内含子在真核生物中普遍存在，但是不同物种之间的内含子数量差异很大，这意味着内含子在进化历史上经历了动态的丢失与获得的变化。之前的研究普遍认为在近期内含子丢失的发生频率高于内含子获得，是塑造基因结构的主要力量。mRNA介导内含子丢失模型被认为是内含子丢失的一种机制。该模型认为，基因转录产生的mRNA分子被逆转录为cDNA，该cDNA与母基因发生重组导致了母基因内含子丢失。由mRNA介导的内含子丢失的特点是：丢失的内含子倾向于落在基因相邻位置上；丢失的内含子倾向于落在基因的3'端；丢失内含子的基因倾向于在性细胞中表达，并且表达量偏高。但是，这些经典证据均存在一些不完善的地方，没有直接从分子机制上证明内含子丢失是由mRNA介导的。 在本研究中，我们提出了一个新的策略用于检验mRNA介导内含子丢失模型：丢失内含子的基因容易产生加工型假基因（以下简称假基因）。这是由于假基因的产生与内含子丢失都需要依赖于mRNA的逆转录。因此，内含子丢失和假基因产生的正相关性可以作为支持mRNA介导内含子丢失模型的有利证据。之前的研究表明，哺乳动物假基因数量较多，而且近期发生在哺乳动物中的内含子丢失符合mRNA介导内含子丢失模型的一些经典特点，暗示这是哺乳动物内含子丢失的主要机制。小鼠和大鼠在哺乳动物中的内含子丢失速率相对较高。因此，我们选取了小鼠和大鼠作为研究的阳性样本，来检测内含子丢失和假基因之间是否存在正相关性。我们的研究结果发现，小鼠和大鼠丢失内含子的基因更容易产生假基因，并且产生假基因的拷贝数偏多；大鼠的假基因数量多于小鼠，而内含子丢失速率也高于小鼠，这些结果说明内含子丢失与假基因产生之间确实存在正相关性。我们还发现，在丢失内含子的基因与含有假基因拷贝的基因中编码可溶性蛋白的比例较高，而逆转录酶倾向于捕获在游离核糖体上翻译的mRNA分子，说明丢失内含子的基因转录的mRNA分子容易被逆转录。最后，我们还发现了三例由假基因和母基因重组而导致母基因内含子丢失的例子，直接说明了mRNA在内含子丢失过程中起到的中介作用。 此外，我们还以拟南芥为对象作了一个阴性对照研究。拟南芥有较高的近期内含子丢失速率，但是之前的研究显示拟南芥的内含子丢失不符合mRNA介导内含子丢失模型的经典证据。相应地，我们的研究结果发现，拟南芥丢失内含子的基因并不产生假基因。相反，拟南芥中存在着较多的不精确内含子丢失，即内含子丢失之后在相邻外显子额外删除或添加了一些DNA片段。这些证据说明拟南芥的内含子丢失主要是基因组水平上的DNA缺失造成的，没有以mRNA作为中介。 我们的结果表明，内含子丢失与假基因之间的相关性可以作为检测mRNA介导内含子丢失模型的新策略。该策略对于进一步研究真核生物基因结构进化的机制和规律具有重要的作用。

While introns are common features of eukaryotes, different species exhibit great varaitions of intron densities, indicating that dynamic losses and gains happen in intron evolutionary history. Previous studies suggest that recent intron losses are more frequent than gains and they are the main force to shape gene structures. The mRNA mediated intron loss model is suggested as one mechanism of intron loss. In this model, the mRNA molecules transcribed by parental genes are reverse-transcribed to cDNA molecules. The intronless cDNA molecules occationally recombine with the corresoponding genomic DNA, resulting in intron losses in parental genes. Several evidences are suggested to support the mRNA mediated intron loss model: Adjacent introns tend to be lost simultaneously; lost introns tend to be biased to 3' end of genes; intron-lost genes are likely to be germline expressed and highly expressed. However, none of these evidences are direct enough to explain that intron loss are actually mediated by mRNA molecules. In this study, we proposed a novel strategy to test the mRNA mediated intron loss model: intron-lost genes are likely to produce processed pseudogenes (PPs). PPs are by-products of reverse-transcription. If reverse-transcription is also required in intron loss, we would expect positive relationship between intron loss and PPs. Previous studies showed that PPs are abaunt in mammalian genomes and recent intron losses in mammals are in accorandce with several traditional evidences of the mRNA mediated intron loss model. Mice and rats have relatively high intron loss rates among mammalian species. Therefore, we used mice and rats to study the relationship of intron loss and PPs. Our results showed that intron-lost genes in mice and rats were more likely to produce PPs and produce more copies of PPs. In the genome-wide level, rats had higher PP abundance and also higher intron loss rate than mice. These evidence showed the positive relationship between intron loss and PPs. Our results also showed that both intron-lost genes and PP-produced genes were tend to encode soluable proteins. Soluable proteins are translationed on free ribosomes, the same place where reverse transcriptase are generated and acted. These results suggest that the mRNA molecules transcribed by intron-lost genes were more likely to be reverse-transcribed. Besides, we discovered three explicit cases where recombination between PPs and parental genes resulting intron losses. These were direct evidences supporing intron losses mediated by mRNA. We also converyed a negative control study using Arabidopsis thaliana, which also has high intron loss rates. Previous studies show that intron losses in Arabidopsis thaliana did not show traditional evidences of mRNA mediated intron loss model. Similarly, we found that none of the intron-lost genes in Arabidopsis thaliana produce any PPs. On the contrary, plenty of imprecise intron losses were discoveredArabidopsis thaliana. These results suggest that intron losses in Arabidopsis thaliana were mainly caused by genomic deletion, not mediated by mRNA. Our study results showed that exploring the relationship between intron losses and PPs could be a novel strategy to test the mRNA mediated model of intron loss. This strategy plays an important role in studying the evolution of eukaryotic gene structures.