P-glycoprotein detoxification by the Malpighian tubules of the desert locust

Detoxification is essential for allowing animals to remove toxic substances present in their diet or generated as a biproduct of their metabolism. By transporting a wide range of potentially noxious substrates, active transporters of the ABC transporter family play an important role in detoxification. One such class of transporters are the multidrug resistance P-glycoprotein transporters. Here, we investigated P-glycoprotein transport in the Malpighian tubules of the desert locust (Schistocerca gregaria), a species whose diet includes plants that contain toxic secondary metabolites. To this end, we studied transporter physiology using a modified Ramsay assay in which ex vivo Malpighian tubules are incubated in different solutions containing the P-glycoprotein substrate dye rhodamine B in combination with different concentrations of the P-glycoprotein inhibitor verapamil. To determine the quantity of the P-glycoprotein substrate extruded we developed a simple and cheap method as an alternative to liquid chromatography–mass spectrometry, radiolabelled alkaloids or confocal microscopy. Our evidence shows that: (i) the Malpighian tubules contain a P-glycoprotein; (ii) tubule surface area is positively correlated with the tubule fluid secretion rate; and (iii) as the fluid secretion rate increases so too does the net extrusion of rhodamine B. We were able to quantify precisely the relationships between the fluid secretion, surface area, and net extrusion. We interpret these results in the context of the life history and foraging ecology of desert locusts. We argue that P-glycoproteins play an important role in the detoxification by contributing to the removal of xenobiotic substances from the haemolymph, thereby enabling gregarious desert locusts to maintain toxicity through the ingestion of toxic plants without suffering the deleterious effects themselves.

generated as a biproduct of their metabolism. By transporting a wide range of potentially noxious 23 substrates, active transporters of the ABC transporter family play an important role in detoxification. 24 One such class of transporters are the multidrug resistance P-glycoprotein transporters. Here, we 25 investigated P-glycoprotein transport in the Malpighian tubules of the desert locust (Schistocerca 26 gregaria), a species whose diet includes plants that contain toxic secondary metabolites. To this end, 27 we studied transporter physiology using a modified Ramsay assay in which ex vivo Malpighian tubules 28 are incubated in different solutions containing the P-glycoprotein substrate dye rhodamine B in 29 combination with different concentrations of the P-glycoprotein inhibitor verapamil. To determine the 30 quantity of the P-glycoprotein substrate extruded we developed a simple and cheap method as an 31 alternative to liquid chromatography-mass spectrometry, radiolabelled alkaloids or confocal 32 microscopy. Our evidence shows that: (i) the Malpighian tubules contain a P-glycoprotein; (ii) tubule 33 surface area is positively correlated with the tubule fluid secretion rate; and (iii) as the fluid secretion 34 rate increases so too does the net extrusion of rhodamine B. We were able to quantify precisely the 35 relationships between the fluid secretion, surface area, and net extrusion. We interpret these results in 36 the context of the life history and foraging ecology of desert locusts. We argue that P-glycoproteins 37 play an important role in the detoxification by contributing to the removal of xenobiotic substances 38

Introduction 42
Insect excretory systems consist primarily of the Malpighian tubules and the hindgut, which act 43 synergistically to regulate haemolymph composition [1,2]. Malpighian tubules are blind ended tubules 44 that float in the haemolymph and empty into the gut at the midgut-hindgut junction, secreting primary tubules are considered analogous to vertebrate nephrons [2]. Cells of the epithelium forming the tubule 47 wall express primary and secondary active transporters that move K + , Na + and Clions into the lumen 48 creating an osmotic gradient that produces water secretion (for a review see [4]). Insects regulate ion 49 and water secretion according to their feeding habits and ecological niche. For example, 50 haematophagous insects must cope with an excess of NaCl and water after a blood meal [5], whereas 51 phytophagous insects must often cope with a diet rich in K + as well as with secondary metabolites [6,7]. 52 In addition to osmoregulation, Malpighian tubules play a fundamental role in the removal of 53 metabolic waste and potentially noxious substances that have been ingested [1,8]. Alkaloids and organic 54 anions and cations are actively transported by ATP-dependant transporters such as the multidrug 55 resistance-associated protein 2 (MRP2) and P-glycoproteins (P-gps, multidrug resistance protein 56 (MDR1) or ABCB1), both members of the ABC transporter family [9,10]. Multidrug resistance-57 different xenobiotics can act synergistically to maximize the efficacy of insecticides in pests or impair 74 the xenobiotic detoxification of beneficial insects such as honey bees [16]. 75 Desert locusts (Schistocerca gregaria) are generalist phytophagous insects with aposematic 76 coloration in the gregarious phase. They feed on a wide range of plants including those, such as 77 Schouwia purpurea and Hyoscyamus muticus, that contain toxins to become unpalatable and toxic to 78 predators [17-20] Nevertheless, it is likely that gregarious desert locusts excrete some of the toxins that 79 they ingest, relying instead on their gut contents to maintain toxicity [21,22]. Two lines of evidence 80 suggest that this excretion is likely to involve P-glycoproteins: (1) they are expressed in the Malpighian 81 tubules of numerous species (e.g. A. domesticus, L. migratoria, P. americana) from orthopteroid orders 82 [15]; and (2) they are expressed in the blood brain barrier of the desert locust [23]. However, P-83 glycoproteins in the Malpighian tubules of desert locusts have not been studied previously. 84 Here we show that xenobiotic transport and extrusion in the Malpighian tubules of the desert 85 locust is an active process dependent upon P-glycoprotein like transporters using isolated tubules to 86 perform a modified Ramsay secretion assay [24]. We evaluated the extrusion of the P-glycoprotein 87 substrate dye rhodamine B (e.g. [25,26]) with or without the addition of the selective P-glycoprotein 88 inhibitor verapamil (e.g. [23,27,28]). Our results suggest that P-glycoprotein transporters may play an 89 important role in the xenobiotic detoxification in the Malpighian tubules of the desert locust. By using 90 linear mixed effect models to account for repeated observations of single tubules and obtaining multiple 91 tubules from single locusts, we found that tubule surface area more accurately predicts fluid secretion 92 rate than diameter or length. Moreover, this statistical model allowed us to quantify the influence of the 93 surface area on the fluid secretion rate in different treatments, and how it changes over time. We found 94 that the surface area of the tubules positively influences their fluid secretion rate and that the fluid 95 secretion rate influences the net extrusion of rhodamine B. We propose that this assay may be used in 96 future to understand the physiology of the P-glycoproteins when exposed to a wide range of different 97 substances. 98 Animals saline. The head was separated from the saline using modelling clay (Plasticine®) (Fig 1B). The 126 preparation was pinned at the cut distal end of the gut to prevent it from floating. 127  The head is separated from the saline bath by a barrier of modelling clay. ( with paraffin oil. We photographed the droplets against a white background at the same light intensity, and white balancing the camera before shooting. All the images were analysed subsequently using 186 ImageJ v.1.51p software [29]. 187 Droplet colour varied from white (transparent droplet at rhodamine B concentration = 0 μM) 188 to intense pink, depending upon the rhodamine B concentration. We split each image into the 189 component colour channels and measured the intensity of the green channel. To control for the 190 background, we compared the mean intensity inside the droplet with that outside using the formula = 191 − , where is the intensity, is the intensity inside the droplet, and is the intensity outside the 192 droplet. We used a range of droplet diameters from 138 μm to 999 μm. 193 To validate the reliability of using the green channel, we also measured the magenta channel  Tukey method. All plots were made using the 'ggplot2' package [36]. 232 To investigate the effect of the treatments on the fluid secretion rate and the net extrusion of 233 rhodamine B, we analysed the interaction between the treatment (categorical), time of incubation 234 (categorical) and the surface area (continuous (mm 2 )). For the rhodamine B concentration, we analysed 235 the interaction between treatment (categorical) and time (categorical). To account for the nested 236 structure of data, we included the individual locust as random intercept in the model. We also included 237 tubule identity as a random intercept and time as random slope to account for the repeated 238 measurements on the same individual tubule. To investigate the effect of the fluid secretion rate on the net extrusion of rhodamine B we analysed the interaction of the variables secretion rate, treatment and 240 time including as before the individual locust as random intercept, and tubule identity as a random 241 intercept and time as random slope. To simplify the interpretation of the regression estimates, we 242 centred the surface variable on its mean. Therefore, all the estimates and comparisons are referred to a 243 tubule with a mean surface area. 244 245

246
We prepared three Malpighian tubules from each locust (see Materials and Methods; Fig 1). Each tubule 247 was punctured near the proximal end to allow the luminal fluid to be secreted and then they were 248 allowed to equilibrate in the saline bath for 30 minutes ( Fig 1D). The saline bath was then replaced with 249 one of four treatments: saline (control); rhodamine B 60 μM (R60); rhodamine B 60 μM + verapamil 250 125 μM (V125); and rhodamine B 60 μM + verapamil 250 μM (V250) (Fig 1E). Six locusts were used 251 for each of the treatments except for the R60 treatment in which eight locusts were used. Every 30 252 minutes the droplet secreted by the tubule during the Ramsay assay was removed. 253 254 Fluid secretion rate and surface area 255 We determined the fluid secretion rate of each tubule from the volume of the droplet secreted after each 256 30-minute interval up to 90 minutes after the start of the treatment. Thus, for each tubule we had three 257 measurements of the secretion rate in each of the four treatments. In total, there were 233 treatment 258 observations (one droplet was lost after 60 minutes for the V250 treatment) from 78 Malpighian tubules. 259 To determine whether the surface area of the Malpighian tubules exposed to the bath solution 260 influences the fluid secretion rate, we measured the length and diameter of each tubule immersed in the 261 saline or treatment. By comparing linear mixed effect models that incorporated these measurements of 262 length, diameter or surface area, we determined that surface area was the best explanatory variable (S1 263   Table). There was no difference in the surface area of Malpighian tubules exposed to the bathing solution among the treatments (F3,22.02=0.488, p=0.694; Control: 2.00 ± 0.09 mm 2 (mean ± S.E.); R60: 265 2.01 ± 0.06 mm 2 ; V125: 2.27 ± 0.06 mm 2 ; V250: 2.29 ± 0.07 mm 2 ). 266 The surface area of the tubule exposed in the bathing solution influenced the fluid secretion 267 rate depending on the treatment (F3,66.29=3.25, p=0.027; Fig 2; Table 1A). Throughout the whole period  The model applied was (Secretion rate ~ surface * treatment + time + (1| locust) + (1+time|tubule)).

(A) 287
Estimates of the influence of each unit of surface (mm 2 ) on the fluid secretion rate for each treatment.

Renewing bath saline increases the fluid secretion rate 306
To exclude the possibility that the decrease in the fluid secretion rate was caused by a damage of the 307 tubules during the Ramsay assay, we replaced the saline bath after 90 minutes with fresh saline, to 308 determine whether tubules would increase their fluid secretion rate to previous levels. We removed 309 three tubules from six locusts (17 tubules in total, one tubule excluded) and incubated them in saline 310 for 90 minutes, removing the droplet secreted every 30 minutes. After 90 minutes the saline bath was 311 removed, replaced with fresh saline. The tubules were then incubated for further 90 minutes removing 312 the droplet secreted every 30 minutes. The secretion rate decreased after 60 and 90 minutes (S4 Fig, S2 313 Table), but increased after 120 minutes following replacement of the saline (S4 Fig, S2 Table). To exclude the possibility that manipulation during the Ramsay assay altered the diameter of the 317 tubules, we measured the diameter of the tubules in vivo, at the beginning, and at the end of the assay. 318 We found that the diameter of the tubules was unaffected by the assay and was comparable to the 319 tubule's diameter in vivo (S5 Fig, S3A,B Table). 320 321 Net extrusion of Rhodamine B 322 We determined the concentration of Rhodamine B in each of the droplets collected from the Malpighian 323 tubules exposed to the R60, V125 and V250 treatments at each time point. There was a significant 324 interaction between treatment and time (F4,73.19=15.19, p<.001, Fig 4A), indicating that the rhodamine 325 B concentration changed over time depending on the treatment ( Table 2). The concentration of 326 rhodamine B in the droplets significantly increased during the incubation time in all the treatments 327 (Table 1B, Fig 4A). In particular, the increase in rhodamine B concentration in the R60 treatment was 328 more pronounced than in either the V125 or V250 treatments (Table 1C, Fig 4A). At each time point, 329 the treatment significantly affected the rhodamine B concentration. Compared to the R60 treatment, the 330 addition of verapamil in the V125 and V250 treatments reduced the rhodamine B concentration of the secreted droplets after 30, 60, and 90 minutes (Table 2D, Fig 4A). However, there was no significant 332 difference in the rhodamine B concentration between the V125 and V250 treatments at any time point 333 (Table 2D, Fig 4A). 334 Table 2 (Table 3A,E, Fig 5A-C). The net extrusion increased significantly between 30 and 60 minutes 359 in all the treatments (Table 3B, Fig 4B). In contrast, between 60 and 90 minutes the net extrusion 360 increased only for the tubules incubated in R60 and V250, whereas it remained steady for V125 361 treatments (Table 3B, Fig 4B). In particular, the net extrusion was more pronounced in the R60 362 treatment compared to V125 and V250 (Table 3C, Fig 4B). At each time point, the treatment affected 363 the net extrusion of rhodamine B (Table 3D, Fig 4B). In comparison to the R60 treatment, the addition 364 of verapamil in the V125 and V250 treatments significantly reduced the net extrusion of rhodamine B 365 after 30, 60 and 90 minutes, however, there was no significant difference between the V125 and V250 366 treatments (Table 3D, Fig 4B).    (Table 3E, Fig 5A-C). In addition, we found that after 30 402 minutes the fluid secretion rate of the tubules incubated in R60 positively influenced the net extrusion 403 of rhodamine B, while there was no significant effect in the V125 and V250 treatments (Table 3F, Fig  404   5D). After 60 and 90 minutes, the fluid secretion rate positively correlated with the net extrusion of 405 rhodamine B in R60 and V250 treatments, but not V125 (Table 3F, Fig 5E,F). Moreover, the secretion 406 rate of the tubules incubated in R60 showed a more pronounced effect on the net extrusion of rhodamine 407 B than that of the tubules incubated in V125 and V250 (Table 3G, Fig 5D-F).  (Table 3A). The 413 addition of verapamil significantly reduced the quantity of rhodamine B extruded but did not block it 414 completely (Table 3A). The rhodamine B concentration of the droplets secreted did not differ between 415 the V125 and V250 treatments (Table 2A). Therefore, we assumed that 125 μM verapamil was 416 sufficient to inhibit all the P-glycoproteins that are verapamil sensitive and that passive diffusion and  (Table 3A, Fig 4B dashed line). In percentage, the P-422 glycoproteins are responsible for the 88%, 77% and 75% (between 0-30, 30-60 and 60-90 minutes 423 respectively) of the total extrusion of rhodamine B. Overall, P-glycoproteins account for the 77% of the 424 total extrusion of rhodamine B, over the 90 minutes of incubation.

427
Our aim was to determine whether P-glycoprotein transporters are involved in the removal of xenobiotic 428 substances by Malpighian tubules from the haemolymph of desert locusts and, if so, how they perform 429 physiologically. To this end, we developed an alternative method to liquid chromatography-mass 430 spectrometry [37], radiolabelled alkaloids [9] or confocal microscopy [11,12] based upon measuring 431 dye concentration. A similar method has been used previously to investigate epithelial transport in 432 tardigrades and desert locusts using chlorophenol red by imaging through the gut or tubules [38]. By 433 imaging extruded drops from Malpighian tubules, we assessed the performance of epithelial 434 transporters more accurately than by imaging the lumen of the tubules. We used the P-glycoprotein 435 substrate rhodamine B [25] and the P-glycoprotein inhibitor verapamil [39] to assay P-glycoprotein 436 function through the colour of the droplets secreted by the Malpighian tubules. Using this strategy, we 437 obtained evidence that desert locust Malpighian tubules express a P-glycoprotein transporter, that the 438 fluid secretion rate of these tubules is proportional to their surface area, and that the fluid secretion rate 439 influences the net extrusion of rhodamine B. Our conclusion that desert locust Malpighian tubules express P-glycoproteins is supported by two lines 444 of evidence. Firstly, these tubules actively extrude the dye rhodamine B, a P-glycoprotein substrate 445 (e.g. [25,26]), when it is present in the solution in which they are incubated. Rhodamine B has been 446 widely used as a substrate for P-glycoproteins in cell culture and blood brain barrier models (e.g. 447 [25,26]). Secondly, the addition of verapamil, a P-glycoprotein inhibitor (e.g. Verapamil inhibits P-glycoproteins and does not interact with other multidrug resistance 472 proteins [44]. This suggests that the effects of verapamil in our experiments are through its specific 473 effects upon P-glycoproteins. The mechanistic basis of verapamil inhibition is unclear but the most 474 widely accepted explanation is that P-glycoproteins extrude both verapamil and their substrate but that 475 verapamil diffuses back across the lipid bilayer much faster than the substrate creating a futile cycle 476 and thereby competing with the substrate transport [45,46]. 477 are involved in the regulation of the fluid secretion [48]. Indeed, an increase of the intracellular Ca 2+ 480 level mediates the effect of diuretic hormones [49]. Verapamil reduced the fluid secretion of Drosophila 481 Malpighian tubules stimulated by peptide agonists (e.g. CAP2b, cGMP) but had no effect on 482 when studying the effect of a substance on active transporters, it is important to take into account the 548 fluid secretion rate because a change in the net extrusion rate of a substrate may be caused not only by 549 a direct effect on the transporters, but also by an indirect consequence of a change in the fluid flow [12]. 550 In our experiment, the fluid secretion rate did not differ between treatments, indicating that the reduction 551 of net extrusion of rhodamine B following exposure to verapamil was caused solely by inhibition of the 552 P-glycoprotein. However, the fluid secretion rate decreased over time, which may produce an 553 underestimation of the net transepithelial transport of rhodamine B. 554 555

Implications for desert locust detoxification
Gregarious desert locusts feed on a broad variety of plants including species containing secondary 557 metabolites such as atropine to become unpalatable to predators [17][18][19][20]. The expression of P-glycoproteins in the Malpighian tubules to extrude noxious substances may be an adaptation to cope 559 with the ingestion of toxic plants. This may also be the reason for expression of P-glycoproteins on the 560 blood brain barrier of desert locusts, which would prevent the uptake of hydrophobic substances in the 561 central nervous system [23]. Yet the relationship between ingesting toxins and detoxification pathways 562 in the Malpighian tubules is not straightforward; some species of Orthoptera, as well as Coleoptera, 563 Lepidoptera, Heteroptera, Hymenoptera and Sternorrhyncha [56], sequester toxins from the plants they 564 ingest to deter predators. However, toxicity may also be conferred by gut contents, rather than through 565 sequestration within bodily tissues. For example, the chemical defence of the spotted bird grasshopper, 566 Schistocerca emarginata (=lineata), is mediated by the contents of toxic plant in its gut [21,22]. This 567 species is a congener of the desert locust, S. gregaria, which suggests that a similar strategy may be 568 involved in the production of toxicity in this species. If this is the case, then the presence of toxins in 569 the haemolymph may be a consequence of ingesting toxic plant material for storage within the gut. In 570 such a scenario, detoxification pathways within the Malpighian tubules would then be essential for 571 ensuring that toxins do not accumulate within the haemolymph to concentrations that would affect 572 physiological processes. 573 The P-glycoprotein detoxification pathway that we have characterised in desert locusts is likely 574 to be highly effective in extruding xenobiotic compounds from haemolymph, especially when the 575 number of Malpighian tubules within an individual locust is considered. However, it is important to 576 consider that the locusts used for our experiments have experienced a diet free of toxins, such as 577 atropine. P-glycoprotein expression can be modulated depending on the diet [57]; Drosophila larvae 578 fed on colchicine increased the expression of the P-glycoprotein gene homologue mdr49 in the brain 579 and gut. Consequently, adult gregarious desert locusts that have fed on a diet including plant toxins may 580 have even stronger P-glycoprotein detoxification pathways. In contrast to their gregarious counterparts,