結合數學模擬與實驗探討線蟲細胞遷移型態的異質性

Abstract

基因表現的異質性已經在細胞間普遍地被發現；然而，基因表現的異質性如何導致生物表現型態的差異仍是未解之謎。利用實驗結合數學模型，我們研究線蟲遠頂細胞中的基因表現噪音(noise)與細胞遷移表現型的差異。在線蟲中， unc-5接受器的表現決定了遠頂細胞在幼蟲時期背向遷移的時間。我們先利用實驗建構了一個調控unc-5接受器表現的基因網路；同時，我們建構的一個模擬遠頂細胞遷移的數學模型，並且決定網路內基因調節的運算邏輯結構。我們發現在調控unc-5接受器表現的基因網路中，結合了許多feed-forward loop的模組結構。我們模擬了在feed-forward loop組合模組中基因表達噪音傳遞的行為，研究結果發現feed-forward loop組合模組可以同時過濾上游傳遞訊號在開啟或關閉時候所產生的噪音。我們假設基因表現噪音導致了在特定突變株線蟲中所觀察到遠頂細胞背向遷移時間的差異；為了驗證這個假設，我們模擬了unc-5在基因網路中隨機表現的過程。我們發現遠頂細胞背向遷移時間點是由穩定表現UNC-5的蛋白質而非震盪激烈的unc-5 mRNA所決定。我們的實驗結合理論發現在遠頂細胞中存在一個UNC-5蛋白質的閾值(threshold)決定遠頂細胞的背向遷移行為，利用所找到的閾值與數學參數，我們的數學模型可以準確地預測在dre-1突變型線蟲中延遲的背向遷移行為。特過實驗結合數學模型模擬的方式，我們發現了一個 blmp-1對lin-29的正向調節機制，使得blmp-1可以促進遠頂細胞的背向遷移行為。blmp-1;daf-12突變型線蟲中，UNC-5蛋白質的噪音在閾值(threshold)間震盪，進而導致了提早與延遲的背向細胞遷移行為在此突變中可以同時發生。上游調節基因的噪音可能傳遞進而影響下游基因的表現，進而導致了背向細胞表現型態的差異。這個研究提出了一個創新的分子機制說明基因動態表現與噪音如何同時導致細胞表現型態的異質性。

Gene expression noise has been observed in the single-cell level, and it’s been studied in the heterogeneity among cells. How such heterogeneity contributes to the phenotypic variations is not clear. Using experimental and computational approaches, we studied the gene expression noise and phenotypic variations of distal tip cells (DTCs) migration in C. elegans. We identified the gene regulatory network that regulates the expression of unc-5 receptor, which spatiotemporally control the dorsalward migration of DTC at the third larva stage. To simulate the DTC migration timings, the regulatory logics in gene regulation were determined. We found that the interlinked feed-forward loops in the network is able to filter noises regardless of the input signals’ states. To test the hypothesis that noise contributes to phenotypic variations in mutants, we simulated stochastic unc-5 expression with its gene regulatory network. In parallel, unc-5 mRNA and protein in single DTC was quantified. Our results showed that the unc-5 transcription is too noisy to directly link to DTC migration phenotype, and suggested that DTC dorsalward turning time is determined by the level of UNC-5 protein reaching a threshold. We successfully determined a UNC-5 threshold in mathematical model and reproduced the DTC migration phenotypes in wild type and mutants. An unexpected retarded DTC migration phenotype in dre-1 mutants was predicted in our model, and further validated by experiments. Both experiments and simulations indicated that blmp-1 can activate lin-29 expression to enhance UNC-5 expression exceeding the threshold. In the blmp-1;daf-12 mutants with heterogeneous phenotypes, UNC-5 is produced near the threshold, and its expression noise brings UNC-5 above or below the threshold, generating phenotypic variations. Suppressing the upstream gene noise reduces heterogeneity in this mutant. Our work reveals a novel mechanism where dynamics and noise together lead to phenotypic variations.