Tfam Knockdown Results in Reduction of mtDNA Copy Number, OXPHOS Deficiency and Abnormalities in Zebrafish Embryos

High mitochondrial DNA (mtDNA) copy numbers are essential for oogenesis and embryogenesis and correlate with fertility of oocytes and viability of embryos. To understand the pathology and mechanisms associated with low mtDNA copy numbers, we knocked down mitochondrial transcription factor A (Tfam), a regulator of mtDNA replication, during early zebrafish development. Reduction of Tfam using a splice-modifying morpholino (MO) resulted in a 42%±4% decrease in mtDNA copy number in embryos at 4 days post fertilization (4 dpf). Morphant embryos displayed abnormal development of the eye, brain, heart and muscle, as well as a 50%±11% decrease in ATP production. Transcriptome analysis revealed a decrease in protein-encoding transcripts from the heavy strand of the mtDNA. In addition, various RNA translation pathways were increased, indicating an upregulation of nuclear and mitochondria-related translation. The developmental defects observed were supported by a decreased expression of pathways related to eye development and haematopoiesis. The increase in mRNA translation might serve as a compensation mechanism, but appears insufficient during prolonged periods of mtDNA depletion, highlighting the importance of high mtDNA copy numbers for early development in zebrafish. For gene expression studies, 12 embryos for the injected controls, 12 embryos for the injected TFAM, and 9 embryos for the non-injected control group were individually collected in sterile tubes. Dye-swap hybridizations (2+2) were performed on microarray slides (4x44K zebrafish V3, Agilent) using gasket slides and a hybridization chamber and incubated for 17 h at 65ºC and 10 rpm in the hybridization oven (Agilent Technologies). Slides were washed with Triton X-102, freshly added to the Wash Buffers. Microarray slides were scanned using a DNA Microarray scanner with SureScan High-Resolution Technology (Model 2565CA, Agilent). The arrays contained 45,220 probes. Each probe identifier was transformed to Ensembl gene IDs (ENSDARG). This resulted in 36,156 probes containing a non-empty transcript ID of which 19,459 were unique transcripts and kept for the analysis. All transcripts were analysed using a multivariate Gaussian linear regression (MVN(mu,Sigma) where mu is the mean and Sigma is the covariance matrix) including slide differences (Slide), dye swap (Dye), background level (Bg), injection (Inj) and a random effect. The inference criterion used for comparing the models is their ability to predict the observed data, i.e. models are compared directly through their minimized minus log-likelihood. When the numbers of parameters in models differ, they are penalized by adding the number of estimated parameters, a form of the Akaike information criterion (Akaike 1973). For each transcript, a model containing the relevant covariates mentioned above (E(y) =Slide+Dye+Bg+Inj) was fitted in order to obtain a reference AIC. Then a model containing the treatment group (Trt) was fitted (E(y)=Slide+Dye+Bg+Inj+Inj:Trt). The transcript under consideration was found to be differentially expressed if the AIC of this second model decreased compared to the model not containing the treatment. These statistical analysis were performed using the freely available program R (Ihaka R 1996) and the publicly available libraries 'rmutil' and ‘growth’ (Lindsey 1999).