Climate change of anthropogenic origin is affecting Earth’s biodiversity and therefore ecosystems and their services. High latitude ecosystems are even more impacted than the rest of Northern Hemisphere because of the amplified polar warming. Still, it is challenging to predict the dynamics of high latitude ecosystems because of complex interaction between abiotic and biotic components. As the past is the key to the future, the interpretation of past ecological changes to better understand ongoing processes is possible. In the Quaternary, the Pleistocene experienced several glacial and interglacial stages that affected past ecosystems. During the last Glacial, the Pleistocene steppe-tundra was covering most of unglaciated northern hemisphere and disappeared in parallel to the megafauna’s extinction at the transition to the Holocene (~11,700 years ago). The origin of the steppe-tundra decline is not well understood and knowledge on the mechanisms, which caused shifts in past communities and ecosystems, is of high priority as they are likely comparable to those affecting modern ecosystems. Lake or permafrost core sediments can be retrieved to investigate past biodiversity at transitions between glacial and interglacial stages. Siberia and Beringia were the origin of dispersal of the steppe-tundra, which make investigation this area of high priority. Until recently, macrofossils and pollen were the most common approaches. They are designed to reconstruct past composition changes but have limit and biases. Since the end of the 20th century, sedimentary ancient DNA (sedaDNA) can also be investigated. My main objectives were, by using sedaDNA approaches to provide scientific evidence of compositional and diversity changes in the Northern Hemisphere ecosystems at the transition between Quaternary glacial and interglacial stages. In this thesis, I provide snapshots of entire ancient ecosystems and describe compositional changes between Quaternary glacial and interglacial stages, and confirm the vegetation composition and the spatial and temporal boundaries of the Pleistocene steppe-tundra. I identify a general loss of plant diversity with extinction events happening in parallel of megafauna’ extinction. I demonstrate how loss of biotic resilience led to the collapse of a previously well-established system and discuss my results in regards to the ongoing climate change. With further work to constrain biases and limits, sedaDNA can be used in parallel or even replace the more established macrofossils and pollen approaches as my results support the robustness and potential of sedaDNA to answer new palaeoecological questions such as plant diversity changes, loss and provide snapshots of entire ancient biota.

In nature, plants often encounter biotic and abiotic stresses, which can cause reduced crop yield and quality, and threaten the nutrition of a growing human population. As heat stress (HS) is one of the main abiotic stresses, and is projected to increase due to global warming, it is necessary to better understand how plants respond and survive under HS. In Arabidopsis thaliana, plants can survive under severe HS if primed by a non-lethal HS, a process called acquisition of thermotolerance. This primed stated can be maintained for several days, and the ability of plants to maintain the primed state is called maintenance of acquired thermotolerance (mATT) or HS memory. According to current research, two Heat shock factors (HSFs) HSFA2 and HSFA3 are known to account for the majority of mATT capability, and there are other HSFs e.g. HSFA1b and HSFA6b in HSF complexes containing HSFA2 and/or HSFA3, however, the roles of these HSFs in HS memory is not clearly understood. Moreover, the mechanism of these HSFs in regulating HS memory is unclear, whether transcriptional machinery e.g. the Mediator complex contributes to transcriptional memory. This work investigates the role of HSFs and Mediator subunits in HS memory in A. thaliana. For the role of HSFs, the interaction between HSFA1b and HSFA2 during HS memory phase was confirmed by in vivo co- immunoprecipitation (Co-IP). HSFA1b, HSFA2, HSFA3 and HSFA6b targeted HS memory-related genes according to DNA affinity purification sequencing (DAP-seq) data, and targets of HSFA1b were confirmed in vivo by chromatin immunoprecipitation qPCR (ChIP-qPCR). The mutant of hsfa6b showed an HS memory deficiency phenotype in mATT survival assay. These data confirmed the role for HSFA2 and HSFA3 in HS memory, and suggest that HSFA1b and HSFA6b also function in HS memory. The Mediator complex functions as an RNA Polymerase II (RNA Pol II) co-regulator, and includes Head, Middle, Tail and Kinase modules. Both MED23 and MED32 belong to the Tail module, and they have a positive role in HS memory. MED23 interacted with HSFA3, as determined by yeast two hybrid (Y2H) and in vivo Co-IP assays. The med23 mutant showed a decreased HS memory phenotype, reduced expression of Type I (sustained expression) memory genes following HS, and reduced accumulation of the memory-associated Tri-methylation of histone H3 lysine 4 (H3K4me3)histone modification at HS memory-related gene loci after HS. MED23 was recruited to HS-inducible memory and non-memory genes after HS, as determined by ChIP-qPCR. The med32 mutant showed a reduced HS memory phenotype, decreased expression of Type I and Type II (hyper-induction) memory genes, and lower accumulation of H3K4me3 at memory gene lociafter HS. However, MED32 did not show interaction with any tested HSF in Y2H or in vivo Co-IP. MED32 regulated the recruitment of RNA Pol II at HS-inducible genes after HS, but was not itself recruited to HS memory genes after HS. These results provided more evidence that the Mediator subunits MED23 and MED32 regulate HS memory on transcriptional and epigenetic levels. In general, this work provides a better insight into the molecular mechanism of how HSFs and Mediator subunits regulate HS memory in plants and will provide new perspectives to breed crops with improved thermotolerance.

Plants can be primed to survive the exposure to a severe heat stress (HS) by prior exposure to a mild HS. The information about the priming stimulus is maintained by the plant for several days. This maintenance of acquired thermotolerance, or HS memory, is genetically separable from the acquisition of thermotolerance itself and several specific regulatory factors have been identified in recent years. On the molecular level, HS memory correlates with two types of transcriptional memory, type I and type II, that characterize a partially overlapping subset of HS-inducible genes. Type I transcriptional memory or sustained induction refers to the sustained transcriptional induction above non-stressed expression levels of a gene for a prolonged time period after the end of the stress exposure. Type II transcriptional memory refers to an altered transcriptional response of a gene after repeated exposure to a stress of similar duration and intensity. In particular, enhanced re-induction refers to a transcriptional pattern in which a gene is induced to a significantly higher degree after the second stress exposure than after the first. This thesis describes the functional characterization of a novel positive transcriptional regulator of type I transcriptional memory, the heat shock transcription factor HSFA3, and compares it to HSFA2, a known positive regulator of type I and type II transcriptional memory. It investigates type I transcriptional memory and its dependence on HSFA2 and HSFA3 for the first time on a genome-wide level, and gives insight on the formation of heteromeric HSF complexes in response to HS. This thesis confirms the tight correlation between transcriptional memory and H3K4 hyper-methylation, reported here in a case study that aimed to reduce H3K4 hyper-methylation of the type II transcriptional memory gene APX2 by CRISPR/dCas9-mediated epigenome editing. Finally, this thesis gives insight into the requirements for a heat shock transcription factor to function as a positive regulator of transcriptional memory, both in terms of its expression profile and protein abundance after HS and the contribution of individual functional domains. In summary, this thesis contributes to a more detailed understanding of the molecular processes underlying transcriptional memory and therefore HS memory, in Arabidopsis thaliana.

As of late, epidemiological studies have highlighted a strong association of dairy intake with lower disease risk, and similarly with an increased amount of odd-chain fatty acids (OCFA). While the OCFA also demonstrate inverse associations with disease incidence, the direct dietary sources and mode of action of the OCFA remain poorly understood. The overall aim of this thesis was to determine the impact of two main fractions of dairy, milk fat and milk protein, on OCFA levels and their influence on health outcomes under high-fat (HF) diet conditions. Both fractions represent viable sources of OCFA, as milk fats contain a significant amount of OCFA and milk proteins are high in branched chain amino acids (BCAA), namely valine (Val) and isoleucine (Ile), which can produce propionyl-CoA (Pr-CoA), a precursor for endogenous OCFA synthesis, while leucine (Leu) does not. Additionally, this project sought to clarify the specific metabolic effects of the OCFA heptadecanoic acid (C17:0). Both short-term and long-term feeding studies were performed using male C57BL/6JRj mice fed HF diets supplemented with milk fat or C17:0, as well as milk protein or individual BCAA (Val; Leu) to determine their influences on OCFA and metabolic health. Short-term feeding revealed that both milk fractions induce OCFA in vivo, and the increases elicited by milk protein could be, in part, explained by Val intake. In vitro studies using primary hepatocytes further showed an induction of OCFA after Val treatment via de novo lipogenesis and increased α-oxidation. In the long-term studies, both milk fat and milk protein increased hepatic and circulating OCFA levels; however, only milk protein elicited protective effects on adiposity and hepatic fat accumulation—likely mediated by the anti-obesogenic effects of an increased Leu intake. In contrast, Val feeding did not increase OCFA levels nor improve obesity, but rather resulted in glucotoxicity-induced insulin resistance in skeletal muscle mediated by its metabolite 3-hydroxyisobutyrate (3-HIB). Finally, while OCFA levels correlated with improved health outcomes, C17:0 produced negligible effects in preventing HF-diet induced health impairments. The results presented herein demonstrate that the beneficial health outcomes associated with dairy intake are likely mediated through the effects of milk protein, while OCFA levels are likely a mere association and do not play a significant causal role in metabolic health under HF conditions. Furthermore, the highly divergent metabolic effects of the two BCAA, Leu and Val, unraveled herein highlight the importance of protein quality.

The complex hierarchical structure of bone undergoes a lifelong remodeling process, where it adapts to mechanical needs. Hereby, bone resorption by osteoclasts and bone formation by osteoblasts have to be balanced to sustain a healthy and stable organ. Osteocytes orchestrate this interplay by sensing mechanical strains and translating them into biochemical signals. The osteocytes are located in lacunae and are connected to one another and other bone cells via cell processes through small channels, the canaliculi. Lacunae and canaliculi form a network (LCN) of extracellular spaces that is able to transport ions and enables cell-to-cell communication. Osteocytes might also contribute to mineral homeostasis by direct interactions with the surrounding matrix. If the LCN is acting as a transport system, this should be reflected in the mineralization pattern. The central hypothesis of this thesis is that osteocytes are actively changing their material environment. Characterization methods of material science are used to achieve the aim of detecting traces of this interaction between osteocytes and the extracellular matrix. First, healthy murine bones were characterized. The properties analyzed were then compared with three murine model systems: 1) a loading model, where a bone of the mouse was loaded during its life time; 2) a healing model, where a bone of the mouse was cut to induce a healing response; and 3) a disease model, where the Fbn1 gene is dysfunctional causing defects in the formation of the extracellular tissue. The measurement strategy included routines that make it possible to analyze the organization of the LCN and the material components (i.e., the organic collagen matrix and the mineral particles) in the same bone volumes and compare the spatial distribution of different data sets. The three-dimensional network architecture of the LCN is visualized by confocal laser scanning microscopy (CLSM) after rhodamine staining and is then subsequently quantified. The calcium content is determined via quantitative backscattered electron imaging (qBEI), while small- and wide-angle X-ray scattering (SAXS and WAXS) are employed to determine the thickness and length of local mineral particles. First, tibiae cortices of healthy mice were characterized to investigate how changes in LCN architecture can be attributed to interactions of osteocytes with the surrounding bone matrix. The tibial mid-shaft cross-sections showed two main regions, consisting of a band with unordered LCN surrounded by a region with ordered LCN. The unordered region is a remnant of early bone formation and exhibited short and thin mineral particles. The surrounding, more aligned bone showed ordered and dense LCN as well as thicker and longer mineral particles. The calcium content was unchanged between the two regions. In the mouse loading model, the left tibia underwent two weeks of mechanical stimulation, which results in increased bone formation and decreased resorption in skeletally mature mice. Here the specific research question addressed was how do bone material characteristics change at (re)modeling sites? The new bone formed in response to mechanical stimulation showed similar properties in terms of the mineral particles, like the ordered calcium region but lower calcium content compared to the right, non-loaded control bone of the same mice. There was a clear, recognizable border between mature and newly formed bone. Nevertheless, some canaliculi went through this border connecting the LCN of mature and newly formed bone. Additionally, the question should be answered whether the LCN topology and the bone matrix material properties adapt to loading. Although, mechanically stimulated bones did not show differences in calcium content compared to controls, different correlations were found between the local LCN density and the local Ca content depending on whether the bone was loaded or not. These results suggest that the LCN may serve as a mineral reservoir. For the healing model, the femurs of mice underwent an osteotomy, stabilized with an external fixator and were allowed to heal for 21 days. Thus, the spatial variations in the LCN topology with mineral properties within different tissue types and their interfaces, namely calcified cartilage, bony callus and cortex, could be simultaneously visualized and compared in this model. All tissue types showed structural differences across multiple length scales. Calcium content increased and became more homogeneous from calcified cartilage to bony callus to lamellar cortical bone. The degree of LCN organization increased as well, while the lacunae became smaller, as did the lacunar density between these different tissue types that make up the callus. In the calcified cartilage, the mineral particles were short and thin. The newly formed callus exhibited thicker mineral particles, which still had a low degree of orientation. While most of the callus had a woven-like structure, it also served as a scaffold for more lamellar tissue at the edges. The lamelar bone callus showed thinner mineral particles, but a higher degree of alignment in both, mineral particles and the LCN. The cortex showed the highest values for mineral length, thickness and degree of orientation. At the same time, the lacunae number density was 34% lower and the lacunar volume 40% smaller compared to bony callus. The transition zone between cortical and callus regions showed a continuous convergence of bone mineral properties and lacunae shape. Although only a few canaliculi connected callus and the cortical region, this indicates that communication between osteocytes of both tissues should be possible. The presented correlations between LCN architecture and mineral properties across tissue types may suggest that osteocytes have an active role in mineralization processes of healing. A mouse model for the disease marfan syndrome, which includes a genetic defect in the fibrillin-1 gene, was investigated. In humans, Marfan syndrome is characterized by a range of clinical symptoms such as long bone overgrowth, loose joints, reduced bone mineral density, compromised bone microarchitecture, and increased fracture rates. Thus, fibrillin-1 seems to play a role in the skeletal homeostasis. Therefore, the present work studied how marfan syndrome alters LCN architecture and the surrounding bone matrix. The mice with marfan syndrome showed longer tibiae than their healthy littermates from an age of seven weeks onwards. In contrast, the cortical development appeared retarded, which was observed across all measured characteristics, i. e. lower endocortical bone formation, looser and less organized lacuno-canalicular network, less collagen orientation, thinner and shorter mineral particles. In each of the three model systems, this study found that changes in the LCN architecture spatially correlated with bone matrix material parameters. While not knowing the exact mechanism, these results provide indications that osteocytes can actively manipulate a mineral reservoir located around the canaliculi to make a quickly accessible contribution to mineral homeostasis. However, this interaction is most likely not one-sided, but could be understood as an interplay between osteocytes and extra-cellular matrix, since the bone matrix contains biochemical signaling molecules (e.g. non-collagenous proteins) that can change osteocyte behavior. Bone (re)modeling can therefore not only be understood as a method for removing defects or adapting to external mechanical stimuli, but also for increasing the efficiency of possible osteocyte-mineral interactions during bone homeostasis. With these findings, it seems reasonable to consider osteocytes as a target for drug development related to bone diseases that cause changes in bone composition and mechanical properties. It will most likely require the combined effort of materials scientists, cell biologists, and molecular biologists to gain a deeper understanding of how bone cells respond to their material environment.