Human de novo genes can originate from neutral long non-coding RNA (lncRNA) loci and are evolutionarily significant in general, yet how and why this all-or-nothing transition to functionality happens remains unclear. Here, in 74 human/hominoid-specific de novo genes, we identified distinctive U1 elements and RNA splice-related sequences accounting for RNA nuclear export, differentiating mRNAs from lncRNAs, and driving the origin of de novo genes from lncRNA loci. The polymorphic sites facilitating the lncRNA–mRNA conversion through regulating nuclear export are selectively constrained, maintaining a boundary that differentiates mRNAs from lncRNAs. The functional new genes actively passing through it thus showed a mode of pre-adaptive origin, in that they acquire functions along with the achievement of their coding potential. As a proof of concept, we verified the regulations of splicing and U1 recognition on the nuclear export efficiency of one of these genes, the ENSG00000205704, in human neural progenitor cells. Notably, knock-out or over-expression of this gene in human embryonic stem cells accelerates or delays the neuronal maturation of cortical organoids, respectively. The transgenic mice with ectopically expressed ENSG00000205704 showed enlarged brains with cortical expansion. We thus demonstrate the key roles of nuclear export in de novo gene origin. These newly originated genes should reflect the novel uniqueness of human brain development.

人类从头基因可以起源于中性长非编码RNA（lncRNA）基因座，并且总体上具有进化意义，但这种全有或全无的功能转变如何以及为何发生仍不清楚。在这里，在 74 个人类/类人猿特异性 de novo 基因中，我们鉴定了独特的 U1 元件和 RNA 剪接相关序列，这些序列负责 RNA 核输出、区分 mRNA 与 lncRNA，并驱动 lncRNA 位点的 de novo 基因起源。通过调节核输出促进 lncRNA-mRNA 转换的多态性位点受到选择性限制，从而维持区分 mRNA 和 lncRNA 的边界。因此，积极通过它的功能性新基因表现出一种预适应起源的模式，因为它们在实现其编码潜力的同时获得了功能。作为概念验证，我们验证了人类神经祖细胞中这些基因ENSG00000205704之一的剪接和 U1 识别对核输出效率的调节。值得注意的是，在人胚胎干细胞中敲除或过度表达该基因分别会加速或延迟皮质类器官的神经元成熟。异位表达ENSG00000205704的转基因小鼠表现出大脑增大和皮质扩张。因此，我们证明了核输出在基因从头起源中的关键作用。这些新产生的基因应该反映了人类大脑发育的新颖独特性。