The piRNA pathway is a conserved, small RNA-based immune system that guards genomic integrity against the harmful effects of transposon mobilisation in animal gonads. The majority of piRNAs originates from discrete source loci named piRNA clusters. In Drosophila germ cells, most piRNAs are produced from both genomic strands of so-called ‘dual-strand’ cluster loci. These loci, embedded in heterochromatin, are licenced by a protein complex composed of Rhino, Deadlock and Cutoff (RDC). Rhino is a fast-evolving heterochromatin protein 1 family member and while it has been shown to bind to the heterochromatic mark H3K9me3 at piRNA clusters, the determinants that specify the exclusive association of RDC with dual-strand clusters remained largely unknown. For the first part of this thesis, I performed a genetic screen in Drosophila germ cells to identify histone marks and modifiers important for Rhino binding at dual-strand clusters. I showed that the Eggless, a H3K9 methyltransferase, and Enhancer of Zeste, a modifier of H3K27, are essential for cluster definition. Using a combination of imaging, genetic and biochemical approaches, I established that both, H3K9me3 and H3K27me3, histone marks are present at dual-strand cluster loci and aimed to characterise RDC binding to these modifications. Transcripts produced from dual-strand clusters do not carry canonical hallmarks of RNA polymerase II transcription. Although capped, piRNA precursors lack splicing signatures and poly(A) tails. These features are essential for canonical mRNA export mediated by Nuclear export factor 1, suggesting that piRNA precursor export requires an alternative machinery. In the second part of this thesis, I show that dual-strand cluster precursor export requires the Drosophila Nuclear export factor family protein Nxf3. Nxf3 is recruited to piRNA clusters by Bootlegger, a novel piRNA factor characterised in this work. Nxf3 specifically binds piRNA cluster transcripts and is responsible for their export to peri-nuclear processing sites via a Crm1-dependent mechanism. This work uncovers a specialised export mechanism that bypasses the requirement for canonical mRNA processing and highlights the co-option of a nuclear export factor to prevent transposon mobilisation.