Transcriptome profiling of female accessory glands plus uterus
To identify genes encoding secreted protein components in the accessory gland, the transcriptome of the dissected accessory glands plus uterus from 5-day old head louse females was analyzed (Fig. 1). Among 8,753 genes annotated, 768 genes were determined to have signal sequences. DEG profiling between accessory gland plus uterus versus ovary (as a control organ) and 0-day old female (at the emergence day) versus 5-day old gravid female revealed that 109 genes were more specifically expressed (more than 10-fold compared to those in ovaries) and that only transcripts of 30 genes occupied 90% of whole expressed transcripts in the accessory gland plus uterus of 5-day old females (Additional file 2: Table S2). The top four genes (LNSP2, LNSP1-like, LNSP1 and LNSP2-like in order of expression) exhibited extremely high expression levels, and are responsible for 75.4% of the whole transcripts expressed in female accessory glands. In addition, the following two hypothetical genes, named as Agp 22 and Agp9, occupied 10.3% of the whole transcripts (Fig. 1b). Therefore, these six genes were presumed to encode the major structural components of nit sheath. It is still unclear, however, whether LNSP1-like and LNSP2-like have interchangeable roles with their intact parental genes or have distinct functions. Nevertheless, the similar combined transcription levels between LNSP1 plus LNSP1-like (387,915 TPM) and LNSP2 plus LNSP2-like (396,387 TPM) and the similarity of gene structures in LNSP1/2 and LNSP1/2-like suggest that both LNSP1-like and LNSP2-like proteins function interchangeably with their respective parental protein. In addition to the putative structural component genes, other genes encoding secretory proteins with putative catalytic or physiological functions included TG, PSI, defensin 1 and defensin 2 (Additional file 2: Table S2). No genes with allergenic properties were identified from the transcriptomes of accessory gland plus uterus.
Temporal transcription profiling, as determined by qPCR, confirmed that the transcription levels of all major accessory gland genes increased over time in an age-dependent manner (Fig. 2). In addition, qPCR results using three different tissues [accessory gland plus uterus, ovary and alimentary tract] as templates revealed that all the genes were exclusively expressed in the accessory gland plus oviduct except for TG and PSI (Additional file 3: Fig. S1). TG showed the highest expression pattern in the alimentary tract, followed by the accessory gland plus uterus (Additional file 3: Fig. S1), suggesting its diverse roles in these tissues.
Spatial transcriptional profiling of defensin 1, defensin 2, Agp22, Agp9 and PSI genes in separated accessory glands and uterus revealed that both defensin 1 and defensin 2 were 6.3- and 1.8-fold more transcribed in the uterus, respectively, suggesting its protective role against pathogens inside the uterus (Additional file 4: Fig. S2). Transcription levels of Agp22 and Agp9 were not significantly different between the two tissues whereas PSI showed a significantly higher transcription level in the accessory glands.
Functional characterization of several nit sheath proteins via RNAi
Knockdown of LNSP1 and LNSP2. In order to investigate the physiological functions of LNSP1 and LNSP2, the corresponding genes were knocked down singly or together. LNSP1 was 57% (P = 0.0053) knocked down by injection of LNSP1 dsRNA without affecting the transcription of LNSP2 (Additional file 5: Fig. S3). LNSP2 was also specifically knocked down (46%, P = 0.0213) by injection of LNSP2 dsRNA without non-specific suppression of LNSP1 transcription (Additional file 5: Fig. S3). Injection of both LNSP1 and LNSP2 dsRNAs resulted in 33% (P = 0.005) and 29% (P = 0.002) of knockdown, respectively (Additional file 5: Fig. S3).
When LNSP1 was knocked down, most eggs became shriveled and desiccated (Fig. 3b), which began to appear from 24 h post-injection (Additional file 6: Fig. S4), compared to the control eggs (Fig. 3a). Scanning electron microscopic observations revealed that the surface microstructures of eggs with LNSP1 knocked down were altered by increased roughness compared to that of control eggs (Additional file 7: Fig. S5). The overall egg number produced from a single female was not affected significantly (P = 0.7296) but hatchability was dramatically reduced (93.6%, P < 0.0001) as the eggs were severely dehydrated (Fig. 3e-f). Thus, LNSP1 appears to function in desiccation resistance, thereby enhancing the survival of embryo. Nevertheless, no apparent impairment was observed in the gluing function of nit sheath.
When LNSP2 transcription was suppressed by RNAi, amorphous glue-like materials were observed to be secreted over hair (Fig. 3c), and overall egg number per female was significantly reduced (80.5%, P = 0.0011) (Fig. 3e). None of the laid eggs hatched out following LNSP2 knockdown (Fig. 3f). Since LNSP2 knockdown females hardly laid any eggs, their body weight significantly increased compared to control females (P = 0.0002) (Fig. 3g). Unlike the control females (Fig. 3h), dissections of gravid females with LNSP2 knocked down clearly showed ruptured uteri, disintegrated ovaries and enlarged accessory glands (Fig. 3i), suggesting that oviposition is blocked when LNSP2 is not sufficient. Taken together, the nit sheath did not form properly without LNSP2, which appeared to be the main reason for the oviposition failure.
Double knockdown of both LNSP1 and LNSP2 resulted in the partial restoration of egg production along with the glue-like materials secreted over hair, but the shape of produced eggs was similar to the shriveled eggs with LNSP1 knocked down (Fig. 3d-e), indicating their susceptibility to desiccation. The egg hatchability was zero as in the case of LNSP2 knockdown (Fig. 3f). Overall, the eggs with double knockdown of LNSP1 and LNSP2 exhibited mixed characteristics of each of single knockdown.
Knockdown of other genes. Knockdown of either Agp22, Agp9 or PSI did not induce any apparent changes in egg number, shape or viability (Additional file 8: Fig. S6), indicating that they may not be as crucial as LNSP1 or LNSP2 in preserving egg viability.
Elucidation of TG-mediated crosslinking of nit sheath
Since the genome-wide survey of possible protein crosslinking systems indicated that TG-mediated crosslinking is the only feasible option, RNAi-based functional characterization of TG was performed. Injection of TG dsRNA significantly suppressed TG transcription (45%, P = 0.0008; Fig. 4a). Knockdown of TG via RNAi resulted in reduced egg number although the difference was statistically insignificant (P = 0.41) (Fig. 4b). Laid eggs appeared normal in shape but the angle between egg and hair was significantly increased (27.6±7.3° compared to 15.7±4.5° in control, P < 0.0001; Fig. 4d) and glue particles were observed at the bottom side of eggs (Fig. 4e), suggesting slowed solidification of nit sheath gel following TG knockdown. The hatchability was dramatically reduced (64.8%, P < 0.0001; Fig. 4c), which was mostly related with the failure of hatching out from eggshells (Fig. 4g) unlike the control eggs (Fig. 4f). Nevertheless, no apparent abnormality in biology (lifespan, viability, etc.) was observed in the females with TG knockdown.
To confirm the role of TG in crosslinking of nit sheath, TG inhibitors were applied via injection or hair coating. Iodoacetamide, a well-known, potent, irreversible TG inhibitor, was too toxic to females. Since the toxic effects of iodoacetamide were exerted within a very narrow window of doses, regardless of treatment routes (hair coating vs. injection), no meaningful dose-response was deduced (Additional file 9: Table S2). Therefore, GGsTop was used as a surrogate inhibitor because it is known as an irreversible inhibitor of gamma-glutamyltransferase (EC 2.3.2.2), which is closely related with TG (EC 2.3.2.13), and because no genes encoding the gamma-glutamyltransferase was annotated from the human louse genome. Injection of GGsTop into gravid females resulted in similar responses as observed in the TG knockdown (Fig. 5). Injection of GGsTop caused a decreased total number of eggs laid (1.4-fold, P = 0.00097; Fig. 5c), a decrease in their hatchability (1.4-fold, P = 0.00014; Fig. 5d) and an increased number of desiccated eggs found on the hair (7.9-fold, P = 0.0049; Fig. 5e). In addition, 12.3% of total eggs laid by GGsTop-injected females were detached from hairs (Fig. 5f). When females laid eggs on the hair tufts coated with GGsTop, thread- or plate-shaped gel materials, produced by oviposition behavior, were observed around the eggs (Fig. 5g), suggesting that solidification of nit sheath was slowed at the post-oviposition stage. However, no reduction in egg hatchability or structural abnormality of eggs were observed, indicating that the egg coating and crosslinking of nit sheath gel inside the uterus during the pre-oviposition stage is sufficient to ensure egg viability. Taken together, these findings suggest that GGsTop inhibited TG activity, and thus suppressed the crosslinking mediated by TG, impairing the egg viability.