Phenotypic analyses, protein localization, and bacteriostatic activity of Drosophila melanogaster transferrin-1

Abstract

Transferrin-1 (Tsf1) is an extracellular insect protein with a high affinity for iron. The functions of Tsf1 are still poorly understood; however, Drosophila melanogaster Tsf1 has been shown to influence iron distribution in the fly body and to protect flies against some infections. The goal of this study was to better understand the physiological functions of Tsf1 in D. melanogaster by 1) investigating Tsf1 null phenotypes, 2) determining tissue-specific localization of Tsf1, 3) measuring the concentration of Tsf1 in hemolymph, 4) testing Tsf1 for bacteriostatic activity, and 5) evaluating the effect of metal and paraquat treatments on Tsf1 abundance. Flies lacking Tsf1 had more iron than wild-type flies in specialized midgut cells that take up iron from the diet; however, the absence of Tsf1 had no effect on the iron content of whole midguts, fat body, hemolymph, or heads. Thus, as previous studies have suggested, Tsf1 appears to have a minor role in iron transport. Tsf1 was abundant in hemolymph from larvae (0.4 μM), pupae (1.4 μM), adult females (4.4 μM) and adult males (22 μM). Apo-Tsf1 at 1 μM had bacteriostatic activity whereas holo-Tsf1 did not, suggesting that Tsf1 can inhibit microbial growth by sequestering iron in hemolymph and other extracellular environments. This hypothesis was supported by detection of secreted Tsf1 in tracheae, testes and seminal vesicles. Colocalization of Tsf1 with an endosome marker in oocytes suggested that Tsf1 may provide iron to developing eggs; however, eggs from mothers lacking Tsf1 had the same amount of iron as control eggs, and they hatched at a wild-type rate. Thus, the primary function of Tsf1 uptake by oocytes may be to defend against infection rather than to provide eggs with iron. In beetles, Tsf1 plays a role in protection against oxidative stress. In contrast, we found that flies lacking Tsf1 had a typical life span and greater resistance to paraquat-induced oxidative stress. In addition, Tsf1 abundance remained unchanged in response to ingestion of iron, cadmium or paraquat or to injection of iron. These results suggest that Tsf1 has a limited role in protection against oxidative stress in D. melanogaster.

Graphical Abstract

1. INTRODUCTION

Iron is an essential micronutrient for all types of organisms, but it can initiate the production of the highly reactive hydroxyl radical (Kosman, 2018). Unsurprisingly, homeostatic processes have evolved in animal lineages to supply an adequate amount of iron to cells while limiting iron toxicity (Anderson and Leibold, 2014; Galay et al., 2015; Lambert, 2012; Tang and Zhou, 2013a). Animals also have immune mechanisms that inhibit the growth of microbes by limiting the availability of iron (Ong et al., 2006). Some of these homeostatic and immune mechanisms involve iron-binding members of the transferrin family of proteins (Lambert, 2012).

The two best-studied transferrins are mammalian serum transferrin and mammalian lactoferrin (Lambert, 2012). Serum transferrin transports iron from one cell to another and also protects against iron-inducted oxidative stress (Galaris et al., 2019; Gkouvatsos et al., 2012). Lactoferrin is present in various extracellular fluids and inhibits pathogen growth by sequestering iron (Farnaud and Evans, 2003). Both proteins are composed of two homologous lobes, each with a high affinity iron-binding site (Aisen et al., 1978; Aisen and Leibman, 1972; Anderson et al., 1987; Bailey et al., 1988; Mizutani et al., 2012). Serum transferrin releases iron at pH 4-6, which enables it to release iron in endosomes and lysosomes; in contrast, iron binding by lactoferrin is less sensitive to low pH (Day et al., 1992).

In insects, a single transferrin family member, transferrin-1 (Tsf1), is implicated in iron transport, immune-related iron sequestration, and protection against oxidative stress (Brummett et al., 2017; Geiser and Winzerling, 2012; Iatsenko et al., 2020; Xiao et al., 2019; Xue et al., 2020). Tsf1 is equally similar to serum transferrin and lactoferrin. For example, the amino acid sequence of Drosophila melanogaster Tsf1 (accession number AAC67389) is 23% identical to human serum transferrin (P02787) and to human lactoferrin (NP_002334). Like both of these mammalian transferrins, Tsf1 is an extracellular, bilobal protein with high affinity for iron; however, many Tsf1 orthologs have only one iron-binding site, and iron coordination by Tsf1 differs from that of serum transferrin and lactoferrin (Weber et al., 2020b, 2020a). Like serum transferrin, Tsf1 has high affinity for iron at neutral pH and low affinity at acidic pH (Weber et al., 2020a).

The role of serum transferrin in iron transport in mammals is well-studied. Iron is transported out of mammalian cells as ferrous ions (Fe²⁺), oxidized to ferric ions (Fe³⁺), and loaded onto serum transferrin (De Domenico et al., 2007; Han, 2011). Iron uptake can occur via two pathways that involve serum transferrin and its receptor. The most widely known mechanism is receptor-mediated endocytic uptake of the Fe³⁺-transferrin-receptor complex (Frazer and Anderson, 2014). A non-endocytic mechanism involves extracellular ferric reduction of iron bound to the transferrin-receptor complex followed by ferrous ion uptake (Kosman, 2020). Animals with a severe deficiency of serum transferrin die shortly after birth with symptoms of anemia and iron overload of various tissues due to a disruption in regulated iron transport (Anderson and Vulpe, 2009; Bernstein, 1987; Hamill et al., 1991).

The role of Tsf1 in iron transport in insects is still poorly understood (Tang and Zhou, 2013b; Xiao et al., 2019). Studies of Tsf1 from several species of insects indicate that Tsf1 is involved in iron transport. In Manduca sexta, cellular uptake of Tsf1-bound iron occurs via an unknown mechanism (Huebers et al., 1988). In Sarcophaga peregrina, iron-bound Tsf1 is transported into oocytes (Kurama et al., 1995). In D. melanogaster, RNAi-mediated knockdown of Tsf1 resulted in changes in iron distribution in the insect body, although a complete absence of Tsf1 had no effect on iron distribution in healthy insects (Iatsenko et al., 2020; Xiao et al., 2019). In addition, in D. melanogaster, a lack of Tsf1 interferes with immune-induced transfer of iron from hemolymph to fat body (Iatsenko et al., 2020). A major question about how Tsf1 functions in iron transport is what is the source of the iron that becomes bound to Tsf1. Insects lack the protein that exports ferrous ions from mammalian cells, and the only known mechanism of iron export from insect cells is the secretion of ferritin loaded with ferric iron, which would need to be reduced to ferrous ions, released from ferritin, and then oxidized prior to being bound by Tsf1 (Nichol and Locke, 1990; Pham and Winzerling, 2010; Tang and Zhou, 2013b; Whiten et al., 2018; Xiao et al., 2014). Another key question is whether iron-bound Tsf1 is endocytosed by cells. Insects lack a mammalian transferrin receptor homolog, and endocytic uptake of Tsf1 has not been observed (Geiser and Winzerling, 2012; Lambert, 2012).

Lactoferrin and Tsf1 appear to have analogous functions. Apo (iron-depleted)-lactoferrin functions as an immune protein in extracellular fluids such as milk, tears, and saliva by chelating iron and, thus, decreasing the availability of iron for microbial growth (Farnaud and Evans, 2003; Jenssen and Hancock, 2009; Orsi, 2004). Tsf1 has been identified in four extracellular fluids: hemolymph, molting fluid, saliva, and seminal fluid (Bonilla et al., 2015; Brummett et al., 2017; Geiser and Winzerling, 2012; Hattori et al., 2015; Qu et al., 2014; Simmons et al., 2013; Zhang et al., 2014). Like lactoferrin, Tsf1 from M. sexta has bacteriostatic activity that depends on its ability to bind iron; therefore, it is likely that Tsf1 participates in immune-related iron sequestration in these extracellular locations (Brummett et al., 2017). Infection induces upregulation of Tsf1, and a reduction or absence of Tsf1 can lead to increased susceptibility to microbes, although whether this is due to iron sequestration, iron transport, or both is unknown (Geiser and Winzerling, 2012; Iatsenko et al., 2020; Kim and Kim, 2010; Lehane et al., 2008).

Serum transferrin, by participating in regulated iron uptake, protects cells from iron overload and, thus, protects against oxidative stress (Galaris et al., 2019). Tsf1 influences the degree of oxidative stress in insects, but the effect can be positive or negative (Kim et al., 2008; Lee et al., 2006; Xue et al., 2020; Zhang et al., 2018). In the beetles Protaetia brevitarsis and Apriona germari, oxidative stress-eliciting treatments such as injection of iron, hydrogen peroxide, or paraquat resulted in upregulation of Tsf1, and Tsf1 RNAi resulted in oxidative stress (Kim et al., 2008; Lee et al., 2006). In contrast to the protective effect of Tsf1 in beetles, D. melanogaster Tsf1 appears to contribute to oxidative stress in flies that are exposed to the insecticide rotenone; in rotenone-treated flies, RNAi-mediated knock down of Tsf1 in neural tissues resulted in less iron in the head and less oxidative stress in the central nervous system (Xue et al., 2020).

The goal of this study was to better understand the physiological functions of Tsf1 in D. melanogaster by using a combination of experimental strategies. Because one of the questions we were interested in was the possible role of Tsf1-mediated transport of iron into developing oocytes, we focused most of our experiments on mated females. We generated a Tsf1 deletion mutant (Tsf1^m10) to study the effect of a lack of Tsf1 on viability, longevity, iron distribution, and susceptibility to oxidative stress. We used immunohistochemistry to establish tissue-specific localization of Tsf1 and to determine whether endocytic uptake of Tsf1 occurs. We measured the concentration of Tsf1 in the hemolymph of larvae, pupae and adults so that we would know what concentrations of Tsf1 are physiologically relevant, and then tested apo- and holo-forms of Tsf1 for bacteriostatic activity. Finally, we used immunoblot analyses to determine whether Tsf1 abundance changed in response to ingestion of iron, cadmium, copper or paraquat or to injection of iron. Our results suggest that, in D. melanogaster, Tsf1 plays a significant role in immune-related iron sequestration, a minor role in iron transport, and no discernible role in protection against oxidative stress.

2. MATERIALS AND METHODS

2.1. Drosophila melanogaster stocks

Insects were cultured at 25.5 °C on K12 High Efficiency diet (United States Biological). The diet components are agar, carob pulp flour, corn flour, p-hydroxybenzoic acid methyl ester, sucrose and yeast. w¹¹¹⁸ (#3605) and y¹ w¹ (#1495) stocks were obtained from the Bloomington Drosophila Stock Center (BDSC) to use as wild-type (WT) lines. Fly lines used for RNAi analyses included actin-Gal4 (act-Gal4) (BDSC #25374), UAS-dsTsf1³⁰⁰⁴⁵ (BDSC #62968), UAS-dsTsf1^R3 (NIG-FLY #6186R-3), UAS-dsTsf1^KK (Vienna Drosophila Resource Center (VDRC) #106479), and UAS-dsTsf1^GD (VDRC #14666) (Dietzl et al., 2007; Zirin et al., 2020). Negative control lines for RNAi experiments included attP30B (VDRC #60100) and attP40 (BDSC #36304). For genotypic identification of larval progeny from heterozygous parents, a GFP-marked CyO balancer chromosome was used (BDSC #4523).

Tsf1^m10, which is a deletion of the entire Tsf1 coding region, was made via CRISPR-Cas9 technology by GenetiVision Corporation. Briefly, guide RNAs (aaccggttgagtcaccacggtgg and gagtggtgtgaaaagccaattgg) and a donor construct containing 3xP3-GFP as a selectable marker were co-injected into y w; nos-Cas9 (y+)/CyO eggs, F1 progeny were screened for GFP-expressing individuals, and three independent stocks were established. PCR using genomic DNA as the template was performed by GenetiVision to confirm that the GFP cassette was inserted at the expected location. To lessen differences in the genetic background of Tsf1^m10 and Tsf1⁺ lines, the Tsf1^m10 line was outcrossed to a w¹¹¹⁸ control line. (Note that w¹¹¹⁸ was chosen as the control genotype because the red eye pigment in wild-type flies interferes with ferrozine-based iron assays.) Tsf1^m10 virgin females were crossed to w¹¹¹⁸ males, and this process was repeated for eight generations. A homozygous w¹¹¹⁸ Tsf1^m10 stock was made without the use of a balancer chromosome. In addition, a homozygous w¹¹¹⁸ stock was created from the eighth generation outcross and was used as a control line instead of the original w¹¹¹⁸ stock (to attempt to control for microbes in the cultures). For simplicity, the w¹¹¹⁸ Tsf1^m10 line is referred to as Tsf1^m10, and the w¹¹¹⁸ line is referred to as WT. Because Tsf1 is an X-linked gene, female flies from the Tsf1^m10 line are homozygous (w¹¹¹⁸ Tsf1^m10/w¹¹¹⁸ Tsf1^m10), and males are hemizygous (w¹¹¹⁸ Tsf1^m10/ Y).

2.2. Verification of the outcrossed Tsf1^m10 line

The Tsf1^m10 mutation in the outcrossed line was verified by RT-PCR and immunoblot analyses. RNA was isolated from pools of 25 female flies (Tsf1^m10 and WT) with the use of the RNAqueous-4PCR kit (Ambion), and first strand cDNA synthesis was done with superscript IV VILO Master Mix (Thermo Fisher). PCR was performed with Tsf1 primers (CCTGCTGAAGAAGAAGTA and CATCATAGCCACTGAACt) for 32 cycles and with gapdh2 primers (CGTTCATGCCACCACCGCTA and CCACGTCCATCACGCCACAA) for 30 cycles. Samples were prepared for immunoblot analysis by homogenizing pools of 10 adult females (Tsf1^m10 and WT) in 600 μl 20 mM Tris, 137 mM NaCl, 1% Triton X-100, 1% glycerol, pH 6.3, then centrifuging at 16,000 × g for 1 minute, and mixing the soluble fraction with an equal volume of 2x SDS sample buffer. Samples (12 μl) were subjected to SDS-PAGE followed by protein transfer to nitrocellulose. Immunodetection of Tsf1 was performed with the use of a 1:2,000 dilution of polyclonal antiserum against D. melanogaster Tsf1 (described in section 2.4), a 1:3,000 dilution of goat anti-guinea pig antibody conjugated to alkaline phosphatase, and a colorimetric detection method. Immunodetection of β-tubulin was performed with the use of a 1:300 dilution of a monoclonal antibody against β-tubulin (described in section 2.4), a 1:3,000 dilution of goat anti-mouse antibody conjugated to alkaline phosphatase, and a colorimetric detection method.

2.3. Purification of Tsf1

Recombinant Tsf1 was expressed using a baculovirus expression system and purified as described previously (Weber et al., 2020a). Briefly, Sf9 cells were infected with a recombinant baculovirus made with a full-length Tsf1 cDNA (LP08340, from the Drosophila Genomics Resource Center), and secreted Tsf1 was purified from the medium with the use of ammonium sulfate precipitation, and anion exchange and size exclusion chromatography.

2.4. Antisera

Purified Tsf1 (0.3 mg) was sent to Cocalico Biologicals for the production of polyclonal antiserum in a guinea pig. We used commercially available rabbit antiserum against D. melanogaster Rab5 (Abcam #ab31261). This antiserum has been used by others to identify endosomes in neural tissue, wing discs, eye discs, and other tissues (Fan et al., 2013; Yuva-Aydemir et al., 2011; Zschätzsch et al., 2014). A mouse monoclonal antibody that recognizes D. melanogaster β-tubulin (Hybridoma Product E7) was purchased from the Developmental Studies Hybridoma Bank at the University of Iowa. E7 was developed by M. McCutcheon and S. Carroll and was deposited by M. Klymkowsky. A mouse monoclonal antibody that recognizes glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was purchased from Thermo Fisher (#MA5-15738).

2.5. Hemolymph collection

Hemolymph was collected from WT larvae, pupae and adults.

Third instar larvae that were still in the feeding stage were collected by adding 20% sucrose solution to a culture, collecting floating larvae, and then selecting third instar larvae based on the morphology of the spiracles. The larvae were rinsed with PBS and blotted dry. The collection apparatus for larval hemolymph was a 0.6 ml microfuge tube with a slit cut in the bottom with a razor blade, placed within a 1.5 ml collection tube. Hemolymph from groups of 12 larvae was collected by piercing larvae near the mouth hooks with forceps, placing the larvae in a cold collection apparatus, and centrifuging at 6,000 × g for 5 seconds at 4 °C. The volume of hemolymph (2-3 μl) was measured and immediately added to 97 μl SDS sample buffer and heated at 95 °C for 5 minutes.

Collection of hemolymph from pupae was challenging, but was accomplished with a slightly modified protocol originally developed to collect hemolymph from Tribolium castaneum (Tabunoki et al., 2019). Pupae in the P4-P7 stage (Bainbridge and Bownes, 1981) were collected with a wet paint brush into a small sieve. Pupae were washed with 0.1% Tween 20, rinsed with tap water for 1 minute, rinsed with 70% ethanol, and blotted dry with a paper towel. The collection apparatus for pupal hemolymph was a 0.6 ml microfuge tube with a hole poked in the bottom with a 23 gauge needle, placed within a 1.5 ml collection tube. Pupae were transferred to a damp filter paper on a glass dish that was sitting on ice. Hemolymph from 100 pupae was collected by piercing the dorsal-lateral thorax with a minuten pin, transferring the pupae to a cold collection apparatus, and centrifuging at 12,000 × g for 10 minutes at 4 °C. The sample in the collection tube had a pellet, a clear layer, and a floating cloudy layer. The clear layer was collected, and the volume of this hemolymph sample (5 μl) was measured and immediately added to an equal volume of 2x SDS sample buffer and heated at 95 °C for 5 minutes.

A previously published method (Tuthill et al., 2020) with slight modifications was used to collect hemolymph from groups of mated females and males that were 6-7 days old. The collection apparatus for adult hemolymph was a 0.6 ml microfuge tube with a hole poked in the bottom with a 23 gauge needle, placed within a 1.5 ml collection tube. Anesthetized flies (~50-60 females or ~70 males) were pierced in the lateral thorax with a minuten pin, placed in a cold collection apparatus, and centrifuged at 7,600 × g for 5 minutes at 4 °C. For immunoblot analyses, 3 μl hemolymph was immediately added to 97 μl SDS sample buffer and heated at 95 °C for 5 minutes.

2.6. Estimating Tsf1 concentration in hemolymph

Immunoblot analyses were used to estimate the concentration of Tsf1 in hemolymph. Hemolymph samples and purified Tsf1 were subjected to reducing SDS-PAGE followed by protein transfer to nitrocellulose. One hemolymph sample from pupae and three independent hemolymph samples from larvae and adults were evaluated. The volumes used were 0.2 μl for larvae, 0.1 μl for pupae, 0.1 μl for adult females, and 0.01 μl for adult males; each gel also included 1, 5, 10, 25 and 50 ng purified Tsf1. Immunodetection was performed as described in section 2.2. Image analysis was performed with Carestream Molecular Imaging Software. A standard curve was used to calculate the concentration of Tsf1 in each hemolymph sample. For larval and adult hemolymph samples, mean ± standard deviation was calculated.

2.7. Immunohistochemistry

To detect Tsf1 in tissues from third instar larvae and adult females, we performed immunostaining of Tsf1 and an early endosome marker, Rab5, in whole mounted tissues from WT individuals. A homozygous y w Tsf1^m10 line was used as a negative control. Insects were dissected in phosphate buffered saline (PBS), and samples were briefly rinsed in fresh PBS before processing. Larval tissues were prepared by snipping the anterior end and inverting the carcass by pushing in on the posterior end. Ovaries were dissected from adult females by making an incision in the cuticle on the ventral side of the abdomen and removing the ovaries. Pericardial cells were prepared in adult females by pinning the down the thorax, making transverse incisions at the anterior and posterior ends of the ventral abdominal cuticle, followed by a longitudinal incision from anterior to posterior ends of the ventral abdominal cuticle, and then pinning down each side of the opening to reveal the abdominal contents. Some fat body and gut tissue was removed to better expose the dorsal vessel and pericardial cells. To analyze adult female midguts, Malpighian tubules, and neural tissues, abdomens were prepared similarly to those used for analyzing pericardial cells, without the removal of these tissues. To detect Tsf1 in tissues from adult male abdomens, we removed abdomens from five day old WT and Tsf1^m10 males, and used forceps to tear the cuticle in two additional locations before fixation. Inoculation of flies with bacteria results in the transfer of iron from hemolymph to fat body (Iatsenko et al., 2020); therefore, to attempt to increase hypothetical endocytic uptake of Tsf1 in adult males, immune induction was accomplished by pricking the lateral thorax with a minuten pin dipped in a pellet of Micrococcus luteus as described previously (Iatsenko et al., 2020). Immune challenge results in endocytic activity within 0.5-3 hours (Sorvina et al., 2013); therefore, fly abdomens were collected at four intervals within a 0.5-3 hours post-inoculation time frame.

The antisera used for immunohistochemistry are described in section 2.4. All washes and incubations were performed with rocking at room temperature unless otherwise noted. Dissected samples were fixed in 4% (w/v) paraformaldehyde in PBS solution for 2 hours at 4 °C. The samples were washed once for ≥5 minutes with PBS and five times for 5 minutes each with PBS plus 0.5% (v/v) Triton X-100 (PBST). Samples were then blocked for 1 hour in PBST and 10% (v/v) goat serum, washed five times for 5 minutes each with PBST, and incubated with Tsf1 antiserum (1:20,000) and Rab5 antiserum (1:1,000) for three days at 4 °C. Four to five 5 minute washes with PBST were followed by a 7-8-hour wash. The samples were then incubated with Alexa Fluor 568-conjugated goat anti-guinea pig secondary antibody (1:1,000; Invitrogen) and Alexa Fluor 488-conjugated goat anti-rabbit secondary antibody (1:1,000; Invitrogen) overnight at 4 °C. Samples were then washed five times for 5 minutes each followed by a 4 hour wash with PBST. To improve the signal to noise ratio, the Tsf1 antiserum used for analysis of male abdominal tissues was pre-adsorbed overnight against paraformaldehyde-fixed abdomens from Tsf1^m10 males, and the final dilution was 1:10,000. Samples were transferred to glass slides and mounted in Prolong Gold Antifade mountant (Invitrogen).

Imaging of larval and adult female tissues was performed using a Zeiss LSM 700 AxioObserver confocal microscope with Plan-Neofluar 20x/0.50 M27 and 40x/1.30 Oil M27 objectives. Imaging of males tissues was performed using a Zeiss LSM 880 AxioObserver confocal microscope with EC Plan-Neofluar 10x/0.3 M27, Plan-Apochromat 20x/0.8 M27, and EC-Plan Neoiluar 40x/1.3 Oil DIC M27 objectives. Alexa Fluor 568 excitation was at 555 nm and emission was at 573 nm. Alexa Fluor 488 excitation was at 488 nm and emission was at 518 nm. The WT and Tsf1^m10 samples were imaged with identical settings. Zen 3.1 lite and Zen 3.4 blue software were used for imaging and processing.

2.8. Effect of apo-Tsf1 and holo-Tsf1 on growth of E. coli

Apo-Tsf1 (Tsf1 with no bound iron) and holo-Tsf1 (ferric-Tsf1) were made from purified recombinant Tsf1 as previously described (Weber et al., 2020a). Wild-type E. coli was obtained from the Coli Genetic Stock Center (strain # 7636). E. coli was cultured in MOPS medium (Neidhardt et al., 1974) until the OD₆₀₀ was 2.0. One ml of E. coli culture was diluted with 20 ml low iron (1 μM) MOPS medium and grown overnight at 37°C. The bacteria were collected by centrifugation, washed and resuspended to an OD₆₀₀ of 1.6 in MOPS medium without added iron. To test for antimicrobial activity of Tsf1, 5 μl of the prepared bacteria were added to 191 μl MOPS without iron, and 4 μl apo- or holo-Tsf1 (final concentration 1 μM) or 4 μl Tsf1 storage buffer (20 mM potassium phosphate, 20 mM sodium chloride buffer, pH 7). Cultures were grown in sterile 96 well plates sealed with a Breathe-Easy gas permeable sealing membrane (Diversified Biotech) in a PowerWave XS microplate reader (BioTek) with constant shaking at 37°C. The OD₆₀₀ of each culture was measured after 8 hours. Three replicate cultures were analyzed per treatment, and mean ± SD was calculated. Differences in means were assessed by performing an unpaired t-test.

2.9. Evaluation of Tsf1 RNAi genotypes

To generate constitutive, ubiquitous knock down of Tsf1, act-Gal4/CyO,GFP females were mated to the following Tsf1 RNAi lines: UAS-dsTsf1^KK, UAS-dsTsf1^GD, UAS-dsTsf1³⁰⁰⁴⁵/CyO,GFP, and UAS-dsTsf1^R3. Control lines included attP30B (for the UAS-dsTsf1^KK line), attP40 (for the UAS-dsTsf1³⁰⁰⁴⁵ line) and w¹¹¹⁸ (for the UAS-dsTsf1^GD and UAS-dsTsf1^R3 lines). (Stock numbers are listed in section 2.1). The degree of Tsf1 knockdown was determined by immunoblot analyses. Wandering larvae (four larvae per sample, n = 3) were homogenized in 50 μl homogenization buffer and 12 μl samples were subjected to SDS-PAGE, protein transfer and immunodetection as described in section 2.2. Image analysis, including normalization of Tsf1 band intensity to β-tubulin, was performed with Carestream Molecular Imaging Software. Mean normalized band intensity was calculated with GraphPad Prism software, and means for each RNAi and corresponding control genotype were used to calculate percent reduction in Tsf1 abundance. To measure the effect of Tsf1 knock down on pupal lethality, three groups of 40 larvae were collected at the wandering stage with a wet paint brush and placed in vials. Pupae were monitored each day until all insects had eclosed or died, and percent pupal mortality was calculated.

2.10. Lethal phase analysis

Lethal phase analyses were done to compare the survival of Tsf1^m10 and WT insects at the embryonic, larval and pupal stages. Differences in survival rates at each stage were assessed by performing an unpaired t-test using GraphPad Prism software.

Survival during the embryonic stage was evaluated by comparing the hatch rate of WT eggs from WT mothers and Tsf1^m10 eggs from Tsf1^m10 mothers. Groups of 50 eggs (n = 9) were collected with fine tip forceps from grape agar plates supplemented with yeast paste and placed on grape agar plates. The number of unhatched eggs was counted after 2 days.

Survival during the larval stages was evaluated by measuring the pupation rate of larvae. Groups of 20 first instar larvae (n = 8) were collected with fine tip forceps from grape agar plates supplemented with yeast paste and placed in 35 mm dishes with food. The larvae were monitored each day, and as the larvae began to pupate, the number of pupae was counted each day until no new pupae were observed.

Survival during the pupal stage was evaluated by measuring the eclosion rate. Groups of 20 larvae (n = 8) were collected at the wandering stage with a wet paint brush and placed in vials. Pupae were monitored each day until all insects had eclosed or died.

2.11. Adult survival analysis

Newly eclosed female and male flies (Tsf1^m10 and WT) were collected, and the sexes were kept together for one day to allow mating to occur. Groups of 25 females or males were placed in a vial with food. Flies were cultured at 25.5 °C and transferred to fresh food every 2 to 3 days. Dead flies were counted each day until all flies had died. Four replicates were done for each genotype, and data were pooled for survival curve analysis, which was done by performing a log-rank (Mantel-Cox) test with the use of GraphPad Prism software.

2.12. Iron content analysis

Whole adult flies, dissected tissues, and eggs were prepared for iron content analysis. In all cases, comparisons were made between Tsf1^m10 and WT samples.

The adults analyzed were mated females and mated males, either 5 or 14 days old. Each sample (n = 7) contained 25 flies.

Heads were collected from five day old mated females. Groups of 100 flies were placed in a 15 ml tube, frozen in liquid nitrogen, vortexed to break the heads off, and placed in a sieve (710 μm) that retained the fly bodies but allowed the heads to fall through to a head collection sieve (212 μm). (Wings and some legs remained in the tubes due to static electricity; most of the remaining legs passed through the head collection sieve.) Each sample (WT, n = 6; Tsf1^m10, n = 7) contained ~100 heads.

Midguts were dissected from five day old mated females in PBS and rinsed one time with PBS to remove hemolymph. Each sample (n = 5) contained ~80 midguts.

Fat body-enriched abdominal carcasses were dissected from five day old mated females in PBS. Abdominal carcasses were prepared by separating the abdomen from the rest of the fly and then removing the ovaries, midgut, hindgut, and Malpighian tubules. Most of the remaining tissue was fat body. For simplicity, fat body-enriched abdominal carcasses are referred to as fat body samples. Each sample (n = 4) contained 90 abdominal carcasses.

Hemolymph was collected from five day old mated females as described in section 2.5. Hemolymph was frozen at −80°C immediately after collection, and four pools of ~35 μl hemolymph (from ~600 flies each) were analyzed.

Newly laid eggs (0-2.5 hours old) were collected from Tsf1^m10 females mated to Tsf1^m10 males and from WT females mated to WT males. Females laid eggs on grape agar plates with a thin layer of yeast paste. Eggs were transferred to a small sieve, rinsed extensively with deionized water, and placed in 1.5 ml tubes. The tubes were briefly centrifuged to pellet the eggs, the remaining water was removed with a fine-tip pipet, and the volume of eggs in each tube was estimated. Sample volumes ranged from 30-60 μl (n = 8).

To estimate the iron content of whole flies, dissected tissues and eggs, we used a ferrozine-based assay that has been described previously (Lang et al., 2012), except that samples were homogenized in PBS plus 0.1% Triton X-100 (Xiao et al., 2019). Briefly, samples were homogenized, centrifuged to remove insoluble matter, and digested with concentrated hydrochloric acid at 95 °C; then iron in the samples was reduced with ascorbic acid, iron-ferrozine complex formation occurred, and absorbance was measured at 562 nm. Blanks were made by substituting homogenates with homogenization buffer. Protein concentrations were determined with a bicinchoninic acid assay (Micro BCA Protein Assay Kit from Thermo Fisher Scientific). Results were reported as nmole iron per mg protein. Differences in iron content were assessed by performing an unpaired t test using GraphPad Prism software.

2.13. Histological staining of iron in midgut cells

To detect iron in cells of the iron region of the midgut, a Prussian blue staining method was used (Mehta et al., 2009). Because little or no staining was observed in the midguts of adults fed a standard diet, the amount of iron in the iron cells was increased by providing food supplemented with iron. Newly eclosed flies were placed on food supplemented with 1 mM ferric ammonium citrate, and after two days were transferred to regular food for 24 hours to clear the high iron food from the gut lumen. Midguts from three day old mated adult females (WT and Tsf1^m10) were dissected and fixed in 4% formaldehyde, permeabilized with 1% Tween-20, incubated with 2% potassium ferrocyanide in 0.24 N HCl, rinsed with water, and observed with a stereomicroscope. Images were taken with an AxioCam camera using AxioVision software. Differences in the number of midguts with detectable blue color was evaluated with a Fisher’s exact test. Twenty-one midguts per genotype were analyzed.

2.14. Measuring the effect of iron, copper, cadmium and paraquat on whole body Tsf1 abundance.

Feeding experiments: Five day old mated females or males were fed 10% sucrose or 10% sucrose with 10 mM ferric ammonium citrate, 15 mM copper sulfate, or 25 mM cadmium chloride. Similarly, WT flies were fed 5% sucrose or 5% sucrose with 20 mM paraquat. Flies were transferred from vials with regular food to bottles with paper wipes saturated with one of the experimental solutions. Green food coloring was added to the solutions to allow verification of feeding. After 24 or 48 hours, flies were collected for immunoblot analysis of Tsf1 expression.

Injection experiments: Five day old mated females were injected in the lateral thorax with 6.9 nl sterile 5 mM ferrous sulfate or 5 mM ferric citrate with the use of a Nanoliter 2000 microinjection system with a Micro4 controller. Control flies were injected with sterile water. After 24 hours, flies were collected for immunoblot analysis.

Bacterial inoculation experiment: Five day old mated females were inoculated by pricking the lateral thorax with a minuten pin dipped in a pellet of a 1:1 ratio of M. luteus and E. coli. The amount of inoculum per fly was variable. After 24 hours, flies were collected for immunoblot analysis.

For each treatment, n = 8 individual whole flies. Flies were homogenized in 50 μl homogenization buffer and samples were subjected to SDS-PAGE, protein transfer and immunodetection as described in section 2.2. Image analysis, including normalization of Tsf1 band intensity to β-tubulin (or, for the bacterial inoculation experiment, GAPDH), was performed with Carestream Molecular Imaging Software. Differences in Tsf1 abundance were assessed by performing unpaired t-tests with GraphPad Prism software.

2.15. Effect of Tsf1^m10 mutation on susceptibility to paraquat.

Newly eclosed flies were collected, and the sexes were kept together for three days to allow mating to occur. Groups of 25 WT or Tsf1^m10 females were placed in a bottle with paper wipes saturated with either 5% sucrose or 5% sucrose with 10 mM or 20 mM paraquat. Green food coloring added to the solutions was visible in the abdomens of the flies, confirming that the flies consumed the solutions. Flies were maintained at 25.5 °C. Each day for four days, live and dead flies were counted. Four replicates were done for each treatment. Data were pooled for survival curve analysis (n = 100 flies). Survival curves were compared by performing a log-rank (Mantel-Cox) test with the use of GraphPad Prism software.

3. RESULTS

3.1. Tsf1^m10 deletion mutation

A Tsf1 deletion mutation, Tsf1^m10, was created by using CRISPR-Cas9 technology to delete the entire coding region of Tsf1 (Figure 1). The Tsf1^m10 mutation was outcrossed for eight generations to a WT line to decrease genetic variability between the Tsf1^m10 and WT lines. A lack of Tsf1 expression in the outcrossed Tsf1^m10 line was verified by RT-PCR and immunoblot analyses (Figure 1).

Figure 1. Verification of the outcrossed Tsf1^m10 line.

A) Location of guide RNAs used for generating a Tsf1 null mutation. The entire Tsf1 coding region was deleted and replaced with a 3xP3-GFP cassette. B) RT-PCR analysis of 25 whole adult females demonstrated a lack of Tsf1 transcripts in the Tsf1^m10 line. Gapdh2 was used as a reference gene. The expected size of the Tsf1 PCR product is 89 bp; the expected size of the gapdh2 product is 72 bp. C) Immunoblot analysis of 10 whole adult females demonstrated a lack of Tsf1 in the Tsf1^m10 line. The expected mass of Tsf1 is 68.7 kDa; the expected mass of β-tubulin, a reference protein, is 55 kDa. Genotypes: WT = w¹¹¹⁸/w¹¹¹⁸; Tsf1^m10 = w¹¹¹⁸ Tsf1^m10/w¹¹¹⁸ Tsf1^m10.

3.2. Tsf1 was present at a high concentration in larval, pupal and adult hemolymph

Serum transferrin and lactoferrin are highly abundant proteins; the concentration of serum transferrin in blood is typically 25-50 μM, and the concentration of lactoferrin in extracellular fluids is approximately 20-60 μM (Anderson and Vulpe, 2009; Weinberg, 2007). Tsf1 is known to be present in the larval and adult hemolymph of D. melanogaster (Iatsenko et al., 2020; Levy et al., 2004). We used immunoblot analyses to determine whether Tsf1 is present in pupal hemolymph and to estimate the concentration of Tsf1 in hemolymph from different developmental stages. We found that Tsf1 is present at a high concentration in the hemolymph of all three stages, but that the concentration was lowest in larvae (0.4 ± 0.17 μM), intermediate in pupae (1.4 μM), higher in adult females (4.4 ± 0.8 μM) and highest in adult males (22.2 ± 4.3 μM) (Figures 2 and S1).

Figure 2. Tsf1 was present in larval, pupal and adult hemolymph.

Immunoblot analysis was performed on 0.2 μl hemolymph from third instar larvae (L), pupae at stages P4-P7 (P), adult males (M), and adult females (F). Twenty-five ng purified Tsf1 (Tsf) was used as a positive control. The positions of molecular mass standards (in kDa) are shown to the left of the blot. The expected mass of Tsf1 is 68.7 kDa. One representative hemolymph sample from each developmental stage is shown. Hemolymph samples were collected from w¹¹¹⁸/w¹¹¹⁸ females and w¹¹¹⁸/Y males.

3.3. Apo-Tsf1 had bacteriostatic activity

To test whether Tsf1 from D. melanogaster has anti-microbial activity, we generated apo- and holo-forms of recombinant Tsf1, and then monitored growth of E. coli in their presence or absence. For this experiment, we used 1 μM Tsf1, which is within the range of concentrations we determined for Tsf1 in hemolymph. Like apo-lactoferrin, apo-Tsf1 inhibited bacterial growth (Figure 3). In contrast, holo-Tsf1 enhanced bacterial growth, presumably because the E. coli were able to use holo-Tsf1 as a source of iron (Figure 3). These results indicate that Tsf1 has the ability to inhibit bacterial growth by sequestering iron.

Figure 3. Apo-Tsf1 had bacteriostatic activity.

E. coli were grown for 8 hours in medium supplemented with apo-Tsf1 (1 μM), holo-Tsf1 (1 μM), or an equivalent volume of buffer. Bacterial growth, which was assessed by absorbance at 600 nm, was inhibited by apo-Tsf1 and enhanced by holo-Tsf1. Data are from three replicate cultures. Means ± SD are shown. Differences in means were assessed by performing unpaired t-tests.

3.4. Tsf1 was detected in tracheae, testes, oocytes and pericardial cells

To identify tissue-specific localization of Tsf1, we performed pilot screens of larval and adult tissues using immunohistochemistry and confocal microscopy to detect Tsf1. We observed Tsf1 in tracheae, testes, oocytes, and two types of nephrocytes (pericardial cells and garland cells) but no other tissue types (unpublished data). To verify positive screening results, we focused on tracheae from larvae, and on testes, oocytes and pericardial cells from adults. Because Tsf1 is a secreted, soluble protein, we would expect it to be mostly extracellular; however, if Tsf1 is endocytosed, we would also expect it to colocalize with the endosome marker Rab5 (Zeigerer et al., 2012).

Because Tsf1 is highly expressed in tracheae (Leader et al., 2018), we predicted that it would be secreted into an extracellular space inside the tracheae, where it could inhibit microbial growth. Our immunohistochemistry results support this prediction (Figure 4A). To demonstrate that the observed fluorescence was not a consequence of autofluorescence or non-specific antibody binding, we verified that trachea from Tsf1^m10 larvae, which were analyzed using identical procedures as WT tracheae, lacked red fluorescence (Figure 4B).

Figure 4. Tsf1 was detected in tracheae.

Tracheae from WT (A – A’”) and Tsf1^m10 (B – B’”) larvae were fixed and processed for immunofluorescence using antibodies against Tsf1 (red) and Rab5 (green). Images are of a single optical slice from the longitudinal center of the tracheal lumen. Scale bars are 50 μm. BF = bright field. Genotypes: WT female = w¹¹¹⁸/w¹¹¹⁸. Tsf1^m10 female = y w Tsf1^m10/y w Tsf1^m10. WT male = w¹¹¹⁸/Y. Tsf1^m10 male = y w Tsf1^m10/Y. (The sex of the larvae is unknown.)

Expression of Tsf1 in testes (Leader et al., 2018) could lead to accumulation of Tsf1 in the testes or possibly the ejaculation of Tsf1 with sperm. We detected Tsf1 in the proximal region of the testis and also in the seminal vesicle, but Tsf1 immunoreactivity was not apparent in accessory glands or ejaculatory ducts (Figure 5). The function of Tsf1 in the testes and seminal vesicles is not obvious, but it may inhibit microbial growth in the extracellular environment where sperm is stored.

Figure 5. Tsf1 was detected in testes and seminal vesicles.

Abdomens from WT (A) and Tsf1^m10 (B) adult males were fixed and processed for immunofluorescence using antibodies against Tsf1 (red) and Rab5 (green). Tsf1 immunoreactivity was observed in WT testes and seminal vesicles. Colocalization of Tsf1 and Rab5 was not observed. Each image is a maximum intensity projection of a z-stack. Scale bars are 100 μm. Abbreviations: ag = accessory gland, ed = ejaculatory duct, sv = seminal vesicle. Genotypes: WT = w¹¹¹⁸/Y. Tsf1^m10 = w¹¹¹⁸ Tsf1^m10/Y.

Although Tsf1 is not expressed in eggs or ovaries, Tsf1 is present in eggs (Brown et al., 2014; Casas-Vila et al., 2017; Graveley et al., 2011; Harizanova et al., 2004), suggesting that Tsf1 in hemolymph is taken up by developing oocytes. We observed colocalization of Tsf1 with Rab5 at the apex of oocytes, and Tsf1 within yolk granules (Figure 6). These results suggest endocytic uptake of Tsf1 by oocytes and storage of Tsf1 in yolk granules.

Figure 6. Tsf1 colocalized with Rab5 in oocytes.

A) A schematic diagram of a developing egg chamber. The surrounding layer of follicle epithelium cells (FE) is shown in tan, the border cell (BC) in yellow, the nurse cells (NC) in light blue, and the oocyte (OC) in green, with nuclei in dark blue. Egg chambers from WT (B and C – C”) and Tsf1^m10 (D) flies were fixed and processed for immunofluorescence using antibodies against Tsf1 (red) and Rab5 (green). Colocalization is yellow. C - C”) A magnified view of a single 0.48 μm optical slice of the WT oocyte. Arrows indicate Tsf1 accumulation in yolk granules. Arrow heads indicate colocalization of Tsf1 and Rab5 at the apex of the oocyte. Nuclei are marked with asterisks. Scale bars are 50 μm for panels B and D and 10 μm for panels C - C”. Genotypes of mothers: WT = w¹¹¹⁸/w¹¹¹⁸. Tsf1^m10 = y w Tsf1^m10/y w Tsf1^m10.

Colocalization of Tsf1 and Rab5 was also detected in pericardial cells (Figure 7). The physiological function of these cells is to filter hemolymph (Denholm and Skaer, 2009); therefore, evidence of endocytic uptake by pericardial cells is not surprising.

Figure 7. Tsf1 colocalized with Rab5 in pericardial cells.

A) A schematic diagram of the location of pericardial cells (PC) surrounding the dorsal vessel (DV) in the abdomen of adult flies. Pericardial cells from WT (B – B” and D – D”) and Tsf1^m10 (C) adult females were fixed and processed for immunofluorescence using antibodies against Tsf1 (red) and Rab5 (green). Colocalization is yellow. B - B” and C) Z-stack of 20-30 optical slices. In B – B”, the boundaries of the dorsal vessels are outlined with blue dashes, and the boundaries of the pericardial cells are outlined with grey dashes. D - D”) Magnified views of a 2 μm optical slice from the center of a wild-type pericardial cell. Scale bars are 50 μm for panels B - B” and C, and 10 μm for panels D - D”. Genotypes: WT = w¹¹¹⁸/w¹¹¹⁸. Tsf1^m10 = y w Tsf1^m10/y w Tsf1^m10.

We were interested in whether Tsf1 is endocytosed by fat body cells because previous studies demonstrated that, at least under some circumstances, Tsf1 is involved in the transfer of iron into the fat body (Huebers et al., 1988; Iatsenko et al., 2020; Xiao et al., 2019); however, our pilot screens failed to detect intracellular Tsf1 in fat body cells from larvae, adult females or adult males (unpublished data). In an attempt to optimize our chances of detecting intracellular Tsf1 in fat body cells, we analyzed adult males, which have the highest concentration of Tsf1 in hemolymph, and inoculated the flies with bacteria because this treatment elicits a Tsf1-mediated transfer of iron from hemolymph to the fat body (although whether this is due to endocytic uptake of Tsf1 is unknown) (Iatsenko et al., 2020). For these experiments, we collected fat body 0.5 – 3 hours after inoculation, which is when endocytic activity in fat body cells in response to immune challenge has been observed (Sorvina et al., 2013). We analyzed 96 fat body samples from inoculated males but did not observe red fluorescence above background level (Figure S2). These negative results may reflect a true lack of endocytic uptake of Tsf1 by fat body cells or they may indicate inadequate sensitivity of our immunohistochemistry methods.

3.5. The Tsf1^m10 mutation had no effect on viability and little effect on longevity

Mice with a severe deficiency in serum transferrin die shortly after birth, whereas mice without lactoferrin are viable; these differences are due to the essential role serum transferrin plays in iron transport and the non-essential role lactoferrin plays in immunity (Bernstein, 1987; Ward et al., 2003). In previous studies, RNAi-mediated knockdown of Tsf1 was associated with pupal mortality, whereas a Tsf1 null mutant line (Tsf1^JP94) was found to be viable; therefore, whether Tsf1 is an essential gene, like serum transferrin, was not certain (Bernstein, 1987; Iatsenko et al., 2020; Xiao et al., 2019). To address this question, we first tested the effect of RNAi-mediated knock down of Tsf1 on pupal mortality by performing crosses between a constitutive, ubiquitous actin-Gal4 driver line and four Tsf1 RNAi lines, and then analyzing the RNAi progeny for mortality. Percent mortality at the pupal stage ranged from 24% – 99%, depending on the Tsf1 RNAi line used (Figure S3). We predicted that the differences in mortality would be related to the degree of knock down of Tsf1. To test this hypothesis, we used western blot analyses to measure the percent reduction of Tsf1 in wandering larvae. Surprisingly, we found that percent mortality did not correlate with percent reduction in Tsf1 (Figure S3). These results suggest that the Tsf1 RNAi genotypes have fitness costs of undetermined origin. To avoid complications related to RNAi genotypes, we decided to use the Tsf1^m10 line instead of Tsf1 RNAi to perform a detailed lethal phase analysis of Tsf1.

Survival during the embryonic stage was evaluated by comparing the hatch rate of Tsf1^m10 eggs laid by Tsf1^m10 mothers and the hatch rate of WT eggs laid by WT mothers. No significant difference in hatch rate was observed (Figure 8A). Survival during the larval stages was evaluated by measuring the pupation rate of larvae selected in the first larval instar. No statistically significant difference in pupation rate was detected (Figure 8B). Survival during the pupal stage was evaluated by measuring the eclosion rate of insects selected as wandering third instar larvae. Almost all individuals of both genotypes survived, and no significant difference in survival was observed (Figure 8C). Taken together, these results demonstrate that Tsf1 is non-essential for survival and, thus, can be considered a non-essential gene.

Figure 8. Tsf1 is not essential for viability and had little effect on adult lifespan.

A) Egg hatch rate of WT eggs from WT mothers and Tsf1^m10 eggs from Tsf1^m10 mothers (n = 9 groups of 50 eggs). B) Pupation rate of WT and Tsf1^m10 larvae (n = 8 groups of 20 larvae). C) Eclosion rate of WT and Tsf1^m10 pupae (n = 8 groups of 20 pupae). Differences in survival rates at each developmental stage were evaluated by performing an unpaired t-test. No statistically significant differences were observed. D) Survival curves for WT and Tsf1^m10 adult females (n = 100 flies). E) Survival curves for WT and Tsf1^m10 adult males (n = 100 flies). The survival curves, shown with standard errors, were compared by performing a log-rank test. Tsf1^m10 females had slightly shorter live spans than WT females (P = 0.016), whereas Tsf1^m10 males survived slightly longer than WT males (P = 0.036). Genotypes: WT female = w¹¹¹⁸/w¹¹¹⁸. Tsf1^m10 female = w¹¹¹⁸ Tsf1^m10/w¹¹¹⁸ Tsf1^m10. WT male = w¹¹¹⁸/Y. Tsf1^m10 male = w¹¹¹⁸ Tsf1^m10/Y.

Given that an absence of Tsf1 may affect iron homeostasis or immunity or may result in oxidative stress, we predicted that Tsf1^m10 adults may have a shorter life span than WT adults. To test this hypothesis, we did survival analyses of mated females and mated males. We found that Tsf1^m10 females had a slightly shorter life span than WT females, whereas Tsf1^m10 males had a slightly longer life span than WT males (Figure 8). These results demonstrate that an absence of Tsf1 has little effect on longevity.

3.6. Whole body iron content and iron distribution were unaffected by a lack of Tsf1

In mice, a severe deficiency of serum transferrin causes iron overload in various tissues, and an absence of transferrin receptor causes iron deficiency in tissues and a decrease in total body iron accumulation (Bernstein, 1987; Levy et al., 1999). Therefore, if iron transport requires Tsf1, a lack of Tsf1 would be expected to affect the distribution of iron in the fly body and possibly the iron content of whole bodies. Previous experiments to test this hypothesis produced conflicting results. RNAi-mediated knock down of Tsf1 resulted in changes in iron distribution in the gut and fat body (as measured by a ferrozine-based assay), whereas the Tsf1^JP94 null mutation did not affect iron content of fat body, hemolymph, ovaries or thoraxes (as measured by inductively coupled plasma optical emission spectrometry) (Iatsenko et al., 2020; Xiao et al., 2019). For this study, we used a ferrozine-based method to compare the iron content of whole bodies, midguts, fat body, hemolymph, and heads from Tsf1^m10 and WT flies, and eggs from Tsf1^m10 and WT mothers.

We observed no differences in the iron content of whole adult mated females or males at 5 or 14 days post-eclosion (Figure 9A–D). We were interested in midguts because they take up iron from the diet, whereas other tissues must be supplied with iron from hemolymph. No difference between Tsf1^m10 and WT midguts was observed (Figure 9E). We also observed no differences in the iron content of fat body, hemolymph or heads (which contain various neural tissues) (Figure 9E–H). Given that Tsf1 appears to be transported into oocytes from the mother’s hemolymph (section 3.4), we predicted that eggs from Tsf1^m10 mothers would have less iron than eggs from WT mothers; however, we detected no difference in iron content (Figure 9I). The fat body and hemolymph results are consistent with those from the Tsf1^JP94 mutant study. Taken together, these results indicate that Tsf1 is not essential for iron transport in healthy adult flies.

Figure 9. The Tsf1^m10 mutation had no effect on the iron content of whole bodies, tissues or eggs.

The iron content of whole body and tissue homogenates from WT and Tsf1^m10 insects was normalized to total protein. A) Five day old females. B) Five day old males. C) Fourteen day old females. D) Fourteen day old males. A – D) Analyses of adults were of n = 7 groups of 25 flies each. E) Midguts from five day old females (n = 5 samples of ~80 midguts). F) Hemolymph from five day old females (n = 4 samples of hemolymph from ~600 flies each). G) Fat body-enriched abdominal carcasses from five day old females (n = 4 samples of 90 abdominal carcasses). H) Heads from five day old females (samples of ~100 heads; WT: n = 6, Tsf1^m10: n = 7). I) WT eggs from WT mothers and Tsf1^m10 eggs from Tsf1^m10 mothers (n = 8 samples of 30-60 μl eggs). Differences in iron content were assessed by performing an unpaired t-test. No statistically significant differences were observed. Genotypes: WT female = w¹¹¹⁸/w¹¹¹⁸. Tsf1^m10 female = w¹¹¹⁸ Tsf1^m10/w¹¹¹⁸ Tsf1^m10. WT male = w¹¹¹⁸/Y. Tsf1^m10 male = w¹¹¹⁸ Tsf1^m10/Y.

3.7. Tsf1^m10 flies had more iron than WT flies in the iron region of the midgut

D. melanogaster, like other cyclorrhaphous dipteran species, have a short, specialized, acidic midgut region that is involved in uptake of iron from the diet (Miguel-Aliaga et al., 2018; Missirlis et al., 2007; Poulson, 1950). Cells in this region (iron cells) accumulate ferritin-bound iron that can be detected in dissected midguts by Prussian blue staining (Mehta et al., 2009). Although the iron content of whole midguts from Tsf1^m10 and WT flies was indistinguishable, we decided to analyze iron accumulation in iron cells because a previous study suggested that Tsf1 influences the export of iron from midgut cells (Xiao et al., 2019). We found that WT flies had little or no detectable Prussian blue staining in the iron cell region, whereas Tsf1^m10 flies had obvious staining of ferritin-bound iron (Figure 10). This result demonstrates an influence of Tsf1 on a particular aspect of iron distribution, and supports the hypothesis that Tsf1 is involved in transporting iron out of iron cells.

Figure 10. Tsf1^m10 flies had more iron than WT flies in the iron region of the midgut.

After two days of a high iron diet followed by one day of a normal diet, midguts from adult females were analyzed by Prussian blue staining to detect ferritin-bound iron. A) Most midguts from WT flies had no detectable staining (top panel), and 14% had faint blue staining (middle panel). In contrast, all midguts from Tsf1^m10 flies had an obvious blue color (lower panel). B) A comparison of the number of midguts with detectable blue color demonstrated a significant difference (P < 0.001) between WT and Tsf1^m10 flies (Fisher’s exact test, n = 21 midguts per genotype). Genotypes: WT = w¹¹¹⁸/w¹¹¹⁸. Tsf1^m10 = w¹¹¹⁸ Tsf1^m10/w¹¹¹⁸ Tsf1^m10.

3.8. Iron, cadmium and paraquat treatments had no detectable effect on Tsf1 abundance

In D. melanogaster, Tsf1 abundance increased in larvae fed a high iron diet, and in the beetle A. germari, Tsf1 abundance increased in response to several types of stress-elicitors, including injection of iron (Lee et al., 2006; Xiao et al., 2019). We used immunoblot analyses of individual flies to test whether Tsf1 abundance increases in response to various types of stress. In D. melanogaster, upregulation of Tsf1 in response to immune challenge is well established (Iatsenko et al., 2020; Yoshiga et al., 1999); therefore, to validate our method, we verified that inoculation with bacteria resulted in an increase in Tsf1 abundance (Figure 11A).

Figure 11. Effect of iron, cadmium, paraquat and copper on Tsf1 abundance.

Immunoblot analyses were used to measure changes in Tsf1 abundance in response to treatment. A) Naive (N) females and females inoculated (In) with bacteria were analyzed 24 h after treatment. As expected, inoculation with bacteria caused an increase in Tsf1 abundance. B) Adult females were injected with water (W), 5 mM ferrous sulfate (Fe²⁺), or 5 mM ferric citrate (Fe³⁺), and flies were analyzed 24 hours after injection. No effect from injection of iron was observed. C - J) Flies were fed sucrose solution (S) or sucrose solution with 10 mM iron (Fe) (C, D), 25 mM cadmium (Cd) (E), 20 mM paraquat (PQ) (F – H), or 15 mM copper (Cu) (I, J). Female (C, E, F, H, I) and male (D, G, J) flies were analyzed after feeding for 24 h (F, G) or 48 h (C-E, H-J). Except for the positive control treatment (A), only ingestion of copper resulted in a change in Tsf1 abundance (I, J). For all treatments, n = 8 individual flies. GAPDH (A) or β-tubulin (B – J) was used to normalize Tsf1 band intensity. Differences in Tsf1 abundance were assessed by performing unpaired t-tests. Comparisons in F and H were done with Welch’s correction because of differences in variances between the two sets of data. (Note that the units for normalized net intensity are of unspecified value; therefore, comparisons between graphs should not be made.) Genotypes: Female = w¹¹¹⁸/w¹¹¹⁸. Male = w¹¹¹⁸/Y.

One possible function of Tsf1 may be protection against iron-induced oxidative stress. We tested this hypothesis by feeding flies 10% sucrose containing 10 mM ferric ammonium citrate (a concentration of iron that is known to reduce longevity in flies (Lang et al., 2012)) and by injecting flies with 5 mM ferrous sulfate or 5 mM ferric citrate. Surprisingly, we observed no change in Tsf1 abundance (Figure 11B–D). These results indicate that the concentration of Tsf1 in adults can remain stable even when the flies are exposed to a large excess of iron.

At very high concentrations, cadmium can cause oxidative stress by inhibition of superoxide dismutase (SOD) and catalase, leading ultimately to the production of reactive oxygen species (Korsloot et al., 2004). To test whether ingestion of cadmium leads to a change in Tsf1 abundance, we fed adult females a high concentration of cadmium chloride in 10% sucrose. Because 25 mM cadmium chloride causes ~ 50% mortality in flies after 4 days of feeding (Zhou et al., 2017), we used this concentration for our experiment. We observed no change in Tsf1 abundance after 48 hours of feeding (Figure 11E).

Paraquat (a quaternary ammonium bipyridyl compound) generates oxidative stress by causing the production of superoxide anion in mitochondria. Feeding flies 20 mM paraquat in 5% sucrose causes oxidative stress within 24 hours (Shukla et al., 2016); therefore, we chose these conditions for our experiments. We observed no change in Tsf1 abundance in males or females after 24 hours of feeding and no change in females after 48 hours of feeding (Figure 11F–H). (Most males had died within 48 hours and so were not analyzed.)

We were interested in the effect of ingestion of copper for two reasons. First, excess copper can cause oxidative stress (Gaetke and Chow, 2003; Korsloot et al., 2004). Second, transcriptomics data suggested that Tsf1 expression may be influenced by copper (Brown et al., 2014). We detected a significant increase in Tsf1 abundance in females and males that had ingested 15 mM copper sulfate for 48 hours (Figure 11 I–J). Given that other oxidative-stress inducing treatments did not lead to an increase in Tsf1, it seems likely that this change in Tsf1 abundance is unrelated to oxidative stress, and we do not know its physiological significance.

3.9. An absence of Tsf1 results in less sensitivity to paraquat

Tsf1 RNAi in A. germari and another beetle, P. brevitarsis, resulted in increased stress-induced apoptosis of fat body cells, suggesting that Tsf1 protects against physiological stress (for example, injection of hydrogen peroxide) in these insects (Kim et al., 2008; Lee et al., 2006). In contrast, Tsf1 RNAi resulted in less oxidative stress in neural tissues of D. melanogaster treated with rotenone (Xue et al., 2020). To test whether an absence of Tsf1 results in a change in susceptibility to oxidative stress conditions, we fed Tsf1^m10 and WT adult females 20 mM paraquat in 5% sucrose solution for 4 days. Green food coloring added to the diet was used to verify that the flies ate the diet containing paraquat (Figure S4). Tsf1^m10 and WT flies fed a control diet of 5% sucrose survived during the course of the experiment (Figure 12A). In contrast, Tsf1^m10 and WT flies fed 20 mM paraquat had high mortality within 4 days, but Tsf1^m10 flies survived better (P < 0.0001) (Figure 12A). To verify that the Tsf1^m10 flies are less susceptible to paraquat than WT flies, we repeated the experiment with 10 mM paraquat, and we observed the same outcome (P < 0.0001).

Figure 12. Tsf1^m10 flies were less susceptible than WT flies to paraquat.

Adult females were fed sucrose or sucrose containing paraquat (PQ) for four days, and each day the number of live and dead flies was counted. WT and Tsf1^m10 flies survived on the sucrose-only diet (A, lines on graph overlap). More WT than Tsf1^m10 flies died when fed 20 mM paraquat (A) (P < 0.0001) or 10 mM paraquat (B) (P < 0.0001). Each treatment group contained 100 flies. Standard errors are shown. Survival curves were compared by performing a log-rank test. Genotypes: WT = w¹¹¹⁸/w¹¹¹⁸; Tsf1^m10 = w¹¹¹⁸ Tsf1^m10/w¹¹¹⁸ Tsf1^m10.

4. DISCUSSION

The goal of this study was to better understand the roles of Tsf1 in iron transport, immunity, and protection against oxidative stress. Our hypotheses were based on the iron transport function of serum transferrin and the immune-related iron sequestration function of lactoferrin, as well as on the results of previous studies of Tsf1 in D. melanogaster and other insects. Although Tsf1 is sometimes viewed as an ortholog of serum transferrin, phylogenetic studies have demonstrated that this is not the case (Bai et al., 2016). Tsf1 arose early in insect evolution, whereas serum transferrin evolved with the placental mammal lineage (Hughes and Friedman, 2014; Najera et al., 2020). The serum transferrin gene appears to have arisen from a duplication of an ancestral gene that encoded lactoferrin (Farnaud and Evans, 2003; Hughes and Friedman, 2014; Ward et al., 2003); therefore, the ancestral role of members of the transferrin family may have been immune-related.

The hypothesis that Tsf1 may have a lactoferrin-like role was proposed based on the observation that a mosquito Tsf1 was upregulated in response to infection (Yoshiga et al., 1997). Since that time, upregulation of Tsf1 in a wide range of insect species in response to a wide range of infectious agents has been shown (Geiser and Winzerling, 2012; Iatsenko et al., 2020). RNAi-mediated knock down of Tsf1 increases immune susceptibility in Glossina morsitans and Plutella xylostella (Hrdina and Iatsenko, 2021), and a Tsf1 null mutation in D. melanogaster causes susceptibility to a subset of pathogenic microbes (Iatsenko et al., 2020). The mechanisms by which Tsf1 protects against infection are not fully understood. The simplest mechanism would be lactoferrin-like iron sequestration. Our finding that D. melanogaster Tsf1 has bacteriostatic activity in its apo-form but not holo-form supports this model, as does a previous study of the bacteriostatic activity of M. sexta Tsf1 (Brummett et al., 2017). A lactoferrin-like mechanism for Tsf1 seems likely in extracellular fluids such as hemolymph, molting fluid, saliva and seminal fluid, which have been shown to contain Tsf1 (Bonilla et al., 2015; Brummett et al., 2017; Geiser and Winzerling, 2012; Hattori et al., 2015; Qu et al., 2014; Simmons et al., 2013; Zhang et al., 2014). Our study demonstrated that Tsf1 is also present in tracheae, testes, and seminal vesicles, and we suggest that Tsf1 may have a lactoferrin-like role in these locations as well. A weakness of a simple iron sequestration mechanism would be if some species of microbe could use iron-bound Tsf1 as an iron source. We demonstrated that holo-Tsf1 increased the growth of E. coli, suggesting that these bacteria are able to scavenge iron from holo-Tsf1. These results are consistent with those of a previous study of Tsf1 from M. sexta (Brummett et al., 2017). The ability of microbes to use holo-Tsf1 as an iron source is also suggested by the discovery that Tsf1 RNAi in Apis mellifera decreased susceptibility to infection by Nosema ceranae and that D. melanogaster holo-Tsf1 promotes growth of Spiroplasma poulsonii (Marra et al., 2021; Rodríguez-García et al., 2021). A more complex mechanism by which Tsf1 may protect against infection is by mediating the transfer of iron from hemolymph to fat body cells in response to an immune challenge (Iatsenko et al., 2020). The movement of iron-bound Tsf1 from hemolymph to an intracellular environment would result in withholding of iron from pathogens present in hemolymph. Although we did not detect evidence of immune-induced endocytic uptake of Tsf1, our immunohistochemistry methods may not have been sensitive enough to detect Tsf1 in endosomes, or Tsf1-mediated iron transfer to the fat body make occur without endocytosis.

In A. germari, Tsf1 was upregulated in response to injection of iron, hydrogen peroxide, or paraquat, and in response to wounding, immune challenge, or heat or cold shock (Lee et al., 2006). In addition, Tsf1 RNAi in A. germari and P. brevitarsis resulted in increased stress-induced apoptosis of fat body cells, suggesting that Tsf1 protects against various types of physiological stress in these beetles (Kim et al., 2008; Lee et al., 2006). The mechanism of protection is unknown, but a lack of Tsf1 may cause oxidative stress in the beetles by causing an increase in non-protein-bound iron that could participate in the Fenton reaction, leading to the production of reactive oxygen species (Kim et al., 2008; Lee et al., 2006). Based on our study, this does not appear to be the case in D. melanogaster. Whereas Tsf1 abundance in beetles was enhanced in response to many types of stress-eliciting treatments, we observed no increase in response to injection of iron or ingestion of iron, cadmium or paraquat. Tsf1 abundance did increase in response to ingestion of copper, but, given the lack of response to other stress-elicitors, it seems unlikely that the response to copper is related to oxidative stress. We also observed little difference in the life span of Tsf1^m10 adults compared with WT adults, and we found that Tsf1^m10 flies were less susceptible than WT flies to ingested paraquat. These results are consistent with our finding that Tsf1^m10 flies did not have an excess of iron in hemolymph, fat body or heads; therefore, it is likely that an absence of Tsf1 does not cause an accumulation of ferrous ions in cells that could participate in the production of reactive oxygen species. Our results are consistent with the finding that Tsf1 RNAi resulted in less oxidative stress in neural tissues of flies treated with rotenone (Xue et al., 2020).

How iron is transported from one cell to another in insects is still not understood (Calap-Quintana et al., 2017; Mandilaras et al., 2013; Tang and Zhou, 2013b; Whiten et al., 2018). Because the main mechanisms of iron transport in mammals involve serum transferrin (Frazer and Anderson, 2014; Kosman, 2020), one reasonable hypothesis has been that Tsf1 plays a similar role in insects. Like serum transferrin, Tsf1 releases iron under moderately acidic conditions, suggesting that Tsf1 could release iron in acidified endosomes or lysosomes (Baker et al., 2002; Weber et al., 2020a). By using immunohistochemistry, we found evidence of endocytic uptake of Tsf1 only in nephrocytes and oocytes. Whether endocytic uptake of Tsf1 by other cell types occurs is not certain, given that our immunohistochemistry methods may be inadequate for detecting small amounts of Tsf1 in endosomes. The lack of an identifiable Tsf1 receptor in insects indicates that, if endocytic uptake of Tsf1 does occur in tissues other than nephrocytes and oocytes, the uptake mechanism must have evolved separately from the mechanisms in mammals.

It was not surprising to observe evidence of endocytic uptake of Tsf1 by nephrocytes, since nephrocytes accumulate hemolymph proteins as they filter hemolymph to regulate its composition (Denholm and Skaer, 2009). Uptake by nephrocytes could provide a mechanism for Tsf1-bound iron to be released and recycled. For example, within the endosomes of a nephrocyte, iron may be released from Tsf1 and then transported into the secretory pathway, where the iron could be bound by ferritin and secreted.

We also observed evidence of endocytic uptake of Tsf1 by oocytes. This result is consistent with previous studies demonstrating that Tsf1 is abundant in D. melanogaster eggs, but that Tsf1 is not expressed in eggs or ovaries (Brown et al., 2014; Casas-Vila et al., 2017; Graveley et al., 2011; Harizanova et al., 2004). Similar results were observed for Riptortus clavatus (bean bug) and Sarcophaga peregrina (flesh fly) (Hirai et al., 2000; Kurama et al., 1995). Whether apo-Tsf1 is taken up by insect oocytes is not known, but iron-bound Tsf1 was shown to be transported into S. peregrina oocytes (Kurama et al., 1995). Based on these observations, we predicted that eggs from Tsf1^m10 mothers would have less iron than eggs from WT mothers; however, eggs from Tsf1^m10 mothers were not iron deficient. This result is consistent with our observation that the eggs from Tsf1^m10 mothers were viable and had a wild-type hatch rate. Our results indicate that Tsf1 is not essential for the uptake of iron by oocytes. Instead, the main function of Tsf1 in oocytes and eggs may be immune-related iron sequestration (similar to the role of mammalian lactoferrin or avian ovotransferrin) (Farnaud and Evans, 2003; Giansanti et al., 2012).

Because serum transferrin is essential for regulated iron transport, mice with a severe deficiency of serum transferrin die shortly after birth (Anderson and Vulpe, 2009; Bernstein, 1987). In contrast, we found that Tsf1^m10 insects are viable at all stages of development. This finding supports a previous observation that the independent Tsf1^JP94 null mutant line is viable (Iatsenko et al., 2020). On the other hand, the viability of Tsf1^m10 flies seems inconsistent with the pupal mortality associated with Tsf1 RNAi genotypes, as observed in this study and a previous study (Xiao et al., 2019). Perhaps physiological stress associated with the Gal4/UAS-mediated RNAi process causes pupae to be sensitive to a lack of Tsf1. Alternatively, because the degree of Tsf1 knock down observed with four Tsf1 RNAi lines did not correlate with percent pupal mortality, the mortality seen in Tsf1 RNAi insects may be due, at least in part, to off-target, positional, or genetic background effects or other complicating factors (see below).

Mice and humans with a severe deficiency of serum transferrin have iron overload of various tissues, and mice with a null mutation in the transferrin receptor gene have iron-deficient tissues and a decrease in whole body iron (Bernstein, 1987; Hamill et al., 1991; Hayashi et al., 1993; Levy et al., 1999). Therefore, we predicted that Tsf1^m10 flies may have abnormal iron distribution in the body and possibly altered iron content of the whole body. Instead, we observed no difference in the iron content of midguts, hemolymph, fat body, heads, eggs or whole bodies. These results extend those of a study by Iatsenko et al. (2020), which found no difference in the iron content of fat body, hemolymph, ovaries and thoraxes from Tsf1^JP94 and control flies. On the other hand, our results may be inconsistent with the finding that larvae with RNAi-mediated knockdown of Tsf1 in the fat body had less iron in the fat body and more iron in the midgut than control flies (Xiao et al., 2019).

We identified one difference in iron content between Tsf1^m10 and WT flies, and that was in the amount of iron in the midgut cells that take up iron from the diet. Somewhat surprisingly, Tsf1^m10 flies had more iron in these cells, suggesting that iron export from these cells was inhibited by the absence of Tsf1. One explanation for this result is the possibility that Tsf1 is involved in iron export. This would be a novel function for a transferrin family member. Previously published support for this hypothesis is the increase in midgut iron in Tsf1 RNAi larvae (Xiao et al., 2019). Authors of that study suggested the interesting possibility that Tsf1 may be secreted from midgut cells in its iron-bound form (Xiao et al., 2019). Secretion of iron-bound Tsf1 could answer the question of how iron-loading of Tsf1 occurs prior to iron transport but would complicate the question of how apo-Tsf1 protects against infection.

As described above, phenotypes of the CRISPR-engineered Tsf1^m10 null line are consistent with phenotypes of the independent Tsf1^JP94 null line; however, Tsf1 mutant phenotypes appear to be in conflict with some of the Tsf1 RNAi phenotypes (Iatsenko et al., 2020; Xiao et al., 2019). RNAi is a powerful method for learning about gene functions; however, it can suffer from false-positive outcomes, including those due to potential off-target effects, positional effects, and fitness costs associated with expression of RNAi transgenes (Heigwer et al., 2018). In addition, given the role of Tsf1 in innate immunity (Iatsenko et al., 2020), differences in the microbiomes of the various fly lines may influence results. By using CRISPR-Cas9 technology to create a deletion mutation, which was then extensively outcrossed, our study avoided some of these complicating factors.

This study and previous studies suggest that Tsf1 influences iron transport but is not essential for iron transport (Iatsenko et al., 2020; Xiao et al., 2019). If there is a robust compensatory iron transport mechanism that functions in individuals deficient in Tsf1, it is possible that Tsf1 plays a role in iron transport than is not apparent from Tsf1 RNAi and mutant analyses. Such a mechanism is unlikely to involve a different transferrin family member because the other D. melanogaster transferrins do not appear to play a significant role in iron homeostasis (Najera et al., 2020). A more likely mechanism would involve the secretion and uptake of ferritin. It is well established that iron is transported out of insect cells and into the hemolymph by the secretion of iron-loaded ferritin (Nichol and Locke, 1990; Pham and Winzerling, 2010; Tang and Zhou, 2013b; Whiten et al., 2018; Xiao et al., 2014). In addition, several lines of evidence suggest that ferritin is involved in the transport of iron from one cell to another (Huebers et al., 1988; Li, 2010; Tang and Zhou, 2013a; Zhou et al., 2007). Therefore, as has been suggested previously, iron transport in insects is likely to be mediated primarily by a ferritin-based mechanism (Xiao et al., 2019). Changes in expression of ferritin in Tsf1 RNAi insects supports this hypothesis (Xiao et al., 2019).

Supplementary Material

1

Figure S1. Quantitation of Tsf1 in hemolymph.

Figure S2. Tsf1 was not detected in fat body.

Figure S3. Percent reduction of Tsf1 in RNAi flies did not correlate with percent mortality in RNAi insects.

Figure S4. Paraquat solution was visible in fly abdomens.

• D. melanogaster transferrin-1 (Tsf1) was detected in hemolymph, testes, tracheae and oocytes

• Apo-Tsf1 had bacteriostatic activity

• A Tsf1 deletion mutation had no effect on viability and little effect on longevity

• The Tsf1 deletion mutant had reduced sensitivity to paraquat-induced oxidative stress

• The Tsf1 mutant had more iron in midgut iron cells but no other differences in iron distribution