Differentiation of mouse induced pluripotent stem cells (iPSCs) into nucleus pulposus-like cells in vitro.

A large percentage of the population may be expected to experience painful symptoms or disability associated with intervertebral disc (IVD) degeneration - a condition characterized by diminished integrity of tissue components. Great interest exists in the use of autologous or allogeneic cells delivered to the degenerated IVD to promote matrix regeneration. Induced pluripotent stem cells (iPSCs), derived from a patient's own somatic cells, have demonstrated their capacity to differentiate into various cell types although their potential to differentiate into an IVD cell has not yet been demonstrated. The overall objective of this study was to assess the possibility of generating iPSC-derived nucleus pulposus (NP) cells in a mouse model, a cell population that is entirely derived from notochord. This study employed magnetic activated cell sorting (MACS) to isolate a CD24(+) iPSC subpopulation. Notochordal cell-related gene expression was analyzed in this CD24(+) cell fraction via real time RT-PCR. CD24(+) iPSCs were then cultured in a laminin-rich culture system for up to 28 days, and the mouse NP phenotype was assessed by immunostaining. This study also focused on producing a more conducive environment for NP differentiation of mouse iPSCs with addition of low oxygen tension and notochordal cell conditioned medium (NCCM) to the culture platform. iPSCs were evaluated for an ability to adopt an NP-like phenotype through a combination of immunostaining and biochemical assays. Results demonstrated that a CD24(+) fraction of mouse iPSCs could be retrieved and differentiated into a population that could synthesize matrix components similar to that in native NP. Likewise, the addition of a hypoxic environment and NCCM induced a similar phenotypic result. In conclusion, this study suggests that mouse iPSCs have the potential to differentiate into NP-like cells and suggests the possibility that they may be used as a novel cell source for cellular therapy in the IVD.

Learning from Geometry

Subspaces and manifolds are two powerful models for high dimensional signals. Subspaces model linear correlation and are a good fit to signals generated by physical systems, such as frontal images of human faces and multiple sources impinging at an antenna array. Manifolds model sources that are not linearly correlated, but where signals are determined by a small number of parameters. Examples are images of human faces under different poses or expressions, and handwritten digits with varying styles. However, there will always be some degree of model mismatch between the subspace or manifold model and the true statistics of the source. This dissertation exploits subspace and manifold models as prior information in various signal processing and machine learning tasks.

A near-low-rank Gaussian mixture model measures proximity to a union of linear or affine subspaces. This simple model can effectively capture the signal distribution when each class is near a subspace. This dissertation studies how the pairwise geometry between these subspaces affects classification performance. When model mismatch is vanishingly small, the probability of misclassification is determined by the product of the sines of the principal angles between subspaces. When the model mismatch is more significant, the probability of misclassification is determined by the sum of the squares of the sines of the principal angles. Reliability of classification is derived in terms of the distribution of signal energy across principal vectors. Larger principal angles lead to smaller classification error, motivating a linear transform that optimizes principal angles. This linear transformation, termed TRAIT, also preserves some specific features in each class, being complementary to a recently developed Low Rank Transform (LRT). Moreover, when the model mismatch is more significant, TRAIT shows superior performance compared to LRT.

The manifold model enforces a constraint on the freedom of data variation. Learning features that are robust to data variation is very important, especially when the size of the training set is small. A learning machine with large numbers of parameters, e.g., deep neural network, can well describe a very complicated data distribution. However, it is also more likely to be sensitive to small perturbations of the data, and to suffer from suffer from degraded performance when generalizing to unseen (test) data.

From the perspective of complexity of function classes, such a learning machine has a huge capacity (complexity), which tends to overfit. The manifold model provides us with a way of regularizing the learning machine, so as to reduce the generalization error, therefore mitigate overfiting. Two different overfiting-preventing approaches are proposed, one from the perspective of data variation, the other from capacity/complexity control. In the first approach, the learning machine is encouraged to make decisions that vary smoothly for data points in local neighborhoods on the manifold. In the second approach, a graph adjacency matrix is derived for the manifold, and the learned features are encouraged to be aligned with the principal components of this adjacency matrix. Experimental results on benchmark datasets are demonstrated, showing an obvious advantage of the proposed approaches when the training set is small.

Stochastic optimization makes it possible to track a slowly varying subspace underlying streaming data. By approximating local neighborhoods using affine subspaces, a slowly varying manifold can be efficiently tracked as well, even with corrupted and noisy data. The more the local neighborhoods, the better the approximation, but the higher the computational complexity. A multiscale approximation scheme is proposed, where the local approximating subspaces are organized in a tree structure. Splitting and merging of the tree nodes then allows efficient control of the number of neighbourhoods. Deviation (of each datum) from the learned model is estimated, yielding a series of statistics for anomaly detection. This framework extends the classical {\em changepoint detection} technique, which only works for one dimensional signals. Simulations and experiments highlight the robustness and efficacy of the proposed approach in detecting an abrupt change in an otherwise slowly varying low-dimensional manifold.

Patient-derived endothelial progenitor cells improve vascular graft patency in a rodent model.

Late outgrowth endothelial progenitor cells (EPCs) derived from the peripheral blood of patients with significant coronary artery disease were sodded into the lumens of small diameter expanded polytetrafluoroethylene (ePTFE) vascular grafts. Grafts (1mm inner diameter) were denucleated and sodded either with native EPCs or with EPCs transfected with an adenoviral vector containing the gene for human thrombomodulin (EPC+AdTM). EPC+AdTM was shown to increase the in vitro rate of graft activated protein C (APC) production 4-fold over grafts sodded with untransfected EPCs (p<0.05). Unsodded control and EPC-sodded and EPC+AdTM-sodded grafts were implanted bilaterally into the femoral arteries of athymic rats for 7 or 28 days. Unsodded control grafts, both with and without denucleation treatment, each exhibited 7 day patency rates of 25%. Unsodded grafts showed extensive thrombosis and were not tested for patency over 28 days. In contrast, grafts sodded with untransfected EPCs or EPC+AdTM both had 7 day patency rates of 88-89% and 28 day patency rates of 75-88%. Intimal hyperplasia was observed near both the proximal and distal anastomoses in all sodded graft conditions but did not appear to be the primary occlusive failure event. This in vivo study suggests autologous EPCs derived from the peripheral blood of patients with coronary artery disease may improve the performance of synthetic vascular grafts, although no differences were observed between untransfected EPCs and TM transfected EPCs.