【作者】 范秀波；

【导师】 刘天庆；

【作者基本信息】 大连理工大学 ， 生物医学工程， 2008， 博士

【摘要】 造血干细胞（Hematopoietic stem cells，HSCs）已被用于基因治疗、肿瘤净化和造血重建等方面，然而由于HSCs资源匮乏，严重限制了其临床应用。虽然前人对HSCs的体外扩增已经做了大量的实验研究，但由于培养条件对HSCs体外扩增的影响非常复杂，在不同的培养条件下，扩增效果差别很大，至今尚未有HSCs体外扩增条件优化方面的报道。因此，如果能够找到一种适宜的数学方法来建立HSCs体外培养条件与扩增效果之间的定量关系，对于优化HSCs体外培养条件以及解决临床应用上所面临的HSCs数量不足的难题无疑具有重大意义。本文首先采用神经网络技术对HSCs的体外扩增能力建立了评价及预测模型。从文献上总结前人的实验结果，共获得341组数据。选取细胞接种密度、细胞因子组合、细胞来源、血清、基质细胞、反应器类型和培养时间等7个影响因子作为网络输入特征参数，分别对有核细胞（Nuclear cells，NCs）、CD34⁺细胞和成集落细胞（Colony-formingunits，CFU-Cs）进行体外扩增能力的评价及预测。选取124、90及86组实验数据分别用于NCs、CFU-Cs和CD34⁺细胞为评价指标的神经网络训练；而17、10及14组实验数据分别用于NCs、CFU-Cs和CD34⁺细胞为评价指标的神经网络预测。结果表明，对NCs、CFU-Cs和CD34⁺细胞的区间训练准确率分别为85．5％、86．1％和86．7％；区间预测准确率分别为82．4％、71．4％和70．0％。由此可知，人工神经网络的非线性模拟使定量描述HSCs体外扩增效果与培养条件间的关系及预测HSCs的最佳体外扩增条件成为可能。除了对文献上已有的实验数据进行评价与预测外，本文还应用所建立的神经网络模型对文献报道之外的培养条件进行优化并对其扩增结果进行了预测。将影响HSCs体外扩增的7个主要影响因素的取值进行定义并对各取值进行组合，共得到18480组培养条件。这其中只有少部分培养条件的实验结果得到实验数据的验证，但绝大部分培养条件的预测结果尚未验证。为了安全可靠地使用神经网络模型所预测的优化培养条件，必须对神经网络的外推预测能力进行实验验证。因此，本研究从18480组培养条件中共筛选出6组具有代表性的培养条件进行了验证实验。由于其中多组实验都需要采用海藻酸钙-壳聚糖（Alginate chitosan，AC）胶珠包埋基质细胞再与HSCs共培养，为此，首先通过正交实验筛选，确定了AC胶珠的最佳制备工艺为：海藻酸钠的浓度为1％（w／v）、氯化钙的浓度为3％（w／v）、壳聚糖的浓度为1％（w／v）、第一步和第二步反应时间均为8分钟。在确定了AC胶珠最佳制备工艺的基础上，进一步通过正交实验的方法确定了AC胶珠包埋兔骨髓间充质干细胞（Mesenchymal stem cells，MSCs）支持脐带血（Umbilical cord blood，UCB）HSCs扩增的最佳条件为：兔骨髓MSCs在海藻酸钠内的接种密度为2×10⁵cells·mL^-1，高低粘度壳聚糖的混合质量比为3：1及单核细胞（Mononuclear cells，MNCs）与兔骨髓MSCs的接种比例为5：1。该结果为验证实验及后续的反应器内共培养UCB-HSCs和UCB-MSCs两种干细胞确定了AC胶珠包埋基质细胞的最佳条件。6组验证实验结果显示，对于CD34⁺细胞、NCs及CFU-Cs和的体外扩增，分别有4、5及6组验证实验结果与预测结果相吻合，准确率分别为67．7％、83．3％及100％。该结果充分证明了本论文所建立的神经网络模型在HSCs体外扩增的模拟分析与预测上的可靠性与适用性。脐带血中所含的另一种干细胞UCB-MSCs，不仅可以作为滋养层细胞支持HSCs在体外的大规模扩增，在造血移植过程中还能够降低并发症的发生率以及加速造血重建功能的恢复。但是目前UCB-MSCs的原代优化培养成功率（Probability）一般只有50％～60％，为了进一步提高培养成功率，本文利用正交实验方法对UCB-MSCs体外培养的主要影响因素：细胞的接种密度、细胞因子的组合及用量、是否添加血清和滋养层细胞，进行正交实验筛选：并对培养出的UCB-MSCs进行了细胞免疫表型分析和多向诱导分化检测，以期获得UCB-MSCs培养的最佳方法。实验结果表明，MSCs在以高细胞密度接种的基础上添加15ng·mL^-1IL-3和5ng·mL^-1GM-CSF，可大大提高UCB-MSCs的原代培养成功率，从50％～60％提高到90％以上。细胞免疫表型分析结果显示，所分离培养的细胞表达间质细胞的相关表面标志CD13、CD29、CD44及CD105，不表达造血细胞的相关表面标志CD34、CD45及HLA-DR；并能够向骨、软骨及脂肪细胞分化，这与骨髓来源的MSCs相一致。利用本文所建立的培养方法，能够为UCB-MSCs的临床应用提供大量优质的种子细胞。最后，对脐带血中所含有的HSCs和MSCs两种干细胞进行共培养研究，无疑具有重要的临床应用价值。本文在不添加血清、只添加细胞因子组合(SCF 15 ng·mL^-1，FL 5ng·mL^-1，TPO 6 ng·mL^-1，IL-3 15 ng·mL^-1，G-CSF 1 ng·mL^-1，GM-CSF 5 ng·mL^-1)及AC胶珠包被基质细胞支持的条件下，采用微载体与生物反应器相结合的策略，考察了UCB-HSCs与UCB-MSCs在转瓶及旋转壁式生物反应器（Rotating wall vessel bioreactor，RWVB）内的共培养。结果表明，在RWVB中的扩增效果最佳，12天内NCs扩增了3．7±0．3倍；CFU-Cs扩增了5．1±1．2倍；CD34⁺CD45⁺CD105^-（HSCs）细胞扩增了5．2±0．4倍；CD34^-CD45^-CD105⁺（MSCs）细胞扩增了13．9±1．2倍。培养结束后，通过自由沉降的方法分离UCB-HSCs和粘附在玻璃包被的苯乙烯聚合物（Glass coated styrenecopolymer，GCSC）微载体表面的UCB-MSCs。同时，细胞多向诱导分化及免疫表型分析结果显示，粘附在GCSC微载体表面上的细胞能够向骨、软骨及脂肪细胞分化；并能够表达间质细胞相关表面标志CD13，CD44，CD73和CD105，而不表达造血细胞的相关表面标志CD34，CD45及HLA-DR，与骨髓MSCs相一致。该结果说明，采用RWVB并在添加细胞因子、基质细胞及微载体支持的条件下，能够实现UCB-HSCs和UCB-MSCs的联合共培养。更多还原

【Abstract】 Hematopoietic stem cells （HSCs） have been used for gene therapy, tumor depollution and hematopoiesis reconstitution. But the shortage of HSCs greatly limits their widespread clinical applications. Though a great deal of research work about ex-vivo expansion of HSCs has been done, expansion results varied in a considerably wide range under different culture conditions because of the complicated relationship between them. Till now, there is no report about optimization for HSCs ex-vivo expansion. Hence, it is important to build up the quantitative relationship between culture conditions and expansion results for HSCs through a suitable mathematic method for optimizing the HSCs ex-vivo expansion culture conditions and solving shortage of HSCs in clinical applications undoubtedly.An evaluating and predictive model for the ex-vivo expansion of HSCs was firstly built up in this study with artifical neural network （ANN） technology. 341 groups of data were summarized from literatures, in which 124, 86 and 90 data were employed to train the network and 17, 14 and 10 data were applied to predict respectively. Expansion folds of nuclear cells （NCs）, CD34⁺ cells and Colony-forming units （CFU-Cs） were chosen as evaluation objectives and inoculated density, cytokines, cell resources, serum, stromal cells, culture time and bioreactor types were chosen as network inputs. The calculated results show that for the training of network, the interval accuracy of the expansion folds for the different cells is 85.5%, 86.1% and 86.7% respectively. While for the prediction of network, the interval accuracy can be up to 82.4%, 71.4% and 70.0% respectively. Therefore this nonlinear modeling makes it possible to describe quantitatively the effects of the culture conditions on the HSCs expansion and to predict the optimal culture conditions for higher ex-vivo expansion of HSCs.Besides the evaluation and prediction for the existing literature data, more optimizations for culture condition and predictions for the HSC expansion were carried out with the founded ANN model. The quantification of all the 7 factors influencing the ex-vivo expansion of HSCs was defined and the case combination of the values taken by these influencing factors could result in 18480 groups of possible culture conditions. Only quite few predictive results of these conditions have been verified with experiments, most of them were still not. In order to use the optimized culture conditions with ANN model safely and reliably, the extrapolated predictive reliability of the network need to be confirmed with more verification experiments. Therefore, 6 groups of representative culture conditions were selected as verification experiments from all these 18480 cases. Since alginate chitosan （AC） beads would be applied in the encapsulation of stromal cells to coculture HSCs in most verification experiments, the optimized preparation parameters for AC beads were firstly determined by fractional factorial experiments, which were 1% （wt%） alginate, 3% （wt%） CaCl₂, 1% （wt%） chitosan, 8 minutes for both the first and the second step reaction. Based on this, the optimized operation conditions for stromal cells （Rabbit mesenchymal stem cells, Rabbit-MSCs） encapsulated in AC beads to support the ex-vivo expansion of UCB-HSCs were also determined by fractional factorial tests, which were 2×l0⁵cells·mL^-1 inoculate density for Rabbit-MSCs in alginate, 3:1 for the weight ratio of high and low molecular weight of chitosan and 5:1 for cell inoculate density ratio of UCB-MNCs and Rabbit-MSCs. Thus the optimized protocol was determined to encapsulate stromal cells in AC beads for the next verification experiments and subsequent coculture of UCB-HSCs and UCB-MSCs in bioreactors.The results of 6 groups of verification experiments showed that there were 4, 5 and 6 experimental results in agreement with the predictive results for the ex-vivo expansion of CD34⁺ cells, NCs and CFU-Cs respectively, which meaned that the predictive accuracy of this ANN was 67.7%, 83.3% and 100%. Therefore the ANN model was reliable and suitable.UCB-MSC is another type of stem cells in UCB, which can not only support the expansion of HSCs in vitro as stromal cells but also alleviate complications and accelerate recovery of hematopoiesis during hematopoietic stem cell transplantation. However it proved challenging to culture MSCs from UCB with a low probability of 50%～60% even after optimized study. In this work, in order to improve the probability of separating and obtaining UCB-MSCs, three fractional factorial experiments were designed and performed to investigate the main influencing factors for the primary culture, the key cytokines and the dose of the suitable cytokines used. The cultured UCB-MSC-like cells were characterized by immunophenotypic and multi-lineage induced differentiation analysis. The experimental results showed that the probability of culture UCB-MSCs could be improved from 50%～60% to 90% with adding 18ng·ml^-1 IL-3 and 5ng·ml^-1 GM-CSF based on high cell inoculated density. Moreover, the UCB-MSC-like cells expressed MSCs-related surface markers of CD13, CD29, CD44 and CD105, but not hematopoietic cells-related surface markers of CD34, CD45 and HLA-DR. Meanwhile, these cells could differentiate into osteoblasts, chondrocytes and adipocytes similarly to MSCs derived from bone marrow. Therefore, it is possible to provide enough UCB-MSCs for clinical application with this opitimized culture method.Lastly, the feasibility of coculture autologous UCB-HSCs and UCB-MSCs was investigated because of the great importance of clinical application. Spinner flasks and rotating wall vessel bioreactor （RWVB） together with glass coated styrene copolymer （GCSC） microcarriers were applied in the study. IMDM medium without serum but supported with the combination of cytokines, including SCF 15ng·mL^-1, FL 5ng·mL^-1, TPO 6ng·mL^-1, IL-3 15ng·mL^-1, G-CSF 1ng·mL^-1 and GM-CSF 5ng·mL^-1, was adopted. And allogeneic adipose derived stem cells encapsulated in AC beads were used for the coculture with UCB-HSCs and UCB-MSCs. Meanwhile diluted-feeding protocol was applied during the culture process. The results indicated that the expansions of total cell number, CFU-Cs, CD34⁺CD45⁺CD105^-（HSCs） cells and CD34^-CD45^-CD105⁺ （MSCs） cells in RWVB were 3.7+0.3-fold, 5.1 + 1.2-fold, 5.2+0.4-fold and 13.9+1.2-fold respectively in RWVB, better than those in spinner flasks. After coculture, UCB-HSCs and UCB-MSCs can be easily separated by gravity sedimentation since UCB-MSCs are adhered on microcarriers. At the same time, the fibroblast-like cells appeared on the surface of GCSC microcarriers can be induced and differentiate into osteoblasts, chondrocytes and adipocytes and expressed MSCs-related surface markers of CD13, CD44, CD73 and CD105, while not hematpopietic cells-related surface markers of CD34, CD45 and HLA-DR, which is similarly to MSCs derived from bone marrow. In conclusion we have developed a feasible culture system to cocultivate UCB-HSCs and UCB-MSCs by adding cytokines and stromal cells together with GCSC microcarriers in RWVB.更多还原