*3.1. Selection and Comparability of Microarray Data Sets*

Recently, we identified a set of retrograde controlled genes that are repressed or induced by biogenic signals in the light when chloroplast biogenesis is blocked at the level of plastid transcription [25]. We were interested in understanding how similar or different these gene groups are in comparison to conditions when the block of chloroplast biogenesis occurs at the level of plastid translation or at a level of general destruction through oxidative

damage. To this end we compared our data with results from two investigating the effects of LIN and NF on nuclear gene expression [26,27]. The experimental design of all three studies was highly similar using *Arabidopsis* seedlings grown for 5–7 days on sugar-supplemented medium in Petri dishes and exposed to light to visualize the effect of blocked chloroplast biogenesis on light-regulated gene expression profiles. In LIN-based experiments seedlings were grown in absence and presence of 0.5 mM LIN under extremely weak blue-red light of 0.5 μE fluence rate for 6 days followed by a shift to 60 μE blue-red light for 24 h [27]. In the NF experiment seedlings were grown in absence and presence of 5 μM NF for 3 days in the dark followed by 3 days in white light [26]. In both studies the inhibitors, thus, had sufficient time to block the respective process before induction of chloroplast biogenesis. In our own experiments with the genetically blocked *pap7-1* mutant seedlings were grown for 5 days directly under light since the genetic inactivation had become effective already during the establishment of the seeds [25]. Therefore, in all three experiments the effect of a block in chloroplast biogenesis on light-controlled gene expression was studied at a comparable developmental stage of the seedlings (2 cotyledons stage). Furthermore, all studies used the Affymetrix ATH1 *Arabidopsis* chips providing high technical comparability. We, therefore, regarded the experimental set-ups as sufficiently similar to provide reliable data for a meta-analysis.
