參考文章是：The cis-regulatory dynamics of embryonic development at single-cell resolution
概述
單細(xì)胞水平的發(fā)育譜系調(diào)控機(jī)制：
Here we investigate the dynamics of chromatin regulatory landscapes during embryogenesis at single-cell resolution
作者利用他們以前開發(fā)的sci-ATAC技術(shù)，對果蠅胚胎超過20000個(gè)單細(xì)胞核追蹤三個(gè)主要發(fā)育階段的染色質(zhì)圖譜：
2–4 h after egg laying (predominantly stage 5 blastoderm nuclei), 此時(shí)的胚胎由約6000個(gè)多能細(xì)胞組成
6–8 h after egg laying (predominantly stage 10–11)，此時(shí)中胚層和內(nèi)胚層的基本譜系已確定
10–12 h after egg laying (predominantly stage 13), 此時(shí)每個(gè)胚胎的20000多個(gè)細(xì)胞正進(jìn)行終端分化
Results
對于每個(gè)發(fā)育階段，所采樣的細(xì)胞核來自幾百個(gè)胚胎（雌雄都有），數(shù)據(jù)的質(zhì)量衡量包括barcode的reads數(shù)分布、片段大小、平均每個(gè)細(xì)胞的reads數(shù)、與以前的研究中定義的DHS（DNase超敏位點(diǎn)）的cover率等等。
通過把基因組分成2-kb的windows，對每個(gè)細(xì)胞的每個(gè)bin進(jìn)行評分，選取其中20,000 most frequently accessible windows使用LSI（latent semantic indexing）算法初步聚類（因?yàn)閱渭?xì)胞數(shù)據(jù)的稀疏特征，需要合并一些bins的數(shù)據(jù)）
如上圖，根據(jù)細(xì)胞全部windows的相似性進(jìn)行聚類，聚類獲得不同cell clade，每個(gè)clade的細(xì)胞的reads即可合并作為peaks calling的input。
clade的處理流程：合并clade的reads，進(jìn)行peaks calling，類似差異分析的方法尋找clade特異的peaks，這些peaks再用于尋找peaks link(比如enhancer和其相關(guān)聯(lián)的promoter)，結(jié)合enhancer胚胎活性數(shù)據(jù)庫、基因表達(dá)數(shù)據(jù)庫驗(yàn)證這些links
每個(gè)clade中富集的特異peaks及l(fā)inks可以用來鑒定不同的clade并用來做注釋，舉例來說：
mesoderm is split into myogenic mesoderm (clade 3) and non-myogenic mesoderm (such as fat body and haemocytes) combined with endoderm (clade 4). The latter indicates that non-myogenic mesoderm and endoderm exhibit similar chromatin accessibility, suggesting a shared developmental program.
注釋以后，作者發(fā)現(xiàn)原來的non-myogenic mesoderm和 endoderm是合并在同一個(gè)clade里面的，說明這兩種類型可能有共有的shared developmental program. 不過就以前的知識，Drosophila的中胚層和內(nèi)胚層并沒有共同的起源，作者給出的解釋是：可能是進(jìn)化遺留的特征
Although, to our knowledge, Drosophila mesoderm and endoderm have not been shown to share a common origin, this is highly reminiscent of the mesendoderm lineage in Caenorhabditis elegans, sea urchins and vertebrates.
為了驗(yàn)證clade assignments的可靠性，作者結(jié)合了FACS和motif enrichment進(jìn)行了驗(yàn)證：
上圖e中用FACS篩選特定細(xì)胞類群，然后進(jìn)行(DNase-seq)，與sci-ATAC的結(jié)果進(jìn)行比較
t-SNE聚類以及每個(gè)cluster的different developmental stages，這里作者對2-4h的細(xì)胞進(jìn)行t-SNE聚類，發(fā)現(xiàn)細(xì)胞的聚類結(jié)果和不同發(fā)育階段的細(xì)胞類型高度一致，在下圖b中，展示了每個(gè)cluster都有的階段特異enhancer活性，說明developmental time是造成這些細(xì)胞聚成不同類的主要因素。
當(dāng)然還是少不了trajectory擬時(shí)分析：可以看到在上圖c中，細(xì)胞最終分成了三支，分別對應(yīng)ectoderm、endoderm、mesoderm，說明在2-4h的發(fā)育階段后期，細(xì)胞逐漸分化成三支
Notably, the trajectory split cells into three major branches that were consistent with our annotations of the major germ layers (neuronal cells are rare at this time point, as expected)
利用擬時(shí)分析，還可以鑒定出那些在擬時(shí)序列中動態(tài)變化的enhancer和gene loci，比如slam這個(gè)基因逐漸關(guān)閉的動態(tài)：
For example, the most significant closing site (P value = 5 × 10?224) is within the slam locus, a gene that is essential for blastoderm cellularization during a very brief temporal window
上圖d展示了隨擬時(shí)序列變化的brancn specific sites，e和f展示的是對應(yīng)的anterior enhancer和posterior enhancer分別驅(qū)動具有空間表達(dá)特異性的gap genes(knirps和giant)的表達(dá)，說明：
sci-ATAC-seq can identify regulatory regions that are specifically accessible in spatially refined subsets of cells without the need for FACS sorting.
胚胎發(fā)育研究中經(jīng)典的lineage-tracing和transplantation experiments都揭示了細(xì)胞命運(yùn)決定主要發(fā)生在blastoderm時(shí)期，因此有了blastoderm fate map這一概念。利用sci-ATAC技術(shù)，進(jìn)一步很好地描繪了胚胎發(fā)育時(shí)期染色質(zhì)可及性的空間異質(zhì)性。spatial heterogeneity in chromatin accessibility.
作者進(jìn)一步對lineage commitment(6–8 h) 和differentiation (10–12 h)兩個(gè)胚胎發(fā)育時(shí)期進(jìn)行探索。通過對較晚期胚胎發(fā)育階段的數(shù)據(jù)應(yīng)用t-SNE，展示出更精細(xì)的細(xì)胞類型、組織圖譜（這也與細(xì)胞進(jìn)入分化的特征相符合）
Clusters were annotated based on overlaps between cluster-enriched peaks and enhancers or genes with known tissue-specific activity.
從上圖中可以發(fā)現(xiàn)clade和更精細(xì)的cell annotation的對應(yīng)關(guān)系、包含關(guān)系（比如a圖的mesendoderm分支成8、16、14三支）
A major advantage of profiling chromatin accessibility is its potential to identify distal regulatory elements that shape gene expression.
為了驗(yàn)證那些組織特異性peaks確實(shí)是一些組織特異的enhancer，作者進(jìn)行了胚胎轉(zhuǎn)基因?qū)嶒?yàn)（體內(nèi)enhancer活性驗(yàn)證，lacZ reporter gene），大概就是對candidate基因組區(qū)域進(jìn)行PCR擴(kuò)增、克隆至hsp70 promoter(驅(qū)動lacZ報(bào)告基因)上游，然后注射入胚胎、整合。后期在candidate有活性的胚胎區(qū)域?qū)袌?bào)告基因表達(dá)
We obtained 31 transgenic lines, representing six candidate regions with specific accessibility in neurogenic ectoderm, ten in non-neurogenic ectoderm, eight in myogenic mesoderm and seven in non-myogenic mesoderm plus endoderm.
candidate的篩選根據(jù)不同clade的開放peaks進(jìn)行選擇。作者發(fā)現(xiàn)一些mesendoderm，clade4的candidates同樣在yolk nuclei也有活性，然而yolk實(shí)際上是胚外組織，理論上不應(yīng)該有報(bào)告基因表達(dá)，作者給出的解釋是：
As the yolk is extra- embryonic, this was unexpected, and suggests a potential regulatory link between the yolk and mesendodermal tissues, which is supported by the role of the GATA transcription factor serpent in both yolk and nonmyogenic mesoderm
總結(jié)
sci-ATAC-seq不僅能解釋胚胎發(fā)育過程中動態(tài)的染色質(zhì)可及性，還能大規(guī)模地預(yù)測體內(nèi)活性的enhancer。作者還提供了一個(gè)網(wǎng)頁工具：http://shiny.furlonglab.embl.de/scATACseqBrowser/
Our ability to understand how changes in the regulatory landscape underlie lineage commitment would be greatly aided by the concurrent measurement of chromatin accessibility and transcription.
In the long term, the integration of chromatin state, transcriptional output, lineage history, and spatial information at single-cell resolution has the potential to unlock how an organism’s genome encodes its development.
整合單細(xì)胞水平的染色質(zhì)狀態(tài)、轉(zhuǎn)錄譜、發(fā)育軌跡、空間信息等數(shù)據(jù)，將進(jìn)一步有利于解答發(fā)育生物學(xué)的問題
參考文獻(xiàn)
The cis-regulatory dynamics of embryonic development at single-cell resolution：https://doi.org/10.1038/nature25981
