天然产物紫杉醇是重要的抗癌药物，经济价值巨大，目前紫杉醇的工业生产依赖两个途径，即从天然红豆杉中提取，或利用紫杉醇前体进行化学半合成，这两个途径都受制于红豆杉树木原料供应，因此药物价格一直很高。利用真菌深层发酵生产紫杉醇可以大大提高供应，降低要价。真菌产生紫杉醇报道已有25年，有超过30种产紫杉醇真菌陆续被报道，但紫杉醇产率低、不稳定，大部分报道过的菌种因退化而失去产紫杉醇的能力，使利用真菌工业化生产紫杉醇一直无法实现。甚至真菌是否能够产生紫杉醇成为一个争议问题。

从植物内生真菌次级代谢产物中分离纯化天然产物是新药的重要来源之一。内生菌小孢拟盘多毛孢（Pestalotiopsis microspora）最初被鉴定为产紫杉醇真菌，这个属的真菌分布广泛，还能够产生丰富的其他次级代谢产物。本实验室分离得到一株能够产生紫杉醇的小孢拟盘多毛孢P. microspora NK17（NK17），但是，NK17产紫杉醇的产量低、不稳定，以至于很难检测到该产物。本论文对NK17中紫杉醇产生的可能影响因素进行探索，研究并揭示了造成紫杉醇低产甚至无法积累的原因。

我们首先对NK17主要次级代谢物进行鉴定，企望发现紫杉烷家族成员。HPLC图谱显示，NK17次级代谢产物主要有5个化合物吸收峰，其中包含已经被鉴定的Dibenzodioxocinone类化合物Pestalotiollide B。我们采用硅胶柱层析和制备型HPLC将这5种化合物分别进行分离纯化，并通过高分辨质谱（HRMS）、核磁分析（NMR）和圆二色谱（CD）对其结构进行鉴定，这几个化合物分别被鉴定为Pestalotiollide B（1），Pestalotiollide C（2），1’,2’-epoxy-3’,4’-didehydropenicillide（3），3’-methoxy-1’2’-dehydropenicillide（4）和1’,2’-dehydropenicillide（5），均为Dibenzodioxocinone类化合物，说明Dibenzodioxocinone类化合物是NK17的优势次级代谢产物。然后，我们对Dibenzodioxocinone低产菌株pgα1Δ进行mRNA-seq分析，发现一个包含21个基因的生物合成基因簇，其基因表达水平整体显著下降。通过基因敲除技术，我们确定其为Dibenzodioxocinone生物合成基因簇，其中一个关键的聚酮合酶基因pks8负责合成碳骨架，多个P450和其他修饰酶负责进行修饰并最终合成Dibenzodioxocinone类化合物。根据以上数据，我们推测了Dibenzodioxocinone类化合物的合成步骤。虽然以上研究获得了Dibenzodioxocinone类天然产物的有用知识，但是阻断其合成，没有观察到对紫杉烷类化合物的产生有积极影响，说明虽然Dibenzodioxocinone类化合物是NK17优势代谢产物，但是其与紫杉醇的生物合成在碳源和能量上可能不存在竞争关系。

目前公认的真菌中紫杉烷类化合物上游合成途径是与萜类合成相关的MVA途径。我们利用RT-PCR技术对NK17中与MVA途径相关的基因的表达情况进行分析，发现MVA途径中7个基因均有表达。紫杉醇的下游合成途径尚不完全明确，红豆杉中紫杉二烯合成酶（TS）催化紫杉醇合成途径中的第一步定向反应，编码该蛋白的基因为ts，我们在NK17中进行蛋白BLAST，结果发现，在NK17基因组中，存在一个基因，pts，与ts具有高度同源性，两者功能域之间具有一定的保守性，但pts缺少ts的一些功能域。为证明pts是否影响了紫杉醇的产生，我们对pts基因进行敲除，发现NK17生长变得非常缓慢，通过GC-MS检测到代谢产物中三萜化合物甾醇的峰消失。然后我们在NK17中异源表达红豆杉ts，发现这些表型均不能被异源表达的ts补偿。这个结果虽然无法确定NK17中的疑似紫杉二烯合成酶是否参与了紫杉醇的合成，但是说明了该基因可能参与了三萜化合物甾醇的生物合成，并与菌体的生长发育有很大关系。

在进行分子生物学研究的同时，我们尝试在NK17发酵时外源加入紫杉醇前体，如10-DAB，Baccatin III和紫杉醇本身，观察是否能够提高紫杉醇的合成。出乎意料的是，添加的紫杉醇前体没有转化成紫杉醇，反而被降解。此外，我们发现常温下NK17发酵液能够缓慢分解紫杉醇，50℃条件下30 min就能将紫杉醇完全分解，说明NK17发酵液中存在未知的能降解紫杉醇/烷的物质。通过大量摸索，最后我们利用葡聚糖凝胶柱层析分离纯化出该未知活性成分，并且通过NMR、质谱、红外、单晶衍射等方法确定其为KHCO₃。然后我们在试管中进行简单的分解试验，发现KHCO₃具有高效降解紫杉醇的特性，通过对降解产生的中间物结构进行分析，我们推测在甲醇溶液中，KHCO₃首先将紫杉醇分解为侧链的甲酯化产物和母环，然后进一步破坏紫杉醇二萜结构，最终导致紫杉醇完全分解。而当我们在培养基中添加CaCl₂时，外源添加的紫杉醇的分解得到有效缓解。然后，我们在NK17固体培养时加入0.2 g/L的CaCl₂，并对其代谢产物进行GC-MS分析，结果在发酵产物中检测到萜烯前体的积累。由于外加的CaCl₂可能会影响NK17的生长，用这种方法抑制HCO₃^—的积累有待更进一步探讨。以上结果表明，HCO₃^—的积累可能是内生真菌不能大量积累紫杉醇的真正原因。考虑到微生物深层培养会产生HCO₃^—，防止发酵液中HCO₃^—的大量积累可能是利用真菌大量生产紫杉醇的一个关键问题。

另外，我们发现在乙腈溶液中，紫杉醇被KHCO₃异构生成7-表紫杉醇，不继续进行分解反应。在该反应中，KHCO₃作为反应物而非催化剂。当KHCO₃浓度为紫杉醇10倍时，转化效率超过80%。7-表紫杉醇具有更强的抗肿瘤活性，这个技术有应用价值，为此申请了专利。

综上所述，本篇论文对产紫杉醇内生真菌NK17次级代谢产物进行系统研究，确定Dibenzodioxocinone类化合物是其优势代谢产物，通过基因敲除确定其生物合成基因簇，并预测其合成途径。此外，本论文还对代谢产物中紫杉醇不能大量积累的原因进行探究，通过分析NK17基因组中已报道的紫杉醇合成基因，发现其唯一的萜类合成酶PTS可能参与了三萜化合物如甾醇的合成，而非紫杉烷的合成；同时，我们发现并鉴定了NK17发酵液中能够分解紫杉醇的活性成分——碳酸氢钾，这可能揭示了真菌不能大量积累紫杉醇的真正原因。

Paclitaxel, (Taxol), was firstly isolated and identified from the bark of yew Taxus sp. Having been used widely in the clinic in the treatment of several types of tumours, taxol has huge economic value. There are two ways to make taxol at present at the industrial level, i.e. extracting from Taxus and semi-synthesis from its precursors Baccatin III or 10-deacetylbaccatin III (10-DAB). Both of the ways limite the production of the drug due to the use of the raw materials of yew trees, thus, resulting in the high cost of treatment with taxol. Large scale production using fungal fermentation promises an alternative technique to increase the suppley of the drug and to lower the treatment cost. The first taxol-producing fungus was reported in 1993; by far more than 30 isolates of taxol-producers were reported all over the world over the past 25 years. However, taxol production in these fungus were low and inconsistent, even in many cases, them lost the ablity to produce taxol, which made it difficult to produce taxol industrially by fungi fermentation. Whether fungi can produce taxol has become a disputable issue.

Natural products from endophytic fungi is one of the main sources for novel drug hunting. Pestalotiopsis microspora NK17 was originally identified as a taxol-producing strain and produces abundant other secondary metabolites as well. In this research, we will explore the factors effecting the biosynthesis of taxol in NK17 for digging out reasons for low yield and loss of the capability of taxol production in fungi.

To address the issues, we firstly purified and characterized the major secondary metabolites of NK17 that were shown high peaks in HPLC, hoping to find the taxanes. Five main compounds were isolated from the secondary metabolites of NK17 and purified by silica gel column chromatography and preparative HPLC. Utilizing techniques including high resolution mass spectrometry (HRMS), nuclear magnetic resonance (NMR) and circular dichroism (CD), we characterized compound 1 as pestalotiollide B (1), which was identified by our lab before, compound 2-5 were pestalotiollide C (2), 1', 2'-epoxy-3', 4'-didehydropenicillide (3) , 3'-methoxy-1'2'-dehydropenicillide (4) and 1', 2'- dehydropenicillide(5), in that order. All the compounds are dibenzodioxocinone derivatives, suggesting that the dibenzodioxocinones are the dominant metabolites in NK17. Dibenzodioxocinones are diphenyl ether lactone polyketides which were characterized as novel inhibitors agains cholesterol acyltransferase and cholesterol ester transfer protein, having potential of curing cardiovascular and cerebrovascular diseases. Further, we idientified and disrupted the biosynthesis pathway of dibenzodioxocinones in NK17 on the rationale that blocking the competing secondary pathways may benefit to the produce of taxanes. Through mRNA-seq profiling in the mutant pgα1Δ in which the G protien α subunit encoding gene was disrupted and produced little dibenzodioxocinones, a gene cluster that putatively harbored 21 genes including a polyketide synthase gene, designated as pks8, was defined. Knocking out genes in this cluster led to the loss of the dibenzodioxocinones. By the way, the biosynthetic pathway of dibenzodioxocinones was speculated. Additionally, some PKS genes over the genome were also affected the biosynthesis of dibenzodioxocinones. Unfortunately, blockage of the biosynthesis of dibenzodioxocinones had little influence on the produce of taxanes.

We also checked the upstream pathway that putatively leads to biosynthetic steps of taxanes, i.e. the MVA pathway for terpenoid biosynthesis. We analyzed the expression of MVA pathway-encoding genes by RT-PCR, to assure its expression under the culture conditions of the fungus. As a result, all genes in the MVA pathway were constitutively expressed. As the route leading to taxol production in yew and in fungi is still incompletely known, we searched for the gene that may encode the first enzyme, taxadiene synthase, in the biosynthesis of taxol, by a comparison of homology with the gene from Taxus (ts). In NK17, an ORF, named pts, shared remarkable similarity to Taxus gene ts. To test the function of the putative equivalent taxadiene synthase in NK17, we constructed pts-deleted strain, ptsΔ. The growth and development of ptsΔ strains was greatly slower, and ptsΔ strains showed weak resistance to fungal antibiotics, high-salt and nucleic acid antibiotics. The production of dibenzodioxocinone was also greatly reduced. In addition, none of these phenotypes can be compensated by expressing ts from Taxus. These results indicated that the taxadiene synthase in fungi not only participates in secondary metabolism, but plays an important role in the growth and development as well. However, the status of PTS in biosynthesis of taxol was still not sure and the genetic basis of taxol biosynthesis in NK17 is still unable to establish with certainty at this moment.

Meanwhile, we found that when taxol precursors, Baccatin III or 10-DAB, were added to the media in order to accumulate the final product taxol, surprisingly, they were decomposed, instead converted to taxol. When taxol was added to the culture, it was degraded in the fermentation broth of NK17. In vitro test showed that taxol could be quickly decomposed by the supernatant of the curlture of NK17 at 50°C for just 30 min. This result suggested that there is an unidentified agent produced by NK17 that has a capability to break down taxol. And this perhaps is the key reason that the fungus lost it capacity to accumulate taxol during growth. We made efforts to purified and acquired approximately 160 mg of this unknown agent, utilizing a number of combined techniques, e.g. Sephadex gel column chromatography from the fermentation broth. Its structure of the agent was determined to be KHCO₃ via NMR, infrared spectroscopy (IR) and single crystal X-ray diffraction (SCXRD) analysis. In the tube, we confirmed that taxol was firstly decomposed by KHCO₃ into two parts, 10-deacetylbaccatin III (10-DAB) and methyl esterized side chain of taxol. And consequently, 10-DAB was completely destroyed. In addition, we could stop the decomposition of the exogenously added taxol by adding CaCl₂ into the medium. This result may explain the underlying cause why endophytic fungi cannot stably produce paclitaxel in deep fermentation. And finding out the conditions for the taxol-producing fungi is a prerequisite for the application of this method in industry.

Notably, we also found that paclitaxel was transformed into 7-epi-paclitaxel, an isomeric form, in the presence of acetonitrile by KHCO₃. In this reaction, KHCO₃ acts as reactant rather than catalyst. Besides, when the concentration of KHCO₃ rises to 10 times of taxol, the conversion efficiency exceeds 80 percent. As 7-epi-paclitaxel shows stronger antitumor activity, which make the finding valuable for clinical treatment, thus we filed a petition for a patent on this finding.