Improving the Accuracy of Spiking Neural Networks for Radar Gesture Recognition Through Preprocessing.

Event-based neural networks are currently being explored as efficient solutions for performing AI tasks at the extreme edge. To fully exploit their potential, event-based neural networks coupled to adequate preprocessing must be investigated. Within this context, we demonstrate a 4-b-weight spiking neural network (SNN) for radar gesture recognition, achieving a state-of-the-art 93% accuracy within only four processing time steps while using only one convolutional layer and two fully connected layers. This solution consumes very little energy and area if implemented in event-based hardware, which makes it suited for embedded extreme-edge applications. In addition, we demonstrate the importance of signal preprocessing for achieving this high recognition accuracy in SNNs compared to deep neural networks (DNNs) with the same network topology and training strategy. We show that efficient preprocessing prior to the neural network is drastically more important for SNNs compared to DNNs. We also demonstrate, for the first time, that the preprocessing parameters can affect SNNs and DNNs in antagonistic ways, prohibiting the generalization of conclusions drawn from DNN design to SNNs. We demonstrate our findings by comparing the gesture recognition accuracy achieved with our SNN to a DNN with the same architecture and similar training. Unlike previously proposed neural networks for radar processing, this work enables ultralow-power radar-based gesture recognition for extreme-edge devices.

Immunomodulatory effects of mixed hematopoietic chimerism: immune tolerance in canine model of lung transplantation.

Long-term survival after lung transplantation is limited by acute and chronic graft rejection. Induction of immune tolerance by first establishing mixed hematopoietic chimerism (MC) is a promising strategy to improve outcomes. In a preclinical canine model, stable MC was established in recipients after reduced-intensity conditioning and hematopoietic cell transplantation from a DLA-identical donor. Delayed lung transplantation was performed from the stem cell donor without pharmacological immunosuppression. Lung graft survival without loss of function was prolonged in chimeric (n = 5) vs. nonchimeric (n = 7) recipients (p < or = 0.05, Fisher's test). There were histological changes consistent with low-grade rejection in 3/5 of the lung grafts in chimeric recipients at > or =1 year. Chimeric recipients after lung transplantation had a normal immune response to a T-dependent antigen. Compared to normal dogs, there were significant increases of CD4+INFgamma+, CD4+IL-4+ and CD8+ INFgamma+ T-cell subsets in the blood (p < 0.0001 for each of the three T-cell subsets). Markers for regulatory T-cell subsets including foxP3, IL10 and TGFbeta were also increased in CD3+ T cells from the blood and peripheral tissues of chimeric recipients after lung transplantation. Establishing MC is immunomodulatory and observed changes were consistent with activation of both the effector and regulatory immune response.

Lentiviral vector conferring resistance to mycophenolate mofetil and sensitivity to ganciclovir for in vivo T-cell selection.

Several clinical studies of gene-modified T cells have shown limited in vivo function of the cells, immunogenicity of the transgene, and lack of a selective advantage for gene-modified T cells. To address these problems, we developed a lentiviral vector (LV) that provides a selectable, proliferative advantage and potentially decreases immunogenicity for transduced T cells. The bicistronic vector expressed two genes linked with an internal ribosomal entry site. One gene is a variant of the inosine monophosphate dehydrogenase 2, inosine monophosphate dehydrogenase (IMPDH(IY)), conferring resistance to the immunosuppressive drug mycophenolate mofetil (MMF). The other is a suicide gene, herpes simplex virus thymidine kinase (HSV-TK), rendering proliferating cells sensitive to ablation with ganciclovir, fused to the selectable transmembrane marker DeltaCD34 (DeltaCD34/TK). Cells transduced with LV-DeltaCD34/TK.IMPDH(IY) were efficiently enriched by immunomagnetic selection for CD34, proliferated in 0.5-5 microM MMF, and were killed by 0.5-25 microg ml(-1) ganciclovir. We demonstrate efficient selection and killing of gene-modified cells and suggest LV-DeltaCD34/TK.IMPDH(IY)-transduced T cells could be used to facilitate allogeneic hematopoietic cell engraftment. The expression of IMPDH(IY) would allow in vivo selection with MMF, and DeltaCD34/TK expression would allow rapid and safe elimination of transduced T cells if graft-versus-host disease developed.

IL-2 does not enhance the conversion to complete donor chimerism following nonmyeloablative hematopoietic cell transplantation in dogs.

A dog model of stable mixed hematopoietic chimerism was established in which leukocyte-antigen-identical littermates receive nonmyeloablative total body irradiation before hematopoietic cell transplantation and postgrafting immunosuppression with mycophenolate mofetil and cyclosporine. Unmodified donor lymphocyte infusion (DLI) into stable mixed chimeras failed to increase donor chimerism, while DLI from donors sensitized to recipient minor-histocompatibility antigens promptly converted all recipients to complete donor chimerism. This established a model for studying approaches to enhance the graft-versus-host (GVH)-effect, a potential surrogate for graft-versus-leukemia activity. We asked if interleukin-2 (IL-2) given after unmodified DLI could result in reliable conversion to complete donor chimerism. IL-2, 4 x 10(5) IU/kg/day, was administered to six mixed chimeric dogs for 14 days. Four dogs received unmodified DLI with IL-2. At 20-40 weeks after DLI, all dogs remained mixed chimeras. For the two recipients of IL-2 only, mixed chimerism also remained unchanged. These results show that IL-2 given with DLI after nonmyeloablative transplantation in dogs is not effective in reliably converting mixed to complete donor chimerism.

Engraftment of DLA-haploidentical marrow with ex vivo expanded, retrovirally transduced cytotoxic T lymphocytes.

Genetically modified donor T cells with an inducible "suicide" gene have the potential to improve the safety and availability of allogeneic hematopoietic stem cell transplantation by enhancing engraftment and permitting control of graft-versus-host disease (GVHD). However, several clinical studies of gene-modified T cells have shown limited to no in vivo function of the ex vivo expanded T cells. Using the well-established dog model of allogeneic marrow transplantation, the question was asked if retrovirally transduced, donor derived, ex vivo expanded cytotoxic T lymphocytes (CTLs) that are recipient specific could enhance engraftment of dog leukocyte antigen (DLA)-haploidentical marrow following a single dose of 9.2 Gy total body irradiation and no postgrafting immunosuppression. In this setting, only 4 of 11 control recipients of DLA-haploidentical marrow without added CTLs engrafted. CTLs did not enhance engraftment of CD34(+) selected peripheral blood stem cells. However, recipient-specific CTLs enhanced engraftment of DLA-haploidentical marrow in 9 of 11 evaluable recipients (P =.049). All dogs that engrafted developed multiorgan GVHD. To facilitate in vivo tracking, 8 dogs received CTLs transduced with a retroviral vector encoding green fluorescent protein (GFP) and neomycin phosphotransferase (neo). Recipients that engrafted had sharp increases in the numbers of circulating GFP(+) CTLs on days +5 to +6 after transplantation. GFP(+) CTLs isolated from blood were capable of recipient-specific lysis. At necropsy, up to 7.1% of CD3(+) cells in tissues were GFP(+) and polymerase chain reaction in situ hybridization for neo showed infiltration of transduced CTLs in GVHD-affected organs. These results show that ex vivo expanded, transduced T cells maintained in vivo function and enhanced marrow engraftment.

Hematopoietic cell transplantation in older patients with hematologic malignancies: replacing high-dose cytotoxic therapy with graft-versus-tumor effects.

Toxicities have limited the use of allogeneic hematopoietic cell transplantation (HCT) to younger, medically fit patients. In a canine HCT model, a combination of postgrafting mycophenolate mofetil (MMF) and cyclosporine (CSP) allowed stable allogeneic engraftment after minimally toxic conditioning with low-dose (200 cGy) total-body irradiation (TBI). These findings, together with the known antitumor effects of donor leukocyte infusions (DLIs), led to the design of this trial. Forty-five patients (median age 56 years) with hematologic malignancies, HLA-identical sibling donors, and relative contraindications to conventional HCT were treated. Immunosuppression involved TBI of 200 cGy before and CSP/MMF after HCT. DLIs were given after HCT for persistent malignancy, mixed chimerism, or both. Regimen toxicities and myelosuppression were mild, allowing 53% of eligible patients to have entirely outpatient transplantations. Nonfatal graft rejection occurred in 20% of patients. Grades II to III acute graft-versus-host disease (GVHD) occurred in 47% of patients with sustained engraftment. With median follow-up of 417 days, survival was 66.7%, nonrelapse mortality 6.7%, and relapse mortality 26.7%. Fifty-three percent of patients with sustained engraftment were in complete remission, including 8 with molecular remissions. This novel allografting approach, based on the use of postgrafting immunosuppression to control graft rejection and GVHD, has dramatically reduced the acute toxicities of allografting. HCT with the induction of potent graft-versus-tumor effects can be performed in previously ineligible patients, largely in an outpatient setting. Future protocol modifications should reduce rejection and GVHD, thereby facilitating studies of allogeneic immunotherapy for a variety of malignancies. (Blood. 2001;97:3390-3400)

Ex vivo expansion of canine dendritic cells from CD34+ bone marrow progenitor cells.

BACKGROUND: The aims of this study were to ex vivo expand canine dendritic cells and determine their phenotype and functional characteristics. METHODS: CD34+-selected cells and CD34+-depleted canine bone marrow (BM) cells were cultured in Iscove's modified medium for 14 days. Cytokines added to the cultures included human granylocyte/macrophage colony-stimulating factor 5 ng/ml, hFlt3 ligand 200 ng/ml, and human tumor necrosis factor-alpha 10 ng/ml. Cultured cells and purified subpopulations were assessed for cell surface antigen expression, morphology, and function by flow cytometric analysis, electron microscopy, and an allogeneic mixed lymphocyte reaction at day 14. RESULTS: Two main cell populations were identified, DR++(bright)/CD14- and DR+(dim)/CD14+. Ex vivo expanded CD34+-selected cells showed increased allostimulatory activity compared to both cultured CD34+-depleted cells and mononuclear cells. In contrast, ex vivo expansion from CD34+-depleted cells was unsuccessful. After sorting cells from the ex vivo expanded CD34+-selected bone marrow to enrich for DR++/CD14- cells, a 42-fold increase (median) of allostimulatory activity was observed as compared with sorted DR+/CD14+ cells (P=0.02). CONCLUSIONS: Cells with dentric cell-like phenotypes and functions can be cultured from canine CD34+-selected bone marrow cells. Future studies will address the roles of these cells in engraftment, graft versus host reactions and graft-host tolerance in a canine hematogoietic stem cell transplantaton model.

Adoptive immunotherapy in canine mixed chimeras after nonmyeloablative hematopoietic cell transplantation.

Development of nontoxic and nonmyeloablative regimens for allogeneic hematopoietic stem-cell transplantation will decrease transplantation-related mortality caused by regimen-related toxic effects. In pursuit of this goal, a dog model of stable mixed hematopoietic chimerism was established in which leukocyte-antigen-identical litter mates are given sublethal total-body irradiation (2 Gy) before stem-cell transplantation and immunosuppression with mycophenolate mofetil and cyclosporine afterward. In the current study, we examined whether donor lymphocyte infusion (DLI) could be used as adoptive immunotherapy to convert mixed to complete donor chimerism. First, 8 mixed chimeras were given unmodified DLI between day 36 and day 414 after stem-cell transplantation. After a 10- to 47-week follow-up period, there were no significant changes in the percentage of donor engraftment. Next, we immunized the donor to the minor histocompatibility antigens (mHA) of the recipient by means of repeated skin grafting. Lymphocytes from the mHA-sensitized donor were infused between day 201 and day 651 after transplantation. All 8 recipients of mHA-sensitized DLI had conversion to greater than 98% donor chimerism within 2 to 12 weeks of the infusion. Complications from mHA-sensitized DLI included graft-versus-host disease in 2 dogs and marrow aplasia in 1. These results showed that the low-dose transplant regimen establishes immune tolerance, and mHA-sensitized DLI is required to break tolerance, thereby converting mixed to complete donor chimerism. We propose that mixed chimerism established after nonmyeloablative allogeneic stem-cell transplantation provides a platform for adoptive immunotherapy that has clinical potential in the treatment of patients with malignant diseases.

Canine T cells transduced with a herpes simplex virus thymidine kinase gene: a model to study effects on engraftment and control of graft-versus-host disease.

BACKGROUND: Alloreactive donor T cells in marrow grafts mediate graft-versus-host disease (GVHD), but T-cell depletion has resulted in increased graft failure. Add-back of gene-modified alloreactive donor T cells could prevent graft rejection. After engraftment, in vivo depletion of those modified T cells with ganciclovir may control GVHD. METHODS: Canine recipient-specific donor cytotoxic T lymphocytes (CTL) were retrovirally transduced with the herpes simplex virus thymidine kinase gene. RESULTS: Gibbon ape leukemia virus-pseudotyped vector yielded primary CTL transduction efficiency of 22.9+/-9.9%. After selection and expansion, 96.7+/-0.8% of CTL expressed retrovirally transferred genes. Recipient-specific cytotoxic activity was maintained with 84.3% specific lysis. After ganciclovir treatment, herpes simplex virus thymidine kinase-transduced CTL proliferation was reduced 98.7+/-0.2% compared with controls. CONCLUSIONS: We have demonstrated efficient ex vivo transduction, expansion, maintenance of alloreactivity, and ganciclovir-mediated ablation of canine CTL, which will permit in vivo studies in the dog, a well-established model for GVHD and engraftment.

Insertions of up to 17 amino acids into a region of alpha-tubulin do not disrupt function in vivo.

Microtubules in yeasts are essential components of the mitotic and meiotic spindle and are necessary for nuclear movement during cell division and mating. The yeast Saccharomyces cerevisiae has two alpha-tubulin genes, TUB1 and TUB3, either of which alone is sufficient for these processes when present in a high enough copy number. Comparisons of sequences from several species reveals the presence of a variable region near the amino terminus of alpha-tubulin proteins. We perturbed the structure of this region in TUB3 by inserting into it 3, 9, or 17 amino acids and tested the ability of these altered proteins to function as the only alpha-tubulin protein in yeast cells. We found that each of these altered proteins was sufficient on its own for mitotic growth, mating, and methods of yeast. We conclude that this region can tolerate considerable variation without losing any of the highly conserved functions of alpha-tubulin. Our results suggest that variability in this region occurs because it can be tolerated, not because it specifies an important function for the protein.