Optical sectioning for 3D super-resolution microscopy

几个世纪以来，显微镜一直是促进我们对生物学的理解的重要工具。最近的发展将显微镜分辨近距离物体的能力提高了20倍(2014年诺贝尔奖表彰)，使人们有可能在纳米尺度上成像薄样品，纳米尺度是单个蛋白质分子的大小。然而，由于来自图像平面上下的大量散焦光以及当光穿过样本时引入的像差，在组织和器官等厚样本中执行细胞结构的超分辨率成像仍然具有挑战性。我们最近建造了一台最先进的超分辨率显微镜，它使用最先为哈勃太空望远镜开发的自适应反射镜技术来消除像差。我们已经证明了这种4PI-SMS显微镜可以对细胞进行分子分辨率的成像，但如果样品太厚，其灵敏度和分辨率会急剧下降。现在，我们建议使用另一种称为时空聚焦的新方法来解决厚样品中背景光离焦的问题。它使用了一种特殊的激光，它产生的光脉冲的波长是激发显微镜探测到的荧光分子所需波长的两倍(能量的一半)。通过整形这些脉冲，我们可以创造出这样的条件，即唯一需要激发的荧光分子位于一层薄片(1-2微米厚)中，来自激光的两个光子同时激活它们。这意味着我们只激发我们想要检测的分子，并将背景的焦外光减少约100倍。将时空聚焦纳入4PI短信息系统将使对更复杂的样品，如组织或有机物质进行定量超分辨率成像成为可能，这将允许解决一系列全新的问题。作为原理的证明，我们将测试这台显微镜能否很好地成像果蝇卵巢和小鼠肠道器官中的单分子，因为这些样本很容易获得，厚度超过50微米。更具体地说，我们计划研究一组保守的极性蛋白质的分子组织，这些蛋白质使细胞的一侧与另一侧不同。这将为时空聚焦4PI-SM提供一个具有挑战性的测试，因为其中几个蛋白质定位于果蝇和哺乳动物上皮细胞的顶端，因此位于样本中超过10微米深。以20纳米的精度可视化这些蛋白质，并能够计算复合体中的分子数量，将使我们能够回答该领域的主要悬而未决的问题。例如，我们计划研究如何确定上皮细胞顶端和侧边之间的边界，以及关键的顶端极性因子--非典型蛋白激酶C如何被招募到顶膜。4PI-SMS显微镜使人们能够看到组织内部用其他光学显微镜方法看不到的结构，因此很难预测我们可能会发现什么新的特征。

Microscopy has been an essential tool in advancing our understanding of biology for centuries. Recent developments have improved the ability of microscopes to resolve close objects by a factor of >20 (recognised with the Nobel Prize in 2014), making it possible to image thin samples at the nanometre scale, which is the size of individual protein molecules. It has remained challenging, however, to perform super-resolution imaging of cellular structures in thick samples, such as tissues and organs, because of the large amount of out of focus light from above and below the image plane and the aberrations that are introduced as the light passes through the sample. We have recently built a state of the art super-resolution microscope that removes the aberrations using adaptive mirror technology that was first developed for the Hubble space telescope. We have demonstrated that this 4Pi-SMS microscope can image cells at molecular resolution, but its sensitivity and resolution deteriorate dramatically if the samples are too thick.We now propose to address the problem of out of focus background light in thick samples using another new approach called spatio-temporal focusing. This uses a special laser that produces pulses of light at twice the wavelength (and half of the energy) needed to excite the fluorescent molecules that the microscope detects. By shaping these pulses, we can produce conditions where the only fluorescent molecules to be excited lie in a thin sheet (1-2 micrometres thick) where two photons from the laser activate them simultaneously. This means that we only excite the molecules that we want to detect and reduces the background out of focus light by a factor of ~100. Incorporating spatio-temporal focusing into the 4pi SMS system will make it possible to perform quantitative super-resolution imaging on more complex samples, such as tissues or organoids, which will allow a whole new range of questions to be addressed. As a proof of principle, we will test how well this microscope can image single molecules in the ovaries of the fruitfly, Drosophila, and in mouse intestinal organoids, as these samples are readily available and are more than 50 micrometres thick. More specifically, we plan to investigate the molecular organisation of a conserved set of polarity proteins that make one side of a cell different from the other. This will provide a challenging test for the spatio-temporal focusing 4Pi-SMS, because several of these proteins are localised on the apical side of epithelial cells in both flies and mammals and therefore lie more than 10 micrometres deep in the sample. Visualising these proteins with 20 nanometre precision and being able to count the number of molecules in a complex will allow us to answer major open questions in the field. For example, we plan to investigate how the boundary between the apical and lateral sides of epithelial cells is specified and how the key apical polarity factor, atypical protein kinase C, is recruited to the apical membrane. The 4Pi-SMS microscope allows one to see structures inside tissues that are not visible with other light microscopy methods, and it is therefore hard to predict what new features we may find.