DHX34 and NBAS form part of an autoregulatory NMD circuit that regulates endogenous RNA targets in human cells, zebrafish and Caenorhabditis elegans

Abstract

The nonsense-mediated mRNA decay (NMD) pathway selectively degrades mRNAs harboring premature termination codons but also regulates the abundance of cellular RNAs. We sought to identify transcripts that are regulated by two novel NMD factors, DHX34 and neuroblastoma amplified sequence (NBAS), which were identified in a genome-wide RNA interference screen in Caenorhabditis elegans and later shown to mediate NMD in vertebrates. We performed microarray expression profile analysis in human cells, zebrafish embryos and C. elegans that were individually depleted of these factors. Our analysis revealed that a significant proportion of genes are co-regulated by DHX34, NBAS and core NMD factors in these three organisms. Further analysis indicates that NMD modulates cellular stress response pathways and membrane trafficking across species. Interestingly, transcripts encoding different NMD factors were sensitive to DHX34 and NBAS depletion, suggesting that these factors participate in a conserved NMD negative feedback regulatory loop, as was recently described for core NMD factors. In summary, we find that DHX34 and NBAS act in concert with core NMD factors to co-regulate a large number of endogenous RNA targets. Furthermore, the conservation of a mechanism to tightly control NMD homeostasis across different species highlights the importance of the NMD response in the control of gene expression.

Figure 1.

DHX34, NBAS and UPF1 regulate a common set of transcripts. Venn diagrams illustrate the overlap among genes upregulated by each of these three NMD factors in (A) human cells (B) Zebrafish and (C) C. elegans. Areas are proportional to the number of regulated genes. In each species, target coregulation is significant (P < 0.0001 for all individual overlaps).

Figure 2.

Scatter plots show a significant positive correlation between the level of regulation of target genes by DHX34 and NBAS compared with the core NMD factor, UPF1 in (A) human cells (B) Zebrafish (C) C. elegans. Each dot corresponds to a regulated gene and black dots represent coregulated genes. Spearman’s rho and P-values are indicated.

Figure 3.

Validation of targets identified in the human microarrays. RNA levels of selected targets of (A) DHX34, (B) NBAS and (C) UPF1 were analyzed by RT-qPCR (black bars) and compared with the microarray data (gray bars) in HeLa cells depleted of the respective factors. In all panels, RNA levels of target genes in mock-depleted HeLa cells were used as a control (white bars). RT-qPCR expression values were normalized against three reference genes with constant expression, and values in control samples were set to one. Matrix GLA protein (MGP) was selected due to its marked upregulation upon NBAS knockdown in the microarray analysis. CTF1 (cardiotrophin 1) was chosen as a control due to its constant expression level in all data sets (Supplementary Table S6). An asterisk next to the gene name indicates the presence of an EnsEMBL-annotated NMD feature.

Figure 4.

NMD regulates the expression of genes associated with cellular stress response pathways and membrane trafficking. Functional categories that are significantly enriched within DHX34, NBAS and UPF1 upregulated genes (relative to all expressed genes) are indicated. Functional categories analyzed were Sequence Feature (SF), GO Biological Process (BP) and KEGG Pathway (KP).

Figure 5.

Autoregulation of the NMD pathway in human cells. DHX34 and NBAS contribute to the negative feedback-loop regulating levels of transcripts that encode NMD factors. RT-qPCR analysis of total cellular RNA from HeLa cells depleted of (A) UPF1, (B) DHX34 and (D) NBAS. The values shown are the average fold-change (mean ± SEM) from three independent experiments relative to mock-depleted cells (control). Statistical analysis was performed using the Student’s t-test. *P < 0.05; **P < 0.01; ***P < 0.001. (C) and (E) Western blot analysis shows the upregulation of UPF1 at the protein level upon depletion of DHX34 or NBAS, respectively, relative to mock-depleted cells (LUC, shRNA against luciferase transcript).

Figure 6.

The NMD negative feedback loop is conserved in Danio rerio. Zebrafish embryos were injected with MO targeting Dhx34 (A), Nbas (B) or Upf1 (C). As a specificity control for each gene, a 5 bp mismatch MO (cont MO) was also injected. The levels of zebrafish NMD factor mRNAs were analyzed by RT-qPCR relative to a control. The values represent an average of three independent injection experiments (mean ± SEM). Because the MOs affect only the protein level of targeted genes, the upregulation of targeted gene transcripts is a further evidence for a feedback loop regulation.