Improving power while controlling the false discovery rate when only a subset of peptides are relevant

The standard proteomics database search strategy involves searching spectra against a peptide database and estimating the false discovery rate (FDR) of the resulting set of peptide-spectrum matches. One assumption of this protocol is that all the peptides in the database are relevant to the hypothesis being investigated. However, in settings where researchers are interested in a subset of peptides, alternative search and FDR control strategies are needed. Recently, two methods were proposed to address this problem: subset-search and all-sub. We show that both methods fail to control the FDR. For subset-search, this failure is due to the presence of “neighbor” peptides, which are defined as irrelevant peptides with a similar precursor mass and fragmentation spectrum as a relevant peptide. Not considering neighbors compromises the FDR estimate because a spectrum generated by an irrelevant peptide can incorrectly match well to a relevant peptide. Therefore, we have developed a new method, “filter then subsetneighbor search” (FSNS), that accounts for neighbor peptides. We show evidence that FSNS properly controls the FDR when neighbors are present and that FSNS outperforms group-FDR, the only other method able to control the FDR relative to a subset of relevant peptides.

Polyclonal PRAME-Specific Cytotoxic T Lymphocytes Generated Using Protein-Spanning Pools of Overlapping Pentadecapeptides Target Chronic Myeloid Leukemia

The cancer testis antigen PRAME is a potential target for adoptive T-cell or vaccine therapy of many hematologic malignancies and solid tumors. PRAME-specific cytotoxic T lymphocytes (PRAME-CTLs) can be detected in patients with hematologic malignancies and we have shown that they can be generated and expanded ex-vivo, using an optimized combination of artificial antigen presenting cells (aAPC) (K562 cell line genetically modified to express the HLA-A*02, CD80, CD40L and OX40L molecules) and cytokines (IL12, IL7 and IL15). Four HLA-A*02 PRAME-derived epitopes (P100, P142, P300 and P425) have previously been identified using a proteosome-mediated digestion analysis. However, this strategy, since relying only on the major cleavage site targeted by the immune-proteosome machinery for epitopes generation, may limit the potential clinical value of the identified peptides. We have now adopted an alternative method that uses a peptide-library consisting of 125 synthetic pentadecapeptides, overlapping by 11 aminoacids, spanning the entire PRAME protein. We evaluated whether novel HLA-A*02 restricted CD8+ T-cell responses to multiple immunogenic epitopes can be identified and used to consistently generate polyclonal PRAME-CTL lines from healthy donors and patients with hematologic malignancies. CD8+ T lymphocytes from 14 HLA-A*02 healthy donors and 3 patients with chronic myelogenenous leukemia (CML) were primed with autologous CD40L-activated B blasts loaded with the PRAME-peptide library in the presence of low doses of IL12, IL7 and IL15, and then expanded by weekly re-stimulation with peptide loaded aAPC and IL-2. The frequency and specificity of PRAME-CTLs were evaluated using IFNg Elispot and 51Cr release assays against PHA-blasts loaded with the PRAME-library. Using this approach we consistently generated PRAME-CTLs in 12 of the 14 HLA-A*02 healthy donors (526±101 SFC/105 cells as assessed by IFNg Elispot assay) compared to an irrelevant peptide-library (7±2 SFC/105). Similarly, PRAME-CTLs were generated from all 3 CML patients (441±250 SFC/105 cells vs 22±10 SFC/105 against an irrelevant peptide-library). These PRAME-CTLs were also able to target autologous tumor blasts (57±6 IFNg SFC/105), demonstrating that the same peptides were processed and presented physiologically. A Cr51 release assay confirmed that the PRAME-reactive T cells were cytotoxic, lysing autologous-PHA blasts loaded with the peptides derived from the PRAME-library (63±14% at a 20:1 E: T ratio), but not with irrelevant peptides (<15%). MHC class-I blocking experiments using specific antibodies showed that both IFNg release and cytotoxic activity were HLA-restricted. Using pentadecapeptides sub-pools, we found that the responses of our expanded PRAME-CTLs were polyclonal, since they consistently released IFNg in response to 1 to 6 pentadecapeptides pools (59% were specific for 1 or 2 pools, 25% to 3 pools, and 16% to 6 pools). Moreover, the approach we describe has allowed us to identify 6 potential new immunogenic 15-mer peptides that are processed and presented by tumor cells, and should facilitate expansion of polyclonal PRAME-CTLs for adoptive transfer or after vaccine administration to patients with PRAME+ hematological malignancy.

Preferentially Expressed Antigen of Melanoma (PRAME)-Specific Cytotoxic T-Lymphocytes (CTLs) and Transgenic T Cells To Target Chronic Myelogenous Leukemia (CML).

Following allogeneic hematopoietic stem cell (HSC) transplantation, a graft versus leukemia (GVL) effect mediated by donor T cells contributes to the maintenance of long-term remission. GVL is likely mediated not only by alloreactive T cells, but also by donor lymphocytes recognizing tumor associated antigens overexpressed by leukemic cells. Selective expansion of such tumor-specific CTLs could augment GVL without increasing the risk of graft versus host disease (GVHD). PRAME is a cancer testis antigen (CTA) that is over-expressed by many hematological malignancies, including CML. We therefore generated PRAME-specific CTLs and evaluated their activity against primary CML tumor cells. CD8+ cells from 9 HLA-A*02 healthy donors and 5 CML patients were primed with autologous CD40L-activated B blasts loaded with HLA-A*02 restricted PRAME-peptides in the presence of low concentrations of IL-12, IL-7 and IL-15. Cells were then re-stimulated once a week using an artificial antigen presenting cell line (aAPC), consisting of K562 cell line that was genetically modified to stably express HLA-A*02, CD80 and CD40L and loaded with the same PRAME-peptides. In 8 out of 9 HLA A*02 healthy donors, we expanded (22±7 fold) T cells specific for the ALY-PRAME peptide. Similarly, PRAME-specific CTLs could be expanded from 3 out of 5 CML pts. When assessed by Elispot assay, these CTLs released IFNγ in response to ALY-peptide (511±260 SFC/105 cells) compared to 30±10 IFNγ SFC/105 against an irrelevant peptide. They also lysed autologous-PHA blasts loaded with ALY (84%±12% specific 51Cr release at 20:1 E:T ratio), but not with an irrelevant peptide (<10%). Both IFNγ release and cytotoxic activity were blocked by MHC class I antibodies, indicating that these effects were MHC-restricted. PRAME-specific CTLs also recognized primary CML cells since they produced IFNγ (320±31 SFC/105) in response to CD33+ CML blasts overexpressing PRAME (as assessed by Real Time PCR) isolated from HLA-A*02+ CML patients. Since routine expansion of PRAME-specific CTLs for clinical trials of adoptive T-cell transfer may remain problematic, we next cloned the αβT cell receptor (TCR) from a CTL clone with high affinity for the ALY-peptide (621 IFNγ SCF/105 cells at 0.2 nM peptide). The α- and β-chains were cloned in the SFG retroviral vector using an IRES sequence, and the transgenic αβTCR was expressed in primary T lymphocytes. T cells expressing the transgenic PRAME-specific αβTCR produced IFNγ on exposure to PRAME+ CML blasts (448± 69 IFNγ SFC/105 cells) and killed PHA blasts loaded with the ALY-peptide (42%±7% at 20:1 E:T ratio). In conclusion, our data show that PRAME is a promising target antigen on CML blasts and that adoptive transfer of CTLs or of T cells expressing a transgenic αβTCR targeting this antigen may be of value for patients with the disease.

Generation and Expansion of PRAME-Specific Cytotoxic T-Lymphocytes for Adoptive T-Cell Therapy of Hematological Malignancies.

The cancer testis antigen PRAME is a potential target for T-cell based adoptive immunotherapy for many myeloid and lymphoid malignancies, since these cells frequently either overexpress the antigen constitutively, or following induction with demethylating agents. Although several HLA class I A*0201 PRAME-derived T cell epitopes have been previously identified, the ex-vivo generation of PRAME-specific cytotoxic T lymphocytes (CTLs) has proved to be a major challenge. We have now optimized a method that consistently and reproducibly generates peptide tumor-specific CTLs. CD8+ cells selected from peripheral blood mononuclear cells. HLA A*0201 donors were first primed with autologous CD40L-activated B blasts loaded with A2-restricted peptides in the presence of low doses of IL-2, IL-7 and IL-15. Following this initial priming, we re-stimulated the T cells with an artificial antigen presenting cell line (AAPC), consisting of the human chronic myelogenous leukemia cell line K562 genetically modified to stably express the HLA-A*0201 molecule and the CD80 co-stimulatory molecule (K562/A*0201/CD80). These AAPC were loaded with HLA-A2 restricted peptides and used to stimulate CTLs in the presence of IL-2. Table 1 shows that this method consistently generated CTLs recognizing multiple HLA-A2 restricted peptides derived from well-characterized tumor-associated antigens including hTERT and PR1. The frequency of expanded T-lymphocytes was evaluated by IFNγ Elispot assay. Table 1. Irrelevant peptide Mart1-ELA Mage3-KVA PR1-VLQ WT1-RMF hTERT-RLV hTERT-ILA Tyr-YMD Number of IFNγ Spot Forming Cells (SFC)/10^5 cells (4 donors) 63±12 895±116 676±37 325±71 560±54 650±70 1019±240 897±127 8±3 810±30 488±19 15±10 48±18 13±7 101±12 503±20 7±2 1379±105 54±11 140±10 631±19 744±33 465±8 1103±97 3±2 1430±52 2±0 3±1 3±1 20±18 376±11 632±121 The specificity of the response was confirmed by tetramer analysis and cytotoxic activity with 51Cr release assay on peptide-loaded PHA blasts. We then evaluated whether this approach could be used to generate and expand PRAME-specific CTLs. In 5 HLA A*0201 healthy donors, after priming with B-blasts and 4 stimulations with K562/A*0201/CD80 cells loaded with the PRAME-derived peptide ALY, we obtained 22±7 fold T-cell expansion. T cells were PRAME specific, as the frequency of IFNγ+ T cells after exposure to the ALY peptide was significantly higher compared to exposure to irrelevant peptide (511±260 IFNg SFC/105 cells vs. 30±10 IFNg SFC/105 cells) in 4 of the 5 donors tested. In addition, CTLs significantly lysed autologous-PHA blasts loaded with ALY (84±12% at 20:1 E:T ratio) while lysis of blasts loaded with the irrelevant peptide was <10%. We then tested whether PRAME-CTLs recognized primary tumor cells, using CD33+ blast cells selected from the peripheral blood of 2 HLA A*0201 patients with PRAME expressing Chronic Myelogenous Leukemia (as assessed by Real Time PCR). The frequency of IFNγ+ T cells was 320±31 SFC/105 when ALY-CTLs were used as effector cells, but only 6±2 SFC/105 for CTLs expanded from the same donors using an irrelevant antigen. In conclusion, our data show that we can efficiently stimulate and expand PRAME-specific CTLs, and suggest that our approach could be developed for clinical application.

CD4 T-Helper Responses to the ALK Protein in Patients with ALK-Positive ALCL.

Anaplstic Lymphoma Kinase (ALK)-positive anaplastic large cell lymphoma (ALCL) has a favourable prognostic outlook compared to ALK-negative AlCL, possibly as a result of the immune recognition of the ALK protein. We have previously shown the presence of both a cytotoxic T cell and an antibody response to the ALK protein in patients with ALK-positive ALCL. The aim of our present study was to investigate the presence of a CD4 T-helper (Th) response in patients with ALK-positive ALCL and in control individuals. Using the TEPITOPE web-based predicitive search algorithm, three 24-mer promiscuous peptides were identified from the ALK sequence as being potentially immunogenic in the context of MHC class II. A gamma-interferon (γ-IFN) and IL-4 ELISPOT assay was used to detect a T cell response in the peripheral blood cells from patients with ALK-positive and ALK- negative ALCL, as well as healthy controls after 6–11 days of culture with the three peptides. ALK278–301 and ALK233–256 were shown to be highly immunogenic in the majority of the ALK-positive patients (see Table). ALK411–434 was immunogenic to T cells from only one of the ALK-positive patients (Patient 4). Cells from none of the two ALK-negative ALCL patients or the five healthy donors showed any reactivity to the ALK peptides. No response to the control irrelevant peptide was observed in any of the ALCL patients or healthy donors. With the exception of one ALK-positive ALCL patient (Patient 2), no significant IL-4 response was recorded in any of the patients or controls. All of the ALK-positive patients presented antibodies to the ALK protein at time of diagnosis.These findings further demonstrate the immunogenicity of the ALK protein and are suggestive of a Th1 type of immune response to the protein. Our findings are of potential prognostic value and open up therapeutic options for those ALK-positive patients who do not respond well to chemotherapy. Summary of the CD4 Th responses to ALK in ALCL patients and healthy donors None PHA (10 μ g/ml) ALK233–256 (10 μM)- (IFN- γ/IL-4) ALK278–301(10 μM)- (IFN- γ/IL-4) ALK411–434 (10 μM)- (IFN- γ/IL-4) Irrelevant peptide (10 μM)- (IFN- γ) Antibody titres to ALK (IgG isotype) ND= Not done. Results are of triplicate cultures ALK+ve patients Patient 1 12 188 56/10 44/18 22/6 8 1/2250 Patient 2 20 240 126/48 78/52 40/26 18 1/2250 Patient 3 14 48 38/ND 24/ND 12/ND 16 1/6750 Patient 4 6 108 64/8 72/8 22/6 10 1/60750 Patient 5 10 48 36/13 26/18 12/12 14 1/6750 Patient 6 15 132 74/ND 58/ND 28/ND 18 1/6750 Patient 7 10 180 34/28 56/32 12/9 12 1/750 ALK-ve patients Patient 8 14 122 12/10 10/6 12/14 22 −ve Patient 9 16 82 14/ND 12/ND 10/ND 24 −ve Healthy Donors Normal 1 22 148 18/14 22/24 26/12 10 −ve Normal 2 12 18 2/4 6/8 12/2 4 −ve Normal 3 10 38 12/10 12/16 9/4 18 −ve Normal 4 9 172 9/ND 10/ND 6/ND 12 −ve Normal 5 4 108 8/12 8/12 2/1 10 −ve

T Cell Receptor Gene Transfer to Produce MHC Class I-Restricted CD4 Helper T Cells: Implications for Immunotherapy of Leukaemia.

CD4 helper T cells play a critical role in the anti-tumour immune response. Cytokines secreted by CD4 T cells can have a direct effect on tumour cells and provide help for CTL priming and effector function. In this study we tested if it was possible to generate MHC class I-restricted helper T cells by retroviral TCR gene transfer into CD4 lymphocytes. Methods: We used a TCR (utilising V11) that recognises the influenza virus A nucleoprotein (NP366–379) peptide in the context of murine Db MHC class I. Murine splenocytes were isolated from C57BL/6 mice (H2b) and activated with conconavalin A and IL-7, and after 48 hours transduced with the pMX-TCR-IRES-TCR retroviral vector. The transduced splenocytes were then cultured in the presence of IL2 for a further 48 hours before staining with anti-murine CD4, CD8 and V11 antibodies and sorting into CD4+ V11+ and CD8+ V11+ populations. Sorted cells were expanded for a further 48–72 hours prior to functional assays. Functional Assays: Purified TCR-transduced (TCR-Td) CD8+ cells and purified TCR-Td CD4+ cells were tested for IFN secretion in response to dendritic cells (DCs) pulsed with NP peptide, an irrelevant peptide (pMDM100) or no peptide. Further experiments examined IFN secretion in response to peptide-loaded tumour cells (EL4 murine lymphoma cells) or transfected tumour cells expressing NP endogenously. Secretion of IFN was measured by ELISA. Results: (1) Antigen-specific IFN secretion was observed by both CD8+ (100% purity) and CD4+ cells (99.93% purity) expressing the class I-restricted TCR when incubated with peptide-loaded DCs. When tested with no peptide or irrelevant peptide, no IFN secretion was observed. The CD8+ cells were more sensitive, recognizing lower concentrations of peptide (10pM) than CD4+ cells (100pM). With peptide-coated EL4 tumour cells as stimulator cells, CD8+ cells showed a peptide-specific response. In contrast, the TCR-Td CD4+ cells were only able to elicit a weak peptide-specific response. Similarly, TCR-Td CD8+ cells were able to recognise NP transfected EL4 tumour cells (EL4NP68), whereas the CD4+ cells were unable to. However, the addition of syngeneic DCs restored the CD4+ cell response to NP transfected EL4 tumour cells to one equivalent to that seen with the TCR-Td CD8+ populations (Table 1). Summary: We have demonstrated that it is feasible to generate MHC class I-restricted CD4+ helper T cells, that are specific for peptide epitopes presented in the context of MHC class I. The CD4+ T cells can recognise antigen-expressing tumour cells in the presence of professional APC, such as DCs. The mechanism by which APC restore tumour recognition may involve trans-costimulation or cross presentation. The data suggest that class I-restricted CD4+ T cells may be able to contribute to enhanced anti-tumour immunity. αββββγγγγγβ γIFN Secretion (ng/ml) After Stimulation with DCs or Tumour Cells T Cell (Responder Cell) Stimulator Cell/s No Peptide NP (100nM) pMDM100 (100nM) Abbreviations: ND not done; DC, EL4 and EL4NP68 as indicated in text. TCR-Td CD8+ DCs 0.1 163.2 0.7 TCR-Td CD8+ EL4 0.1 19.9 0.2 TCR-Td CD8+ EL4NP68 16.6 ND ND TCR-Td CD8+ EL4NP68 + DCs 31.2 ND ND TCR-Td CD4+ DCs 0.1 163.9 0.2 TCR-Td CD4+ EL4 0.1 0.8 0.0 TCR-Td CD4+ EL4NP68 0.2 ND ND TCR-Td CD4+ EL4NP68 + DCs 25.3 ND ND

Unspecific Labeling of Pancreatic Islets by Antisera Against Fibroblast Growth Factors and Their Receptors

Six distinct fibroblast growth factors (FGF5) have been detected in pancreatic islets by immunohistochemistry (IHC) using commercially available antisera. We show here that these antisera are useful for Western blotting but that only two are suited for IHC. By Western blotting, these antisera detect recombinant FGFs. Detection can be eliminated by preabsorption with immunizing peptide but not with irrelevant peptide. By IHC we find specific labeling of islets with anti-FGF1 and anti-FGF2 antisera. Labeling can be abolished by preabsorption with the immunizing peptides. In contrast, prominent staining of islets by anti-FGF4, −FGF5, −FGF7, and −FGF10 antisera is unspecific because the staining cannot be competed by preabsorption with the immunizing peptides.

Differential detection of rat islet and brain glutamic acid decarboxylase (GAD) isoforms with sequence-specific peptide antibodies.

We studied the distribution of the M(r) 65,000 and M(r) 67,000 isoforms of glutamic acid decarboxylase, GAD65 and GAD67, in rat islets and brain by immunocytochemistry. Synthetic peptides representing selected GAD65 or GAD67 sequences were used to produce sequence-specific antibodies, allowing differential immunocytochemical detection of the two isoforms. GAD-specific reactivity of each peptide antiserum was confirmed by ELISA, immunoblotting, and immunoprecipitation. Immunostaining specificity was verified by displacement with either immunizing or irrelevant peptide. Dual immunostaining with GAD isoform-specific antibodies and polyclonal antibodies to glucagon showed that GAD65 was primarily detected in rat pancreatic islet beta-cells, whereas alpha-cells had weak GAD65 staining. In contrast, GAD67 was detected primarily in alpha-cells. In rat brain, GAD65 and GAD67 were present in neuron cell bodies and processes. These data demonstrate that antibodies raised against the N-terminus of GAD allow differential immunocytochemical identification of GAD67 and GAD65. Differential expression of GAD isoforms within islet alpha- and beta-cells supports the role of GAD65 in autoimmune diabetes and stiff-man syndrome.