Synthesis of a dsRNA molecule complementary to D. melanogaster gene CG13617

In order to synthesize a specific dsRNA molecule for the D. melanogaster gene CG13617, a 1154 bp cDNA fragment comprising exons 3-5 was amplified by RT-PCR with primers DmE1-DmE2 (this gene has five exons in D. melanogaster, see Results) (FIGURE 11).

We chose the final part of the gene to design the dsRNA because it is much less conserved among species and it does not contain any of the characterized functional domains of the protein. This is important to avoid that the small RNA fragments that will be generated from the long dsRNA molecule bind additional mRNAs, which could cause unspecific silencing of other genes. This RT-PCR product was then cloned into the pGEM-T Easy vector (Promega) and the plasmid DNA was used as a template in a PCR reaction with primers

T7 promoter

Amplification product / Template DNA

T7 RNA pol RNA dsRNA

Embryo microinjection CG13617silencing by RNAi

D. melanogaster CG13617 _{DmE1 DmE2}

587 bp

T7DmE2 T7DmE3

RT-PCR + cloning

PCR

in vitrotranscription

GCTTCTAATACGACTCACTATAG T7 promoter

cDNA

pGEM-T vector

rNTPs

37 ºC 4 h

5’ 3’

FIGURE 11 | dsRNA synthesis in D. melanogaster. Blue rectangles represent CG13617 exons and green rectangles correspond to the T7 promoter sequences introduced in the PCR product. Primers are depicted as short black arrows. Yellow circles represent T7 RNA polymerase and two oppositely oriented parallel arrows symbolize the synthesized dsRNA molecule. Green arrows indicate the order of the different procedures.

T7DmE2-T7DmE3, which have a T7 promoter sequence (GCTTCTAATACGACTCACTATAG) added at their 5’ end, just after the corresponding CG13617 specific sequence (TABLE 5). We used cloned cDNA as a template in the PCR reaction (instead of cDNA directly) to be able to obtain easily large amounts of the amplification product (since the cDNA is amplified in the bacteria, the quantity of template no longer depends on the expression level of the gene). As only half of the primer sequence is able to bind the template DNA at the beginning of the amplification reaction, the annealing temperature was 57 ºC during the first 10 cycles and 60 ºC during the remaining 25 cycles. A PCR product containing 587 bp of CG13617 cDNA and a T7 promoter sequence at each end was generated. This PCR product was purified from an agarose gel using the QIAquick^® Gel Extraction Kit (Qiagen), cleaned with Phase Lock Gel^TM Light 1.5 ml tubes (Eppendorf), precipitated with ammonium acetate and ethanol, and resuspended in RNase-free water to be used as template in an in vitro transcription reaction with T7 RNA polymerase (MEGAscript^®

T7 Kit, Ambion). In this case, transcription starts at both ends of the DNA template because they both possess a promoter recognized by T7 RNA polymerase. The two complementary RNA molecules bind each other in the same synthesis reaction at 37 ºC to form dsRNA. After the in vitro transcription reaction, we used RNeasy^® Mini Kit (Qiagen) to remove proteins, ribonucleotides and template DNA and the newly synthesized dsRNA was dissolved in RNase-free injection buffer (0.1 mM sodium phosphate pH 7.8, 5 mM KCl). Finally, the dsRNA molecule was run in an agarose gel where it is expected to migrate in a way similar to dsDNA. The general guidelines for this procedure were obtained from CARTHEW (2003) and are illustrated in FIGURE 11.

2.11.2 Microinjection

D. melanogaster embryos for microinjection were collected when they were less than 1 h old in plates containing 1.5% agar dissolved in apple juice. It is essential to collect the embryos at a very early stage of development to ensure that they are in the syncytial blastoderm stage (a phase in which incomplete cell division causes the embryo to have many nuclei contained within a common cytoplasm), so the whole embryo acts as a single cell and the injected material can reach all the future cells of the individual. Embryos were dechorionized manually using double-sided adhesive tape adhered on a slide, were desiccated for 10 min at room temperature (to allow the introduction of the volume we want to inject), and were covered with halocarbon oil to avoid any posterior desiccation. Dechorionized embryos were then injected with the dsRNA complementary to the gene CG13617 at a concentration of 430 ng/μl, which was experimentally determined to be a dsRNA concentration that produced an effective gene silencing without having an extremely toxic effect for the embryos. Control embryos were also microinjected using a solution only with buffer to avoid introducing differences between samples caused by the injection process (rather than by the effects of the dsRNA itself), which could result in changes when expression profiles of the embryos are compared. After microinjection, embryos were kept in a humid chamber at room temperature covered in halocarbon oil until collection. It is important to take into account that a variable fraction of the microinjected embryos died during the process due to any of the manipulations they endured (dechorionation, desiccation or microinjection).

Samples of microinjected individuals were initially collected at three different developmental stages to check for the effectiveness of CG13617 silencing: ~20 h old embryos (including individuals that are already dead as well as alive embryos because it is impossible to distinguish them at this stage), first instar larvae that have just hatched, and third instar larvae that had been transferred to a vial with medium after hatching to continue their development.

For embryos and first instar larvae, all the embryos in the same slide (20-30 individuals) were collected as one sample. For third instar larvae, 2 individuals were enough to be able to extract sufficient RNA. For further analysis, we chose to use only the first instar larvae samples because we can ensure that we are working with RNA from tissues of individuals that had survived the microinjection process and that were alive at the moment of collection. Also, these larvae are closest to the embryonic stage, where CG13617 is known to be silenced in D.

buzzatii. Since first instar larvae samples were intended to be used for different gene expression experiments (microarrays, real-time RT-PCR) a larger amount of RNA was needed and 50-100 larvae were pooled together and collected as a single sample. After hatching, first instar larvae were collected following a protocol described in CARTHEW (2003) that includes a wash with heptane to eliminate halocarbon oil. First instar larvae were stored at -80 ºC until enough quantity was obtained to proceed with RNA isolation.

BOX 2 | RNA interference

Double-stranded RNA (dsRNA) has been revealed in recent years to be an important regulator of gene expression in many eukaryotes. It triggers different types of gene silencing that are collectively referred to as RNA silencing or RNA interference. The main feature that characterizes this mechanism is the presence of small ~20-30 nt non-coding RNAs able to regulate gene expression at different levels. As a general rule, this small RNAs serve as specificity factors that direct bound effector proteins to target nucleic acid molecules via base-pairing interactions.

The result of this process is an inhibitory effect on the gene expression of the target gene. The discovery of these mechanisms has led to the development of experimental techniques to knock out specific genes based on the introduction of dsRNAs to silence their expression. These procedures are known as RNA interference (RNAi) techniques and are widely used in multiple model organisms such as Caenorhabditis elegans, Drosophila or even human cells.

There are three main categories of short non-coding RNAs: short interfering RNAs (siRNAs), microRNAs (miRNAs) and piwi-interacting RNAs (piRNAs).

siRNAs and miRNAs come from double-stranded RNA precursors and are broadly distributed both phylogenetically and within the tissues of an organism.

Instead, piRNAs are found primarily in animals, they only function in the germ line, and seem to be derived from single-stranded precursors. It is also important to take into account that these two groups of molecules bind to distinct sets of effector proteins. There are also some differences between siRNAs and miRNAs:

miRNAs are processed from stem-loop precursors with incomplete double-stranded character that are purposefully expressed. They derive from a type of regulatory genes known as microRNA genes that produce non-coding transcripts with an imperfectly-paired hairpin structure. On the other hand, siRNAs are primarily exogenous in origin (they come from TEs, viruses, or transgenes) although they can be

BOX 2 | RNA interference (continued)

F^IGURE 12 | RNA interference mechanism. Several different categories of transcripts can adopt dsRNA structures that can be processed by Dicer into short (~21-23 nt) siRNAs (or miRNAs, if microRNA genes are the origin of the hairpin dsRNAs). To the left side are the exogenous sources of dsRNA molecules and to the right the endogenous ones. These RNA duplexes can be intra or intermolecular and although most are perfectly base-paired, some are not (for example, those coming from gene/pseudogene complexes or microRNA hairpin precursors). A siRNA or miRNA consists of a guide strand (red) which assembles into a functional RISC, and a passenger strand (blue), which is ejected and degraded. All forms of RISC contain the small RNA bound to an Ago protein and some additional factors. Target RNAs are then recognized by base-pairing, and silencing occurs through one of several mechanisms. In many species, the siRNA populations that engage a target can be amplified by the action of RNA-dependent RNA polymerase (RdRP) enzymes, strengthening and perpetuating the silencing response. Adapted from CARTHEW and SONTHEIMER (2009).

BOX 2 | RNA interference (continued)

endogenous too (see below) and they are excised from long, fully complementary dsRNAs. Here we will focus our attention on siRNAs and how they regulate gene expression in Drosophila.

The signature components of the RNA silencing machinery are: the Dicer enzymes, the Ago proteins and the ~21-23 nt siRNAs. The trigger is the presence of dsRNA molecules that can arise from multiple sources, which can be exogenous or endogenous. The exogenous dsRNA sources include transgene transcripts (trangenes can insert in the genome forming arrays made up of multiple copies with different orientations and their transcription can result in dsRNA formation), viral RNAs and experimentally introduced dsRNAs. The endogenous dsRNA molecules can be originated from convergent transcripts or natural sense-antisense pairs, the pairing of gene/pseudogene transcripts, microRNA genes and other hairpin RNA structures, or from repeated sequences that can be transcribed, like TEs or centromeres (F^IGURE 12, top).

All these dsRNA precursors need to be processed by the Dicer enzymes to form the small ~21-23 nt siRNAs. These enzymes have a PAZ domain that binds the end of the dsRNA molecule and two RNase III domains that cleave one strand each. The resulting product is a short dsRNA molecule ~21-23 nt long with ~2 nt 3’ overhangs. Many different siRNAs can be excised from a single long dsRNA molecule. Some organisms have only one Dicer enzyme, but others are capable of producing several of them. In D.

melanogaster, there are two distinct Dicers with functional specialization: Dicer-1 is required for miRNA biogenesis, while Dicer-2 is devoted mostly to the siRNA pathway. Dicer enzymes are usually associated to another protein with dsRNA-binding domains. In the case of Drosophila Dicer-2 this protein is R2D2.

The next step consists in the incorporation of these small RNAs into an RNA-induced silencing complex (RISC). The Argonaute (Ago) proteins are the central defining components of the various forms of RISC.

They also have a PAZ domain and possess a PIWI domain exclusive of this protein family. In Drosophila there are five Argonaute proteins with functional specialization, but numbers vary in different organisms. However, double-stranded siRNAs generated by Dicer cannot load directly into Argonaute proteins. These siRNAs enter into a RISC assembly pathway that involves duplex unwinding and culminates in the stable association of only one of the two strands with the Ago effector protein. This

Ago-associated strand will become the guide strand that directs target recognition by base pairing, while the other passenger strand is discarded. In Drosophila, in the first place, the R2D2/Dicer-2 heterodimer binds a siRNA duplex and then unknown factors are added to form the RISC-loading complex (RLC). This complex assembles with Ago2 to form pre-RISC and finally, Ago2 cleaves the passenger strand, that is ejected, and a functional RISC complex is formed. Several dsRNA-binding proteins are involved in this process. Strand selection is dictated by the relative thermodynamic stabilities of the two duplex ends: whichever strand has its 5’ terminus at the less stably base-paired end will be favored as the guide strand. siRNAs with equal base-pairing stabilities at their ends will incorporate either strand into RISC with approximately equal frequency.

The main mechanism of action of siRNAs is to cause gene silencing at the post-transcriptional level through the degradation of target mRNAs. The identities of the genes to be silenced are specified by this small RNA component of RISC. The siRNA guide strand directs RISC to perfectly complementary mRNA targets, which are then degraded by the PIWI domain of the Ago protein that cleaves the linkage between target nucleotides paired to siRNA nucleotides 10 and 11 (counting from the 5’ end). Then cellular exonucleases attack the fragments to complete the degradative process. The mRNA target dissociates from RISC after cleavage, leaving the protein complex free to cleave additional target molecules. Mismatches at or near the center of the siRNA/target duplex suppress the endonucleolytic cleavage, but the gene can still be silenced at post-transcriptional level by other mechanisms, such as translational repression, a pathway generally used by miRNAs. Silencing of imperfectly matched mRNAs in a way similar to how miRNAs act appears to account for most “off-target”

effects of siRNAs and is therefore of considerable importance.

In some organisms like C. elegans, primary siRNAs can induce the synthesis of secondary siRNAs through the action of an RNA-dependent RNA polymerase (RdRP) that uses siRNAs as primers to synthesize dsRNA with the target transcript as a template. This secondary siRNAs can amplify and sustain the silencing response, making it very strong. This amplification process involves the appearance of siRNAs corresponding to regions not included in the initial dsRNA trigger, so the lack of RdRP in insects and vertebrates makes the silencing mechanisms more specific in these species.

BOX 2 | RNA interference (continued)

As illustrated in FIGURE 12 (bottom), there are multiple mechanisms for siRNAs or miRNAs to cause the silencing of the target genes. Another important pathway known to take place in many species is the induction of heterochromatin formation with the consequent silencing of the affected genes. This transcriptional silencing was first reported in Saccharomyces pombe and also in plants as transcriptional gene silencing (TGS). It has been better characterized in S. pombe, where a RITS (RNA-induced transcriptional silencing) complex containing Ago1 is guided to specific loci such as centromeric repeats by bound siRNAs where it recognizes nascent transcripts thanks to an interaction between RITS and RNA pol II. RITS association promotes histone H3 methylation on lysine 9 by histone methyltransferases, which leads to the recruitment of Swi6 protein and chromatin condensation. Engagement of RITS to nascent

transcripts also activates RdRP, which generates secondary siRNAs able to spread the silencing.

Heterochromatinization causes DNA to be inaccessible for the transcription machinery and therefore, any genes included in the heterochromatin will be silenced to some extent.

In summary, in just a few years it has become clear that RNA interference pathways provide not only a completely new and unexpected mechanism to regulate gene expression and to defend the genome from invasive nucleic acids, but also have proven to be a very useful tool for biological research. These silencing mechanisms acting at some of the most important levels of genome function constitute a very active area of research and as more details of how this pathways work are discovered, new applications, such as their clinical use, will also be developed.

2.12 Microarrays

Gene expression levels of first instar larvae samples microinjected with the dsRNA that silences gene CG13617 expression (DSRNA) and control samples injected only with buffer (CONTROL) were compared using D. melanogaster oligonucleotide microarrays (GeneChip^® Drosophila Genome 2.0 Array, Affymetrix ). These microarrays contain 18880 probe sets (14 probes 25-nt long each) able to interrogate ~18500 transcripts. Labeling and hybridization were performed following the instructions of the manufacturer, starting from 3 μg of total RNA of four different samples: DSRNA1, DSRNA2, CONTROL1, CONTROL2 (FIGURE 13). Each sample was processed independently and hybridized to a different array.

For the four arrays quality measures fell within the usual limits and were similar between them.

To take into account the diversity of methods available to process the array information and calculate expression values, array results were analyzed using three different programs: GENECHIP^® OPERATING SOFTWARE (GCOS) version 1.4 (Affymetrix), RMA (IRIZARRY et al. 2003) as implemented in Bioconductor (GENTLEMAN et al. 2004), and DCHIP

version 2004 (LI and WONG 2001). In the GCOS analysis, first all arrays were normalized

separately to a same average intensity of 500 and pair-wise comparisons between the arrays of the control and embryos injected with dsRNA were generated. Then, probe sets showing expression differences between them were identified using the BULLFROG 5.3 program (ZAPALA et al. 2002) with the following criteria: a consistent call of increase/marginal increase or decrease/marginal decrease in the four pair-wise comparisons and an average fold change greater than 1.8. For the Bioconductor analysis, arrays were normalized by quantile normalization and expression values were calculated using the RMA method. The resulting expression values were then analyzed with the SAM (Significance Analysis of Microarrays) program (TUSHER et al. 2001) as two class unpaired data using default parameters. The list of differentially expressed genes was obtained by fixing a false discovery rate (FDR) of 10%

(delta = 0.5843) and all those with an average fold change between DSRNA and CONTROL

arrays greater than 1.8 were selected. In the DCHIP analysis, arrays were normalized to that with median intensity and expression values for each probe set were calculated as a model-based expression index using the PM/MM difference model and the program default parameters. The criteria used to identify probe sets with signal differences between experimental conditions were a fold-change greater than 1.5 using the lower bound of the 90% confidence interval, absolute difference between means greater than 50, and a t-test P-value lower than 0.05. The lists obtained with each method were combined and those probe sets differentially expressed in at least two of the three independent analyses were considered to be significant. Cluster analysis was carried out by average linkage hierarchical clustering using the CLUSTER and TREEVIEW programs (EISEN et al. 1998). Prior to the clustering, the hybridization signal from each probe set was median-centered and normalized across the samples, and the uncentered Pearson correlation was used as similarity metric. The gene ontology analysis of the differentially expressed genes was performed using the Functional Annotation Clustering tool at the DAVID Bioinformatics Resources NIAID/NIH webpage (DENNIS et al. 2003, HUANG et al. 2009) using the 41 differentially expressed probe sets and medium classification stringency (see Results).

FIGURE 13 | Labeling of RNA samples to be hybridized in Affymetrix microarrays for gene expression profiling. The different procedures are indicated at the left of the figure. Starting from total RNA samples, double stranded cDNA is synthesized adding a T7 promoter at the 3’ end. This double stranded cDNA is then used as a template in an in vitro transcription reaction with labeled ribonucleotides. In this reaction, biotin-labeled antisense complementary RNAs (cRNA) are generated. These cRNA molecules are the complementary strand to the initial cellular RNAs. After fragmentation, cRNAs are able to hybridize with the oligonucleotide probes contained in the microarray (which possess the original mRNA sequence). Figure from http://www.affymetrix.com.

BOX 3 | Oligonucleotide microarrays

DNA oligonucleotide microarrays are a technology used in molecular biology to assess the expression levels of thousands of genes simultaneously in a given

DNA oligonucleotide microarrays are a technology used in molecular biology to assess the expression levels of thousands of genes simultaneously in a given