staining of peripheral blood lymphocytes in healthy donors (Molldrem, et al 1999), but are increased in leukaemia patients and induce specific lysis of neutrophil precursors and leukaemia cells that aberrantly express PRTN3 and NE (Molldrem, et al 2000). Because NE is aberrantly expressed in granulocytes of CN and SCN patients, we reasoned that PR1-CTL might be increased in the HLA-A2+ patients. We have evaluated a 25 year old, HLA-A2+, man, who was diagnosed with CN after he developed frequent infections during childhood and was observed to have an absolute neutrophil count (ANC) ranging between 0.1 and 6.0 ×109 cells/L over an 18–21 day period. He has been treated with G-CSF every other day, which has shortened his neutropenic cycles to 15–18 days and reduced the frequency of skin and pulmonary infections. He and his parents gave informed consent for obtaining peripheral blood in accordance with the M.D. Anderson Cancer Center Institutional Review Board. To look for abnormal protease expression, we examined peripheral blood neutrophils from the patient and an age-matched healthy control by Wright stain and confocal microscopy. In control granulocytes NE is predominantly confined to primary granules, whereas in the patient NE is predominantly localized outside primary granules near or in the nucleus (Figure 1A–H). Fig 1 Confocal microscopy reveals aberrant NE localization in patient granulocytes. Granulocytes were permeabilized and labeled with directly conjugated antibodies to myeloperoxidase (MPO) (Alexa-555) and NE (Alexa-647) for 30 min at 4°C. The cells ... Although mutations in ELA2 and associated regulatory genes in SCN lead to altered NE expression with an unfolded protein response (UPR), DNA from the coding regions of ELA2, the beta subunit of AP3, and PRTN3 were not mutated in the patient or his parents; however, one allele of GFI1 contained two de-novo mutations at transcript base 1145 (AAC→AGC = N382S) and transcript bases 1209 and 1210 (AAG→AGA = K403R) that had each been previously described in two different SCN patients (Person, et al 2003). These mutations were found in neither parent, and the patient has no siblings. While defective lymphocyte proliferation was observed in SCN patients with single GFI1 mutations we found no defect in lymphocyte proliferation after T cell receptor (TCR) cross-linking in this patient (Figure 1I) (Person, et al 2003). However, T cell development appears abnormal since more naive T cells (75% vs. 32%, P=0.0025) and fewer central memory T cells (9% vs. 42%, p = 0.0233) were seen in the patient compared to age-matched controls (Figure 1J). We sought to determine whether PR1-CTL were detectable in this patient with aberrant NE expression since he is HLA-A2+. Over a single 15-day cycle of neutropenia, the percentage of PR1-CTL was increased compared to HLA-A2+ healthy subjects, and both parents who lacked detectable PR1-CTL. The PR1-CTL also showed cycling that lagged neutrophil cycling by 4–10 days (Figure 2A–C). Although patient lymphocytes were defective in IFN-γ or TNF-α secretion in response to TCR cross-linking (Figure 2D), they did produce IFN-γ, and to a lesser extent TNF-α, following stimulation with PR1. Fig 2 PR1-specific CD8+ T cells vary in number during a neutropenic cycle. 1 × 106 patient PBMC were labelled with aqua live/dead dye (Invitrogen, Carlsbad CA), CD4 PE-Cy5, CD14 PE-Cy5, CD16 PE-Cy5, CD19 PE-Cy5, CD8 FITC, and PE-conjugated PR1/HLA-A2 ... Studies in this CN patient who possesses a double de novo GFI1 mutation demonstrate specific T lymphocyte immunity to the myeloid-restricted antigens PRTN3 and NE. Autoimmune neutropenia is well described, and cases of adult onset cyclic neutropenia in patients with immune dysregulation (large granular lymphocytes) have been reported (Berliner, et al 2004, Loughran and Hammond 1986). Therefore, multiple non-exclusive mechanisms (e.g. UPR, Csf1/Csf1r derepression, NE potentiated self-repression, auto-immunity) can result in neutropenia syndromes (Kollner, et al 2006, Loughran and Hammond 1986, Salipante, et al 2009, Zarebski, et al 2008). The studies in this patient suggest that in addition to UPR-induced apoptosis of granulocyte precursors due to aberrantly expressed NE, there might be an additional mechanism of granulocyte-mediated autoimmunity caused by PR1-CTL in some CN patients. An autoimmune mechanism that leads to regular oscillations of neutropenia can be mathematically modeled using a series of ordinary differential equations in a manner analogous to that used in modeling “predator-prey” simulations or oscillating chemical reactions (supplemental information) (Schnakenberg 1979). In this model, aberrantly expressed NE results in PR1 peptide being presented by HLA-A2 on the cell surface. The number of PR1-CTL subsequently increases and after sufficient granulocyte progenitors are eliminated, the number of PR1-CTL contracts, allowing recovery of granulocyte progenitors. This process has similar kinetics to those of a generalized immune reaction. This hypothesized mechanism is consistent with the observed data in the patient we have investigated, but does not exclude other proposed mechanisms of neutropenia in other patients.