E. coli is a commonly used bacterium for dsRNA expression because of its clear genetic background and convenient genetic manipulation. E. coli strain HT115 (DE3), which is RNase III deficient, and L4440 vector with a pair of oppositely oriented T7 promoters (one on each side of the multiple cloning site) are widely used as an expression strain and vector for dsRNA production (Timmons et al., 2001; Voloudakis et al., 2015). After introduction of the L4440 vector ligated with the target fragment into strain HT115 (DE3), large amounts of T7 RNA polymerase can be synthesized on induction by isopropyl β-D-1-thiogalactopyranoside (IPTG); the T7 RNA polymerase binds to the T7 promoter in L4440, which mediates the transcription of downstream DNA sequences into RNA. As a result, two complementary RNAs are synthesized, which in turn form the target dsRNA (Voloudakis et al., 2015).

The production of dsRNA using engineered bacterial expression was first attempted by Timmons and Fire (1998), and the corresponding RNAi phenotype was verified after feeding to the nematode Caenorhabditis elegans, showing that dsRNA expressed by bacteria can induce RNAi effects (Timmons et al., 2001). Using this dsRNA generation strategy, RNAi effects were induced in a variety of insects, such as Spodoptera exigua (Tian et al., 2009), Bactrocera dorsalis (Li et al., 2011), Chilo infuscatellus (Zhang et al., 2012), Spodoptera exigua (Vatanparast and Kim, 2017), Plagiodera versicolora (Zhang et al., 2019), Tuta absoluta (Bento et al., 2020), Harmonia axyridis (Ma et al., 2020), Spodoptera littoralis (Caccia et al., 2020), Agrilus planipennis (Leelesh and Rieske, 2020). In addition, expressed virus dsRNA can protect a plant or animal against viral infection. For example, E. coli strain HT115 (DE3) was used to express dsRNA of the Chinese Sacbrood Virus (CSBV) VP1, which was fed to Chinese honeybees (Apis cerana) and effectively prevented the virus infecting the bees (Zhang et al., 2016). Treating Nicotiana benthamiana with dsRNAs of fragments of two major plant viruses, Pepper Mild Mottle Virus (PMMoV) and Plum Pox Virus (PPV), effectively reduced the infection of N. benthamiana by these two viruses (Tenllado et al., 2003). All these results show that engineered E. coli can synthesize dsRNAs, and the dsRNA produced can induce RNAi effects in the corresponding target organisms.

The yield of dsRNA synthesized in E. coli has been improved over time. An average of 4 μg of dsRNA was obtained per ml of E. coli culture in 2003 (Tenllado et al., 2003), and 45 μg hairpin dsRNA per ml of bacteria (optical density at 600 nm = 1) in 2013 (Posiri et al., 2013). The increase of dsRNA yield is due to the fermentation methods and operation parameters used (Thammasorn et al., 2015; Papic et al., 2018). dsRNA production using batch fermentation and fed-batch fermentation was compared in a 10 L fermenter, and the dsRNA titer in fed-batch fermentation (95.0 ± 21.5 μg/ml) was nearly 30-fold that found in batch fermentation (3.4 ± 0.5 μg/ml) (Thammasorn et al., 2015). The nutrition can also affect the final dsRNA yield, and the production of dsRNA using Terrific broth (TB) (6.2 ± 0.2 μg/ml) was higher than that using Luria-Bertani (LB) broth (2.6 ± 0.8 μg/ml). After further optimization, the yield was close to 0.06 g/g, the maximum production rate reached 11.1 mg L⁻¹ h⁻¹ by batch fermentation and 15.2 mg L⁻¹ h⁻¹ by fed-batch fermentation (Papic et al., 2018). Therefore, the dsRNA yield is related to bacterial growth, and fed-batch fermentation resulted in a higher dsRNA yield by sustainably supplying nutrition.

Modification of the expression vector and host strain can further improve the efficiency of dsRNA synthesis. dsRNA production using a new E. coli expression system, pET28-BL21 (DE3) RNase III- was thrice than that of L4440-HT115 (DE3) (Ma et al., 2020).

Moreover, extraction methods are closely linked to the yield of dsRNA. The titer of dsRNA extracted from E. coli by ultrasonic crushing and phenol extraction was 19.5 μg/ml, while sonication and heating before dsRNA extraction increased the titer of dsRNA by 2.5- to 5- fold (Ahn et al., 2019).

Nowadays, large-scale synthesis of dsRNA in E. coli has developed, but further increasing the titer, rate, and yield (TRY) of dsRNA production is essential for future applications.

The model eukaryotic species Saccharomyces cerevisiae has also been used as a chassis for dsRNA production. S. cerevisiae has a clear genetic background, easy genetic engineering methods, and well-developed fermentation processes (Nandy and Srivastava, 2018). Besides, S. cerevisiae does not contain the core genes Dicer-2 and Argonaute-2 of the RNAi pathway (Drinnenberg et al., 2009), which allows efficient dsRNA synthesis in S. cerevisiae compared with E. coli and other bacterial species (Zhong et al., 2019). Similarly, plant chloroplast does not contain RNAi pathway, and dsRNA can be enriched to 0.4% of total RNA in plant chloroplast; expression of dsRNA in plant chloroplast can be used to protect plants from being fed by insects, which would be more efficient than expressing dsRNA form the plant leaves (Zhang et al., 2015).

Feeding the fruit fly Drosophila suzukii with recombinant yeast expressing insect dsRNA targeting y-Tubulin resulted in a significant reduction in larval survivorship, adult motility, and reproduction (Murphy et al., 2016). Moreover, feeding D. suzukii with genetically modified S. cerevisiae expressing dsRNA (targeting y-tubulin23C) resulted in a significant decrease in the fitness of D. suzukii in the environment (Abrieux and Chiu, 2016).

The expression of dsRNA in S. cerevisiae has also been validated in the mosquito Aedes aegypti. Fez2 and lrc were selected as target genes in A. aegypti, and shRNAs of these genes were expressed in S. cerevisiae. When the genetically-modified S. cerevisiae was heated, dried, and fed to insects, this led to >95%mortality of A. aegypti (Hapairai et al., 2017). The same effects were observed with Aedes albopictus, Anopheles gambiae, and Culex quinquefasciatus (Mysore et al., 2017; Mysore et al., 2019a; Mysore et al., 2019b). In this way, biocontrol strategies for specific mosquito species can be developed, to effectively suppress human diseases transmitted by mosquitoes.

Many insects, livestock, aquaculture species and humans consume yeast. Therefore, developing efficient yeast expression systems might increase the possibility of applying yeast-derived dsRNA commercially (Duman-Scheel, 2019). The dsRNA produced by S. cerevisiae can also serve as a potential oral delivery system for shRNA to mammalian cells (mouse intestinal DCs) and be used in human disease therapeutics (Zhang et al., 2014; Duman-Scheel, 2019). Several companies have developed yeast dsRNA expression systems. In May 2019, Renaissance BioScience filed a patent application for the production and delivery of bioactive dsRNA ingredients using yeasts.

In the future, large-scale production of dsRNA in S. cerevisiae can be enhanced by improving the expression vectors (Crook et al., 2014), the promoters for the dsRNA transcription (Voineagu et al., 2008), the length of the hairpins (Yoshimatsu and Nagawa, 1989), and the sites of integration positions (Kim et al., 2015).

Some B. subtilis strains are classified as probiotics for human and animal consumption (Rosales-Mendoza and Angulo, 2015). Therefore, this species has also been selected for dsRNA expression. The dsRNA (daf-2, unc-62) expression vector pBSR was introduced into B. subtilis, and feeding C. elegans this genetically-modified B. subtilis strain induced RNAi effects (Lezzerini et al., 2015). A B. subtilis strain with dsVP28 expression was able to effectively prevent shrimp infection with white spot syndrome virus (WSSV); the survival rate of shrimp treated with the B. subtilis strain was 91.67%, while that in the control group was only 28.57% (Saelim et al., 2020).

Bacillus thuringiensis (Bt) is an effective biopesticide production strain that has been widely used for control of lepidopteran pests. Bt has been used as an expression host for dsRNA production. In the vector pBTdsSBV-VP1, two spore-producing-dependent cyt1Aa promoters in opposite direction were linked to the VP1 gene of Sacbrood virus (SBV), and a Shine-Dalgarno sequence (GAAAGGAGG) was added at specific positions, which increased the stability of the RNA. Transfer of pBTdsSBV-VP1 into Bt strain 4Q7 led to the expression of dsRNA. Feeding the total RNA extracted from this Bt strain to Apis cerana (honeybees) infected with SBV virus significantly reduced the viral infection of the insects (Park et al., 2020).

The Bt-based dsRNA production platform has some advantages compared with other platforms. The cry sporulation-dependent gene promotor was used for dsRNA expression, and the dsRNA could be produced during the sporulation phase of Bt. Moreover, other expression systems (like E. coli, B. subtiis, S. cerevisiae expression systems) require an inducer (IPTG or others) to induce dsRNA expression, but no inducer is needed for expression in Bt. Finally, Bt cells can undergo enzyme-associated autolysis after sporulation, thus cell lysis is not required for dsRNA extraction (Park et al., 2020).

With the increase of insect resistance to Bt, the use of Bt as a platform for dsRNA expression would help with pest control via a Bt + RNAi strategy (Caccia et al., 2020; Kang et al., 2021). Therefore, the Bt dsRNA expression system could be a useful dsRNA production platform for the introduction of RNAi in organisms.

There are abundant symbiotic bacteria in the oral tract and gut of insects, and they interact directly with the insects and plants. Some symbiotic bacteria can easily be genetically manipulated, so they might be potentially efficient dsRNA production platforms for insect control. The use of insect-symbiotic bacteria to express dsRNA for insect control is known as symbiont-mediated RNAi (SMR) (Taracena et al., 2015; Whitten et al., 2016; Whitten and Dyson, 2017; Whitten, 2019; Asgari et al., 2020).

Rhodococcus rhodnii (R. rhodnii), a symbiotic bacterium of the triatomine Rhodnius prolixus, was used to express RHBP-specific hairpin RNA; the gene expression products of RHBP can suppress R. prolixus by affecting its adult oviposition (Taracena et al., 2015). Subsequently, two symbiotic bacterial strains, R. rhodnii and BFo2 (a member of the Enterobacteriales), were isolated from the insects R. prolixus and Frankliniella occidentalis (western flower thrips), respectively. The RNase III gene was knocked out and dsRNA expression cassettes was expressed in these two insect symbiotic bacteria; when the engineered bacteria were taken up by insects, the dsRNA functioned in the hosts, inducing RNAi effects (Whitten et al., 2016).

Snodgrassella alvi, a core gut symbiotic bacterium of the honeybee A. mellifera, was modified as a dsRNA-producing host. The dsRNA produced by the engineered S. alvi can suppress gene expression in A. mellifera. Moreover, this dsRNA can suppress genes of parasitic Varroa mites and kill them, which protects the honey-bees from the Varroa mites, the most threatening pest to the world’s beekeeping industry (Leonard et al., 2020). Based on this technology, a new bioproduct, “BioDirect” was registered as dsRNA for the prevention and control of Varroa mites. This is the first dsRNA biopesticide active ingredient submitted to the U.S. Environmental Protection Agency (EPA) for exogenous application in agriculture.

Thus, SMR is not only potential pest control agents, but can also be beneficial for insect protection. SMR depends on both the specificity of RNAi toward the targeted insect gene, and the specificity of the symbiotic bacterium for its host. This dual specificity makes SMR a precision control tool, and this tool is obviously different from chemical insecticides. However, there are issues that need to be addressed before symbiotic bacteria can reliably serve as dsRNA expression hosts. The first is to find suitable symbiotic bacteria that stably colonize the host insects; the bacterial content should also be relatively high in the host insect. The second is that the symbiotic bacteria must be able to express dsRNA efficiently and stably. Thus, the acquisition and modification of symbiotic bacteria and colonization of the engineered symbiotic bacteria in the host need to be addressed before applying SMR for dsRNA production.

Nevertheless, this SMR strategy is specific for pest control without increasing environmental stress, and it might be widely used in the future.