Fast-Training Deep Learning Algorithm for Multiplex Quantification of Mammalian Bioproduction Metabolites via Contactless Short-Wave Infrared Hyperspectral Sensing.

Within the biopharmaceutical sector, there exists the need for a contactless multiplex sensor, which can accurately detect metabolite levels in real time for precise feedback control of a bioreactor environment. Reported spectral sensors in the literature only work when fully submerged in the bioreactor and are subject to probe fouling due to a cell debris buildup. The use of a short-wave infrared (SWIR) hyperspectral (HS) cam era allows for efficient, fully contactless collection of large spectral datasets for metabolite quantification. Here, we report the development of an interpretable deep learning system, a convolution metabolite regression (CMR) approach that detects glucose and lactate concentrations using label-free contactless HS images of cell-free spent media samples from Chinese hamster ovary (CHO) cell growth flasks. Using a dataset of <500 HS images, these CMR algorithms achieved a competitive test root-mean-square error (RMSE) performance of glucose quantification within 27 mg/dL and lactate quantification within 20 mg/dL. Conventional Raman spectroscopy probes report a validation performance of 26 and 18 mg/dL for glucose and lactate, respectively. The CMR system trains within 10 epochs and uses a convolution encoder with a sparse bottleneck regression layer to pick the best-performing filters learned by CMR. Each of these filters is combined with existing interpretable models to produce a metabolite sensing system that automatically removes spurious predictions. Collectively, this work will advance the safe and efficient adoption of contactless deep learning sensing systems for fine control of a variety of bioreactor environments.

Changes in Structural and Thermodynamic Properties of Starch during Potato Tuber Dormancy.

The main reserve polysaccharide of plants-starch-is undoubtedly important for humans. One of the main sources of starch is the potato tuber, which is able to preserve starch for a long time during the so-called dormancy period. However, accumulated data show that this dormancy is only relative, which raises the question of the possibility of some kind of starch restructuring during dormancy periods. Here, the effect of long-term periods of tuber rest (at 2-4 °C) on main parameters of starches of potato tubers grown in vivo or in vitro were studied. Along with non-transgenic potatoes, Arabidopsis phytochrome B (AtPHYB) transformants were investigated. Distinct changes in starch micro and macro structures-an increase in proportion of amorphous lamellae and of large-sized and irregular-shaped granules, as well as shifts in thickness of the crystalline lamellae-were detected. The degree of such alterations, more pronounced in AtPHYB-transgenic tubers, increased with the longevity of tuber dormancy. By contrast, the polymorphic crystalline structure (B-type) of starch remained unchanged regardless of dormancy duration. Collectively, our data support the hypothesis that potato starch remains metabolically and structurally labile during the entire tuber life including the dormancy period. The revealed starch remodeling may be considered a process of tuber preadaptation to the upcoming sprouting stage.

Distribution of the amelogenin protein in developing, injured and carious human teeth.

Amelogenin is the major enamel matrix protein with key roles in amelogenesis. Although for many decades amelogenin was considered to be exclusively expressed by ameloblasts, more recent studies have shown that amelogenin is also expressed in other dental and no-dental cells. However, amelogenin expression in human tissues remains unclear. Here, we show that amelogenin protein is not only expressed during human embryonic development but also in pathological conditions such as carious lesions and injuries after dental cavity preparation. In developing embryonic teeth, amelogenin stage-specific expression is found in all dental epithelia cell populations but with different intensities. In the different layers of enamel matrix, waves of positive vs. negative immunostaining for amelogenin are detected suggesting that the secretion of amelogenin protein is orchestrated by a biological clock. Amelogenin is also expressed transiently in differentiating odontoblasts during predentin formation, but was absent in mature functional odontoblasts. In intact adult teeth, amelogenin was not present in dental pulp, odontoblasts, and dentin. However, in injured and carious adult human teeth amelogenin is strongly re-expressed in newly differentiated odontoblasts and is distributed in the dentinal tubuli under the lesion site. In an in vitro culture system, amelogenin is expressed preferentially in human dental pulp cells that start differentiating into odontoblast-like cells and form mineralization nodules. These data suggest that amelogenin plays important roles not only during cytodifferentiation, but also during tooth repair processes in humans.

EMMPRIN/CD147 deficiency disturbs ameloblast-odontoblast cross-talk and delays enamel mineralization.

Tooth development is regulated by a series of reciprocal inductive signaling between the dental epithelium and mesenchyme, which culminates with the formation of dentin and enamel. EMMPRIN/CD147 is an Extracellular Matrix MetalloPRoteinase (MMP) INducer that mediates epithelial-mesenchymal interactions in cancer and other pathological processes and is expressed in developing teeth. Here we used EMMPRIN knockout (KO) mice to determine the functional role of EMMPRIN on dental tissue formation. We report a delay in enamel deposition and formation that is clearly distinguishable in the growing incisor and associated with a significant reduction of MMP-3 and MMP-20 expression in tooth germs of KO mice. Insufficient basement membrane degradation is evidenced by a persistent laminin immunostaining, resulting in a delay of both odontoblast and ameloblast differentiation. Consequently, enamel volume and thickness are decreased in adult mutant teeth but enamel maturation and tooth morphology are normal, as shown by micro-computed tomographic (micro-CT), nanoindentation, and scanning electron microscope analyses. In addition, the dentino-enamel junction appears as a rough calcified layer of approximately 10±5µm thick (mean±SD) in both molars and growing incisors of KO adult mice. These results indicate that EMMPRIN is involved in the epithelial-mesenchymal cross-talk during tooth development by regulating the expression of MMPs. The mild tooth phenotype observed in EMMPRIN KO mice suggests that the direct effect of EMMPRIN may be limited to a short time window, comprised between basement membrane degradation allowing direct cell contact and calcified matrix deposition.

Distribution of syndecan-1 protein in developing mouse teeth.

Syndecan-1 is a cell surface proteoglycan involved in the regulation of various biological processes such as proliferation, migration, condensation and differentiation of cells, intercellular communication, and morphogenesis. The extracellular domain of syndecan-1 can bind to extracellular matrix components and signaling molecules, while its intracellular domain interacts with cytoskeletal proteins, thus allowing the transfer of information about extracellular environment changes into the cell that consequently affect cellular behavior. Although previous studies have shown syndecan-1 expression during precise stages of tooth development, there is no equivalent study regrouping the expression patterns of syndecan-1 during all stages of odontogenesis. Here we examined the distribution of syndecan-1 protein in embryonic and post-natal developing mouse molars and incisors. Syndecan-1 distribution in mesenchymal tissues such as dental papilla and dental follicle was correlated with proliferating events and its expression was often linked to stem cell niche territories. Syndecan-1 was also expressed in mesenchymal cells that will differentiate into the dentin producing odontoblasts, but not in differentiated functional odontoblasts. In the epithelium, syndecan-1 was detected in all cell layers, by the exception of differentiated ameloblasts that form the enamel. Furthermore, syndecan-1 was expressed in osteoblast precursors and osteoclasts of the alveolar bone that surrounds the developing tooth germs. Taken together these results show the dynamic nature of syndecan-1 expression during odontogenesis and suggest its implication in various processes of tooth development and homeostasis.