Mapping of Environmental Data Using Kernel-Based Methods

Recently, kernel-based Machine Learning methods have gained great popularity in many data analysis and data mining fields: pattern recognition, biocomputing, speech and vision, engineering, remote sensing etc. The paper describes the use of kernel methods to approach the processing of large datasets from environmental monitoring networks. Several typical problems of the environmental sciences and their solutions provided by kernel-based methods are considered: classification of categorical data (soil type classification), mapping of environmental and pollution continuous information (pollution of soil by radionuclides), mapping with auxiliary information (climatic data from Aral Sea region). The promising developments, such as automatic emergency hot spot detection and monitoring network optimization are discussed as well.

Mult-class differentiation of cannabis seedlings in a forensic context

This article presents an experimental study about the classification ability of several classifiers for multi-classclassification of cannabis seedlings. As the cultivation of drug type cannabis is forbidden in Switzerland lawenforcement authorities regularly ask forensic laboratories to determinate the chemotype of a seized cannabisplant and then to conclude if the plantation is legal or not. This classification is mainly performed when theplant is mature as required by the EU official protocol and then the classification of cannabis seedlings is a timeconsuming and costly procedure. A previous study made by the authors has investigated this problematic [1]and showed that it is possible to differentiate between drug type (illegal) and fibre type (legal) cannabis at anearly stage of growth using gas chromatography interfaced with mass spectrometry (GC-MS) based on therelative proportions of eight major leaf compounds. The aims of the present work are on one hand to continueformer work and to optimize the methodology for the discrimination of drug- and fibre type cannabisdeveloped in the previous study and on the other hand to investigate the possibility to predict illegal cannabisvarieties. Seven classifiers for differentiating between cannabis seedlings are evaluated in this paper, namelyLinear Discriminant Analysis (LDA), Partial Least Squares Discriminant Analysis (PLS-DA), Nearest NeighbourClassification (NNC), Learning Vector Quantization (LVQ), Radial Basis Function Support Vector Machines(RBF SVMs), Random Forest (RF) and Artificial Neural Networks (ANN). The performance of each method wasassessed using the same analytical dataset that consists of 861 samples split into drug- and fibre type cannabiswith drug type cannabis being made up of 12 varieties (i.e. 12 classes). The results show that linear classifiersare not able to manage the distribution of classes in which some overlap areas exist for both classificationproblems. Unlike linear classifiers, NNC and RBF SVMs best differentiate cannabis samples both for 2-class and12-class classifications with average classification results up to 99% and 98%, respectively. Furthermore, RBFSVMs correctly classified into drug type cannabis the independent validation set, which consists of cannabisplants coming from police seizures. In forensic case work this study shows that the discrimination betweencannabis samples at an early stage of growth is possible with fairly high classification performance fordiscriminating between cannabis chemotypes or between drug type cannabis varieties.

Machine learning for geospatial data : algorithms, software tools and case studies

Résumé Cette thèse est consacrée à l'analyse, la modélisation et la visualisation de données environnementales à référence spatiale à l'aide d'algorithmes d'apprentissage automatique (Machine Learning). L'apprentissage automatique peut être considéré au sens large comme une sous-catégorie de l'intelligence artificielle qui concerne particulièrement le développement de techniques et d'algorithmes permettant à une machine d'apprendre à partir de données. Dans cette thèse, les algorithmes d'apprentissage automatique sont adaptés pour être appliqués à des données environnementales et à la prédiction spatiale. Pourquoi l'apprentissage automatique ? Parce que la majorité des algorithmes d'apprentissage automatiques sont universels, adaptatifs, non-linéaires, robustes et efficaces pour la modélisation. Ils peuvent résoudre des problèmes de classification, de régression et de modélisation de densité de probabilités dans des espaces à haute dimension, composés de variables informatives spatialisées (« géo-features ») en plus des coordonnées géographiques. De plus, ils sont idéaux pour être implémentés en tant qu'outils d'aide à la décision pour des questions environnementales allant de la reconnaissance de pattern à la modélisation et la prédiction en passant par la cartographie automatique. Leur efficacité est comparable au modèles géostatistiques dans l'espace des coordonnées géographiques, mais ils sont indispensables pour des données à hautes dimensions incluant des géo-features. Les algorithmes d'apprentissage automatique les plus importants et les plus populaires sont présentés théoriquement et implémentés sous forme de logiciels pour les sciences environnementales. Les principaux algorithmes décrits sont le Perceptron multicouches (MultiLayer Perceptron, MLP) - l'algorithme le plus connu dans l'intelligence artificielle, le réseau de neurones de régression généralisée (General Regression Neural Networks, GRNN), le réseau de neurones probabiliste (Probabilistic Neural Networks, PNN), les cartes auto-organisées (SelfOrganized Maps, SOM), les modèles à mixture Gaussiennes (Gaussian Mixture Models, GMM), les réseaux à fonctions de base radiales (Radial Basis Functions Networks, RBF) et les réseaux à mixture de densité (Mixture Density Networks, MDN). Cette gamme d'algorithmes permet de couvrir des tâches variées telle que la classification, la régression ou l'estimation de densité de probabilité. L'analyse exploratoire des données (Exploratory Data Analysis, EDA) est le premier pas de toute analyse de données. Dans cette thèse les concepts d'analyse exploratoire de données spatiales (Exploratory Spatial Data Analysis, ESDA) sont traités selon l'approche traditionnelle de la géostatistique avec la variographie expérimentale et selon les principes de l'apprentissage automatique. La variographie expérimentale, qui étudie les relations entre pairs de points, est un outil de base pour l'analyse géostatistique de corrélations spatiales anisotropiques qui permet de détecter la présence de patterns spatiaux descriptible par une statistique. L'approche de l'apprentissage automatique pour l'ESDA est présentée à travers l'application de la méthode des k plus proches voisins qui est très simple et possède d'excellentes qualités d'interprétation et de visualisation. Une part importante de la thèse traite de sujets d'actualité comme la cartographie automatique de données spatiales. Le réseau de neurones de régression généralisée est proposé pour résoudre cette tâche efficacement. Les performances du GRNN sont démontrées par des données de Comparaison d'Interpolation Spatiale (SIC) de 2004 pour lesquelles le GRNN bat significativement toutes les autres méthodes, particulièrement lors de situations d'urgence. La thèse est composée de quatre chapitres : théorie, applications, outils logiciels et des exemples guidés. Une partie importante du travail consiste en une collection de logiciels : Machine Learning Office. Cette collection de logiciels a été développée durant les 15 dernières années et a été utilisée pour l'enseignement de nombreux cours, dont des workshops internationaux en Chine, France, Italie, Irlande et Suisse ainsi que dans des projets de recherche fondamentaux et appliqués. Les cas d'études considérés couvrent un vaste spectre de problèmes géoenvironnementaux réels à basse et haute dimensionnalité, tels que la pollution de l'air, du sol et de l'eau par des produits radioactifs et des métaux lourds, la classification de types de sols et d'unités hydrogéologiques, la cartographie des incertitudes pour l'aide à la décision et l'estimation de risques naturels (glissements de terrain, avalanches). Des outils complémentaires pour l'analyse exploratoire des données et la visualisation ont également été développés en prenant soin de créer une interface conviviale et facile à l'utilisation. Machine Learning for geospatial data: algorithms, software tools and case studies Abstract The thesis is devoted to the analysis, modeling and visualisation of spatial environmental data using machine learning algorithms. In a broad sense machine learning can be considered as a subfield of artificial intelligence. It mainly concerns with the development of techniques and algorithms that allow computers to learn from data. In this thesis machine learning algorithms are adapted to learn from spatial environmental data and to make spatial predictions. Why machine learning? In few words most of machine learning algorithms are universal, adaptive, nonlinear, robust and efficient modeling tools. They can find solutions for the classification, regression, and probability density modeling problems in high-dimensional geo-feature spaces, composed of geographical space and additional relevant spatially referenced features. They are well-suited to be implemented as predictive engines in decision support systems, for the purposes of environmental data mining including pattern recognition, modeling and predictions as well as automatic data mapping. They have competitive efficiency to the geostatistical models in low dimensional geographical spaces but are indispensable in high-dimensional geo-feature spaces. The most important and popular machine learning algorithms and models interesting for geo- and environmental sciences are presented in details: from theoretical description of the concepts to the software implementation. The main algorithms and models considered are the following: multi-layer perceptron (a workhorse of machine learning), general regression neural networks, probabilistic neural networks, self-organising (Kohonen) maps, Gaussian mixture models, radial basis functions networks, mixture density networks. This set of models covers machine learning tasks such as classification, regression, and density estimation. Exploratory data analysis (EDA) is initial and very important part of data analysis. In this thesis the concepts of exploratory spatial data analysis (ESDA) is considered using both traditional geostatistical approach such as_experimental variography and machine learning. Experimental variography is a basic tool for geostatistical analysis of anisotropic spatial correlations which helps to understand the presence of spatial patterns, at least described by two-point statistics. A machine learning approach for ESDA is presented by applying the k-nearest neighbors (k-NN) method which is simple and has very good interpretation and visualization properties. Important part of the thesis deals with a hot topic of nowadays, namely, an automatic mapping of geospatial data. General regression neural networks (GRNN) is proposed as efficient model to solve this task. Performance of the GRNN model is demonstrated on Spatial Interpolation Comparison (SIC) 2004 data where GRNN model significantly outperformed all other approaches, especially in case of emergency conditions. The thesis consists of four chapters and has the following structure: theory, applications, software tools, and how-to-do-it examples. An important part of the work is a collection of software tools - Machine Learning Office. Machine Learning Office tools were developed during last 15 years and was used both for many teaching courses, including international workshops in China, France, Italy, Ireland, Switzerland and for realizing fundamental and applied research projects. Case studies considered cover wide spectrum of the real-life low and high-dimensional geo- and environmental problems, such as air, soil and water pollution by radionuclides and heavy metals, soil types and hydro-geological units classification, decision-oriented mapping with uncertainties, natural hazards (landslides, avalanches) assessments and susceptibility mapping. Complementary tools useful for the exploratory data analysis and visualisation were developed as well. The software is user friendly and easy to use.

Unraveling the complex network of cuticular structure and function.

A hydrophobic cuticle is deposited at the outermost extracellular matrix of the epidermis in primary tissues of terrestrial plants. Besides forming a protective shield against the environment, the cuticle is potentially involved in several developmental processes during plant growth. A high degree of variation in cuticle composition and structure exists between different plant species and tissues. Lots of progress has been made recently in understanding the different steps of biosynthesis, transport, and deposition of cuticular components. However, the molecular mechanisms that underlie cuticular function remain largely elusive.

Microbial influences on epithelial integrity and immune function as a basis for inflammatory diseases.

Certain autoimmune diseases as well as asthma have increased in recent decades, particularly in developed countries. The hygiene hypothesis has been the prevailing model to account for this increase; however, epidemiology studies also support the contribution of diet and obesity to inflammatory diseases. Diet affects the composition of the gut microbiota, and recent studies have identified various molecules and mechanisms that connect diet, the gut microbiota, and immune responses. Herein, we discuss the effects of microbial metabolites, such as short chain fatty acids, on epithelial integrity as well as immune cell function. We propose that dysbiosis contributes to compromised epithelial integrity and disrupted immune tolerance. In addition, dietary molecules affect the function of immune cells directly, particularly through lipid G-protein coupled receptors such as GPR43.

Abrogation of HMX1 function causes rare oculoauricular syndrome associated with congenital cataract, anterior segment dysgenesis, and retinal dystrophy.

PURPOSE: To define the phenotypic manifestation, confirm the genetic basis, and delineate the pathogenic mechanisms underlying an oculoauricular syndrome (OAS). METHODS: Two individuals from a consanguineous family underwent comprehensive clinical phenotyping and electrodiagnostic testing (EDT). Genome-wide microarray analysis and Sanger sequencing of the candidate gene were used to identify the likely causal variant. Protein modelling, Western blotting, and dual luciferase assays were used to assess the pathogenic effect of the variant in vitro. RESULTS: Complex developmental ocular abnormalities of congenital cataract, anterior segment dysgenesis, iris coloboma, early-onset retinal dystrophy, and abnormal external ear cartilage presented in the affected family members. Genetic analyses identified a homozygous c.650A>C; p.(Gln217Pro) missense mutation within the highly conserved homeodomain of the H6 family homeobox 1 (HMX1) gene. Protein modelling predicts that the variant may have a detrimental effect on protein folding and/or stability. In vitro analyses were able to demonstrate that the mutation has no effect on protein expression but adversely alters function. CONCLUSIONS: Oculoauricular syndrome is an autosomal recessive condition that has a profound effect on the development of the external ear, anterior segment, and retina, leading to significant visual loss at an early age. This study has delineated the phenotype and confirmed HMX1 as the gene causative of OAS, enabling the description of only the second family with the condition. HMX1 is a key player in ocular development, possibly in both the pathway responsible for lens and retina development, and via the gene network integral to optic fissure closure.

Immunocytochemical localisation of microtubule-associated proteins 1b and 2 in the developing rat spinal cord.

The straightforward anatomical organisation of the developing and mature rat spinal cord was used to determine and interpret the time of appearance and expression patterns of microtubule-associated proteins (MAP) 1b and 2. Immunoblots revealed the presence of MAP1b and 2 in the early embryonic rat spinal cord and confirmed the specificity of the used anti-MAP mouse monoclonal antibodies. The immunocytochemical data demonstrated a rostral-to-caudal and ventral-to-dorsal gradient in the expression of MAP1b/2 within the developing spinal cord. In the matrix layer, MAP1b was found in a distinct radial pattern distributed between the membrana limitans interna and externa between embryonal day (E)12 and E15. Immunostaining for vimentin revealed that this MAP1b pattern was morphologically and topographically different from the radial glial pattern which was present in the matrix layer between E13 and E19. The ventral-to-dorsal developmental gradient of the MAP1b staining in the spinal cord matrix layer indicates a close involvement of MAP1b either in the organisation of the microtubules in the cytoplasmatic extensions of the proliferating neuroblasts or neuroblast mitosis. MAP2 could not be detected in the developing matrix layer. In the mantle and marginal layer, MAP1b was abundantly present between E12 and postnatal day (P)0. After birth, the staining intensity for MAP1b gradually decreased in both layers towards a faint appearance at maturity. The distribution patterns suggest an involvement of MAP1b in the maturation of the motor neurons, the contralaterally and ipsilaterally projecting axons and the ascending and descending long axons of the rat spinal cord. MAP2 was present in the spinal cord grey matter between E12 and maturity, which reflects a role for MAP2 in the development as well as in the maintenance of microtubules. The present description of the expression patterns of MAP1b and 2 in the developing spinal cord suggests important roles of the two proteins in various morphogenetic events. The findings may serve as the basis for future studies on the function of MAP1b and 2 in the development of the central nervous system.

Structure-function relationships of the alpha1b-adrenergic receptor.

The alpha1b-adrenergic receptor (AR) is a member of the large superfamily of seven transmembrane domain (TMD) G protein-coupled receptors (GPCR). Combining site-directed mutagenesis of the alpha1b-AR with computational simulations of receptor dynamics, we have explored the conformational changes underlying the process of receptor activation, i.e. the transition between the inactive and active states. Our findings suggest that the structural constraint stabilizing the alpha1b-AR in the inactive form is a network of H-bonding interactions amongst conserved residues forming a polar pocket and R143 of the DRY sequence at the end of TMDIII. We have recently reported that point mutations of D142, of the DRY sequence and of A293 in the distal portion of the third intracellular loop resulted in ligand-independent (constitutive) activation of the alpha1b-AR. These constitutively activating mutations could induce perturbations resulting in the shift of R143 out of the polar pocket. The main role of R143 may be to mediate receptor activation by triggering the exposure of several basic amino acids of the intracellular loops towards the G protein. Our investigation has been extended also to the biochemical events involved in the desensitization process of alpha1b-AR. Our results indicate that immediately following agonist-induced activation, the alpha1b-AR can undergo rapid agonist-induced phosphorylation and desensitization. Different members of the G protein coupled receptor kinase family can play a role in agonist-induced regulation of the alpha1b-AR. In addition, constitutively active alpha1b-AR mutants display different phosphorylation and internalization features. The future goal is to further elucidate the molecular mechanism underlying the complex equilibrium between activation and inactivation of the alpha1b-AR and its regulation by pharmacological substances. These findings can help to elucidate the mechanism of action of various agents displaying properties of agonists or inverse agonists at the adrenergic system.

Getting in touch: segregated somatosensory what and where pathways in humans revealed by electrical neuroimaging.

Whether the somatosensory system, like its visual and auditory counterparts, is comprised of parallel functional pathways for processing identity and spatial attributes (so-called what and where pathways, respectively) has hitherto been studied in humans using neuropsychological and hemodynamic methods. Here, electrical neuroimaging of somatosensory evoked potentials (SEPs) identified the spatio-temporal mechanisms subserving vibrotactile processing during two types of blocks of trials. What blocks varied stimuli in their frequency (22.5 Hz vs. 110 Hz) independently of their location (left vs. right hand). Where blocks varied the same stimuli in their location independently of their frequency. In this way, there was a 2x2 within-subjects factorial design, counterbalancing the hand stimulated (left/right) and trial type (what/where). Responses to physically identical somatosensory stimuli differed within 200 ms post-stimulus onset, which is within the same timeframe we previously identified for audition (De Santis, L., Clarke, S., Murray, M.M., 2007. Automatic and intrinsic auditory "what" and "where" processing in humans revealed by electrical neuroimaging. Cereb Cortex 17, 9-17.). Initially (100-147 ms), responses to each hand were stronger to the what than where condition in a statistically indistinguishable network within the hemisphere contralateral to the stimulated hand, arguing against hemispheric specialization as the principal basis for somatosensory what and where pathways. Later (149-189 ms) responses differed topographically, indicative of the engagement of distinct configurations of brain networks. A common topography described responses to the where condition irrespective of the hand stimulated. By contrast, different topographies accounted for the what condition and also as a function of the hand stimulated. Parallel, functionally specialized pathways are observed across sensory systems and may be indicative of a computationally advantageous organization for processing spatial and identity information.

Trapping of naive lymphocytes triggers rapid growth and remodeling of the fibroblast network in reactive murine lymph nodes.

Adaptive immunity is initiated in T-cell zones of secondary lymphoid organs. These zones are organized in a rigid 3D network of fibroblastic reticular cells (FRCs) that are a rich cytokine source. In response to lymph-borne antigens, draining lymph nodes (LNs) expand several folds in size, but the fate and role of the FRC network during immune response is not fully understood. Here we show that T-cell responses are accompanied by the rapid activation and growth of FRCs, leading to an expanded but similarly organized network of T-zone FRCs that maintains its vital function for lymphocyte trafficking and survival. In addition, new FRC-rich environments were observed in the expanded medullary cords. FRCs are activated within hours after the onset of inflammation in the periphery. Surprisingly, FRC expansion depends mainly on trapping of naïve lymphocytes that is induced by both migratory and resident dendritic cells. Inflammatory signals are not required as homeostatic T-cell proliferation was sufficient to trigger FRC expansion. Activated lymphocytes are also dispensable for this process, but can enhance the later growth phase. Thus, this study documents the surprising plasticity as well as the complex regulation of FRC networks allowing the rapid LN hyperplasia that is critical for mounting efficient adaptive immunity.

Neuromodulation by oxytocin and vasopressin in the central nervous system as a basis for their rapid behavioral effects

The last several years have seen an increasing number of studies that describe effects of oxytocin and vasopressin on the behavior of animals or humans. Studies in humans have reported behavioral changes and, through fMRI, effects on brain function. These studies are paralleled by a large number of reports, mostly in rodents, that have also demonstrated neuromodulatory effects by oxytocin and vasopressin at the circuit level in specific brain regions. It is the scope of this review to give a summary of the most recent neuromodulatory findings in rodents with the aim of providing a potential neurophysiological basis for their behavioral effects. At the same time, these findings may point to promising areas for further translational research towards human applications.

Essential role of CD8 palmitoylation in CD8 coreceptor function.

To investigate the molecular basis that makes heterodimeric CD8alphabeta a more efficient coreceptor than homodimeric CD8alphaalpha, we used various CD8 transfectants of T1.4 T cell hybridomas, which are specific for H-2Kd, and a photoreactive derivative of the Plasmodium berghei circumsporozoite peptide PbCS 252-260 (SYIPSAEKI). We demonstrate that CD8 is palmitoylated at the cytoplasmic tail of CD8beta and that this allows partitioning of CD8alphabeta, but not of CD8alphaalpha, in lipid rafts. Localization of CD8 in rafts is crucial for its coreceptor function. First, association of CD8 with the src kinase p56lck takes place nearly exclusively in rafts, mainly due to increased concentration of both components in this compartment. Deletion of the cytoplasmic domain of CD8beta abrogated localization of CD8 in rafts and association with p56lck. Second, CD8-mediated cross-linking of p56lck by multimeric Kd-peptide complexes or by anti-CD8 Ab results in p56lck activation in rafts, from which the abundant phosphatase CD45 is excluded. Third, CD8-associated activated p56lck phosphorylates CD3zeta in rafts and hence induces TCR signaling and T cell activation. This study shows that palmitoylation of CD8beta is required for efficient CD8 coreceptor function, mainly because it dramatically increases CD8 association with p56lck and CD8-mediated activation of p56lck in lipid rafts.

Interim analysis of the Reitan Catheter Pump (RCP) heart failure efficacy study: RCP improves cardiovascular and renal function in acute decompensated heart failure (ADHF)

Background: The RCP is a 14 French collapsable percutaneous cardiovascular support device positioned in the descending part of the thoracic aorta via the femoral artery. A 10 patient first in man study demonstrated device safety and significant improvement in renal function among high risk PCI patients. We now report haemodynamic and renal efficacy in patients with ADHF.Methods: Prospective non randomised study seeking to recruit 20 patients with ADHF with a need for inotropic or mechanical circulatory support with: i) EF < 30% ii)Cardiac index(CI) < 2.2 L / min / m2 Outcome measures included: 1) Cardiac index (CI) 2) Pulmonary Capillary Wedge Pressure (PCWP) 3) Urine output / serum creatinine 4) Vascular / device complications 5) 30 day mortalityResults: INTERIM ANALYSIS (n=12) The mean age of the study group was 64 years, with a mean baseline creatinine of 193 umol/L, eGFR 38 ml/min. The intended RCP treatment period was 24 hours. During RCP treatment there was a significant mean reduction of PCWP at 4 hours of 17% (25 to 21 mmHg p=0.04). Mean CI increased at 12 hours by 11%, though not reaching significance (1.78 to 1.96 L/min/m2 p=0.08). RCP insertion prompted substantial diuresis. Urine output tripled over the first 12 hours compared to baseline (55 ml/hr vs 213 ml/hr p=0.03). This was associated with significantly improved renal function, a 28% reduction in serum creatinine at 12 hours (193 to 151 umol/L p=0.003), and a increase in eGFR from 38 ml/min to 50 ml/min (p=0.0007). 2 patients previously refused cardiac transplantation were reassessed and successfully transplanted within 9 months of RCP treatment on the basis of demonstrable renal reversibility. There were no vascular or device complications. There were 2 deaths at 30 days, one from multi-organ failure and sepsis, and one from intractable heart failure - neither were device related.Conclusion: RCP support in ADHF patients was associated with improved haemodynamics, and an improvement in renal function. The Reitan Catheter Pump may have a role in providing percutaneous cardiovascular and renal support in the acutely decompensated cardiac patient, and may have a role in suggesting renal reversibility in potential cardiac transplant patients. Further data will be reported at recruitment completion.

Circadian regulation of renal function.

Urinary excretion of water and all major electrolytes exhibit robust circadian oscillations. The 24-h periodicity has been well documented for several important determinants of urine formation, including renal blood flow, glomerular filtration, tubular reabsorption, and tubular secretion. Disturbance of the renal circadian rhythms is increasingly recognized as a risk factor for hypertension, polyuria, and other diseases and may contribute to renal fibrosis. The origin of these rhythms has been attributed to the reactive response of the kidney to circadian changes in volume and/or in the composition of extracellular fluids that are entrained by rest/activity and feeding/fasting cycles. However, numerous studies have shown that most of the renal excretory rhythms persist for long periods of time, even in the absence of periodic environmental cues. These observations led to the hypothesis of the existence of a self-sustained mechanism, enabling the kidney to anticipate various predictable circadian challenges to homeostasis. The molecular basis of this mechanism remained unknown until the recent discovery of the mammalian circadian clock made of a system of autoregulatory transcriptional/translational feedback loops, which have been found in all tissues studied, including the kidney. Here, we present a review of the growing evidence showing the involvement of the molecular clock in the generation of renal excretory rhythms.

Distinct profiles of cytotoxic granules in memory CD8 T cells correlate with function, differentiation stage, and antigen exposure.

Cytotoxic CD8 T cells exert their antiviral and antitumor activity primarily through the secretion of cytotoxic granules. Degranulation activity and cytotoxic granules (perforin plus granzymes) generally define CD8 T cells with cytotoxic function. In this study, we have investigated the expression of granzyme K (GrmK) in comparison to that of GrmA, GrmB, and perforin. The expression of the cytotoxic granules was assessed in virus-specific CD8 T cells specific to influenza virus, Epstein-Barr virus (EBV), cytomegalovirus (CMV), or human immunodeficiency virus type 1 (HIV-1). We observed a dichotomy between GrmK and perforin expression in virus-specific CD8 T cells. The profile in influenza virus-specific CD8 T cells was perforin(-) GrmB(-) GrmA(+/-) GrmK(+); in CMV-specific cells, it was perforin(+) GrmB(+) GrmA(+) GrmK(-/+); and in EBV- and HIV-1-specific cells, it was perforin(-/+) GrmB(+) GrmA(+) GrmK(+). On the basis of the delineation of memory and effector CD8 T cells with CD45RA and CD127, the GrmK(+) profile was associated with early-stage memory CD8 T-cell differentiation, the perforin(+) GrmB(+) GrmA(+) profile with advanced-stage differentiation, and the GrmB(+) GrmA(+) Grmk(+) profile with intermediate-stage differentiation. Furthermore, perforin and GrmB but not GrmA and GrmK correlated with cytotoxic activity. Finally, changes in antigen exposure in vitro and in vivo during primary HIV-1 infection and vaccination modulated cytotoxic granule profiles. These results advance our understanding of the relationship between distinct profiles of cytotoxic granules in memory CD8 T cells and function, differentiation stage, and antigen exposure.