Modulation of sleep by trafficking of lipids through the Drosophila blood-brain barrier

  1. Fu Li
  2. Gregory Artiushin
  3. Amita Sehgal  Is a corresponding author
  1. Howard Hughes Medical Institute and Chronobiology and Sleep Institute, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Perelman School of Medicine, University of Pennsylvania, United States

Abstract

Endocytosis through Drosophila glia is a significant determinant of sleep amount and occurs preferentially during sleep in glia of the blood-brain barrier (BBB). To identify metabolites whose trafficking is mediated by sleep-dependent endocytosis, we conducted metabolomic analysis of flies that have increased sleep due to a block in glial endocytosis. We report that acylcarnitines, fatty acids conjugated to carnitine to promote their transport, accumulate in heads of these animals. In parallel, to identify transporters and receptors whose loss contributes to the sleep phenotype caused by blocked endocytosis, we screened genes enriched in barrier glia for effects on sleep. We find that knockdown of lipid transporters LRP1&2 or of carnitine transporters ORCT1&2 increases sleep. In support of the idea that the block in endocytosis affects trafficking through specific transporters, knockdown of LRP or ORCT transporters also increases acylcarnitines in heads. We propose that lipid species, such as acylcarnitines, are trafficked through the BBB via sleep-dependent endocytosis, and their accumulation reflects an increased need for sleep.

Editor's evaluation

Through a convincing series of in vivo experiments, this work demonstrates that lipid- and carnitine transporters in glial cells of the Drosophila blood-brain barrier play important roles in the regulation of sleep. The data are consistent with metabolite clearance, in particular, the sleep-dependent transport of acyl-carnitine across the BBB, being an important function of sleep.

https://doi.org/10.7554/eLife.86336.sa0

Introduction

Although the regulation of sleep is normally studied on a behavioral and circuit level, there is increasing evidence for a role of basic cellular physiology. For instance, we found that disruption of endocytic and trafficking pathways in glia increases sleep in Drosophila (Artiushin et al., 2018). Glia of the Drosophila blood/hemolymph-brain barrier (BBB) emerged as a new cellular locus of sleep regulation in this study, such that genetically manipulating endocytosis in these cells alone was sufficient to increase sleep. As the increased sleep appeared to reflect higher sleep need, we asked if sleep enhances endocytosis through the BBB and found that this was indeed the case. However, the nature of the molecules trafficked through the BBB in a sleep-dependent manner was not known.

Barriers that separate the solutes of blood/hemolymph of the periphery from the interstitial fluid of the central nervous system display a rich profile of transporters, receptors, and trafficking proteins, which often reflect their unique functions. The Drosophila barrier glia populations share many conserved features with vertebrate barriers that employ endothelial and astrocytic populations (DeSalvo et al., 2014, Weiler et al., 2017). For instance, both are capable of moving lipids and carbohydrates, ions, amino acids, and xenobiotics (Weiler et al., 2017). Furthermore, the fly barrier populations may serve specialized roles in metabolism, as not only the conduit of energy sources from the periphery, but also by containing the enzymatic machinery necessary for processing energy sources (Volkenhoff et al., 2015), and secreting signals in reference to nutritional state (Chell and Brand, 2010; Spéder and Brand, 2014).

Given that much of the traffic through the barrier involves energetic substrates, we conducted metabolomic profiling to identify candidate metabolites whose trafficking may be inhibited by an endocytosis block in glia. As genetic manipulations to block endocytosis can directly impact endocytosis-dependent carrier traffic as well as indirectly affect levels of membrane-associated transporters/receptors by altered recycling, we asked if specific glial proteins mediate effects of trafficking on sleep. Thus, we complemented the metabolomic approach with a knockdown screen to broadly search for barrier genes involved in sleep regulation. We report here that specific lipid and carnitine transporters act in barrier glia to affect sleep, and that disrupting expression of these transporters or of endocytosis leads to an accumulation of acylcarnitines in the head.

Results

Blocking endocytosis in glia induces acylcarnitine accumulation in fly heads

Our previous work showed that expression of shibire (shi), a dominant negative dynamin that blocks endocytosis, in all or BBB glia increases sleep. To attain an unbiased, global assessment of metabolites that may be relevant to the increased sleep seen in Repo-GAL4>20xShi.ts1 flies (hereafter referred to as Repo>Shi1), we conducted LC-MS analysis. Heads of male and female Repo>Shi1 flies as well as GAL4 and UAS controls were collected on dry ice with sieves and immediately frozen at –80°C. Each sample contained 200 fly heads (equal male and female), with five samples per genetic condition.

As an initial analysis, raw signal was scaled per each metabolite in reference to other samples within the dataset, and comparisons were made between Repo >Shi flies and each control, as well as controls to each other by Welch’s t-test (Supplementary file 1).

Metabolites of interest were those for which signal from the experimental samples was significantly different, in the same direction, when compared to both controls, while controls compared to each were not significant. Of secondary interest were metabolites where a difference was seen in controls, but was proportionally smaller than consistent differences of each control to the experimental samples.

In surveying this dataset, the outstanding functional category, which contained multiple metabolites whose signal was consistently different in experimental animals versus controls, were the acylcarnitines (Table 1). Furthermore, the fold changes for given metabolites in this group, which consists of fatty acids conjugated to carnitine, were the highest overall. Carnitinylation occurs on fatty acids of various chain lengths, but only a subset of chain lengths in this dataset had sufficient signal, therefore we statistically compared Repo>Shi1 flies to each of the parental controls for metabolites that had signal in at least three of five biological replicates for each genotype. Expression of Shi1 in glia increased abundance in fly heads of the following acylcarnitine species: C2, C16, C16:1, C17, C18:1, C18:2 (Figure 1A). The only metabolite of this group with less signal in experimental animals was the longer chain, C24 (Figure 1B). Carnitine and deoxycarnitine were not significantly altered as compared to both controls.

Table 1
Fatty acid acylcarnitine accumulation in Repo>Shi1 fly heads.

Acylcarnitine species in Repo>Shi1 relative to parental controls. Welch’s t-test was performed on the scaled signal for each metabolite, comparing the conditions shown.

Sub pathwayBiochemical nameGal4>UASGal4CtrlGal4>UASUASCtrlUASCtrlGal4Ctrl
Fatty acid metabolism
(acylcarnitine)
Acetylcarnitine (C2)4.252.681.59
Myristoylcarnitine (C14)28.8126.391.09
Palmitoylcarnitine (C16)4.521.742.60
Palmitoleoycarnitine (C16:1)26.5010.352.56
Margaroylcarnitine (C17)4.051.472.76
Stearoylcarnitine (C18)1.431.061.34
Linoleoylcarnitine (C18:2)6.773.142.15
Oleoylcarnitine (C18:1)4.001.922.08
Arachidoylcarnitine (C20)1.281.171.09
Behenoylcarnitine (C22)1.311.091.20
Eicosenoylcarnitine (C20:1)2.281.241.84
Lignoceroylcarnitine (C24)0.720.830.87
Margaroylcarnitine (C17)4.051.472.76
Nervonoylcarnitine (C24:1)2.361.261.87
Cerotoylcartinine (C26)0.850.970.87
Ximenoylcarnitine (C26:1)0.750.880.85
Figure 1 with 1 supplement see all
Acylcarnitine levels are increased in Repo>Shi1 fly heads.

(A) Short and long chain length or (B) very long chain length acylcarnitines from Repo>Shi1 fly heads and parental controls. The raw signal from LC/MS is plotted, n=3–5 samples, of 200 fly heads each. One-way ANOVA, with Holk-Sidak post hoc comparisons. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Error bars represent standard error of the mean (SEM).

Figure 1—source data 1

Acylcarnitine measurements in Repo>Shi1 and control fly heads.

https://cdn.elifesciences.org/articles/86336/elife-86336-fig1-data1-v3.xlsx

To determine whether the blocking of endocytosis in glia affects acylcarnitine species specifically in heads or also in bodies, we measured the acylcarnitine abundance in bodies of Repo>Shi1 and control flies. Bodies of male and female Repo>Shi1 flies and parental controls were collected on dry ice with sieves and immediately frozen at –80°C. Each sample contained 50 fly bodies (male and female in equal numbers), with a total of five samples per genetic condition. Expression of Shi1 in all glia increased abundance in fly bodies of the following acylcarnitine species: C4 butyryl, C4 isobutyryl, C4-OH butyryl, C4-OH isobutyryl, C12-OH, C14:1-OH, C14-OH,16:1-OH, C16-OH (Figure 1—figure supplement 1). These acylcarnitine species are different from those enriched in the head (Figure 1A and B); in particular, hydroxylated acylcarnitines are enriched in the body with a block in glial endocytosis. These data indicate that endocytosis block, which increases sleep need, also affects lipids in the body, but the nature of the lipids affected is different.

Identification of barrier glia genes that affect sleep

In addition to identifying metabolites that accumulate as a result of blocked glial endocytosis, we sought to identify glial molecules whose function might be impacted by the block in endocytosis and thereby contribute to the effect on sleep. As noted above, glia of the BBB are the most relevant glial subtype for sleep-dependent endocytosis. We identified genes enriched in barrier glia by referring to transcriptional profiling that compared expression in the two glial populations that comprise the Drosophila BBB—subperineurial and perineurial glia (SPG+PG)— to all neurons, and all glia (DeSalvo et al., 2014). Preference was given to previously studied genes, particularly transporters, receptors, and those involved in trafficking, although many genes among the top 50 highly expressed in the barrier glia populations were also tested for effects on sleep. Of the genes enriched in barrier glia, we focused on those that showed low variability in expression from sample to sample. UAS-RNAi constructs for candidate genes were expressed with Repo-GAL4 or RepoGS drivers. Sleep in Repo-GAL4 lines was compared with that in GAL4 and UAS alone controls, while sleep changes with RepoGS were determined by comparing flies maintained on RU486 with control flies tested without RU486. Knockdown of most genes did not produce a significant phenotype, but sleep was increased with knockdown of some transporter genes CG3036, CG6126, mnd, VMAT, CG6836, Rh50, CG4462 (Figure 2—figure supplement 1A), cytoskeleton/trafficking factors CG8036, Vha16, nuf (Figure 2—figure supplement 1B), as well as lsd-2 (Figure 2—figure supplement 1A), acon, and MtnA (Figure 2—figure supplement 1D). Meanwhile, knockdown of the transporter gene CG16700 (Figure 2—figure supplement 1A), cytochrome P450 gene Cyp6a20 (Figure 2—figure supplement 1D), and trafficking factor Cln7 decreased total sleep (Figure 2—figure supplement 1B).

Since the candidate genes were selected based on enrichment within the barrier glia, we chose to examine and secondarily validate promising phenotypes through knockdown with more limited, barrier glia drivers. Thus, we screened a subset of the genes suggested by results of the pan-glial screen (Figure 2—figure supplement 1) with drivers NP6293-GAL4 (Awasaki et al., 2008) and Moody-GAL4 (Strauss et al., 2015) that target PG and SPG, respectively. Knockdown of MtnA (105011 KK), CG6386 (108502 KK), CG4462 (105566 KK), or lsd-2 (102269 KK) did not significantly alter sleep when expressed in either of the barrier glial populations (data not shown). Reduction of cyp6a20 in the PG population inconsistently reproduced the pan-glial sleep loss phenotype, so this was not pursued further (data not shown). The VMAT gene produces two isoforms, one of which is thought to be specific to glia (Romero-Calderón et al., 2008). Knockdown of VMAT (TRiP HMC02346) in the PG increased total sleep (Figure 2A), with sleep increases evident in both day- and nighttime (Figure 2C and G). However, locomotor activity during wake (activity index) was not different from that of controls (Figure 2F). Knockdown of VMAT in PG also did not significantly alter the average daytime or nighttime sleep bout length or bout number, although there was a trend toward increased bout length during the day (Figure 2D, E, H, I). At night, the increase seemed to result primarily from earlier onset of sleep after lights off.

Figure 2 with 1 supplement see all
Change in sleep with knockdown of specific barrier-enriched genes in barrier glia.

Total sleep (A and B) and sleep architecture (C–I) in female flies with knockdown of VMAT, UAS-HMC02346 driven by (perineurial glia [PG]) NP6293-GAL4. Mean sleep, sleep bout length and bout number are shown for daytime sleep (C–E) and nighttime sleep (G–I). Activity index, which depicts activity per waking minute, was not different from both controls (F). n=45–48 per genotype. One-way ANOVA, with Holk-Sidak post hoc comparisons. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Error bars represent standard error of the mean (SEM).

Figure 2—source data 1

Sleep data of flies with VMAT knockdown in perineurial glia.

https://cdn.elifesciences.org/articles/86336/elife-86336-fig2-data1-v3.xlsx

In the SPG, VMAT knockdown had no effect (data not shown), which is consistent with protein expression, as the VMAT-B antibody specifically marks the PG (DeSalvo et al., 2014).

Simultaneous knockdown of Lrp1 and Megalin in barrier glia increases sleep

Although the candidate screen identified barrier genes whose knockdown increases sleep, as does blocking endocytosis in barrier glia, these genes were not obviously linked to the metabolite profile seen with a block in glial endocytosis. The results of metabolomic screening showed changes in lipid, and particularly acylcarnitine, trafficking. Therefore, we reassessed our screen candidates to consider transporters and receptors which may function in these pathways and could have been missed due to redundancy/lethality. Lrp1 and Megalin (Lrp2) are two LDL receptor-related protein members involved in the transport of lipid carrier proteins at the fly barrier (Brankatschk et al., 2014). Expression is likewise found in mammals at the endothelial barrier (Herz, 2003).

Knocking down Lrp1 and Megalin (Lrp2) individually in the pan-glial screen did not significantly alter total sleep time (Figure 2—figure supplement 1A). However, both Lrp1 and Megalin (Lrp2) have been considered to be complementary (Eraly and Nigam, 2002; Brankatschk et al., 2014), therefore it is possible that inhibition of a single gene is insufficient to appreciably affect transport. Simultaneously knocking down Lrp1 and Megalin in all glia with Repo-GAL4 driver increased total sleep time, with increases in both day- and nighttime (Figure 3A and Figure 3—figure supplement 1A and B). Increases in sleep at night resulted from longer bout lengths, with a decrease in bout numbers (Figure 3—figure supplement 1B). Locomotor activity during wake (activity index) did not differ significantly from that of controls, indicating that sleep increase was not due to lethargy/sickness (Figure 3—figure supplement 1C). When both Lrp genes were knocked down with barrier glia drivers, a significant increase in total sleep was seen with the PG driver (NP6923) (Figure 3B), but not with either of the SPG drivers moody or Rab9 (Strauss et al., 2015; Artiushin et al., 2018; Figure 3C and D). As with pan-glial knockdown, night sleep and bout length were increased and night sleep bout number was decreased when Lrp was knocked down with the NP6293-GAL4 driver (Figure 3—figure supplement 1B and E). Meanwhile, the daytime sleep, bout length and bout number did not change (Figure 3—figure supplement 1D). Also, activity index of flies with Lrp knocked down in PG was not significantly different from that of controls (Figure 3—figure supplement 1F).

Figure 3 with 8 supplements see all
Sleep amount changes with knockdown of Lrp1 and Megalin genes in all glia or barrier glia.

Total sleep in female flies with knockdown of (A) Lrp1 (8397 GD) and Megalin (105071 KK) RNAi driven by Repo-GAL4. n=45–48 per genotype; (B) Lrp1 (8397 GD) and Megalin (105071 KK) RNAi driven by NP6293-GAL4 (perineurial glia [PG]). n=45–48 per genotype; (C) Lrp1 (8397 GD) and Megalin (105071 KK) RNAi driven by moody-GAL4 (subperineurial glia [SPG]) n=45–48 per genotype; (D) Lrp1 (8397 GD) and Megalin (105071 KK) RNAi expressed by Rab9-GAL4 (SPG), n=45–48 per genotype; (E) Lrp1 (8397 GD) and Megalin (105071 KK) RNAi driven by 9–137-GAL4 (PG and SPG) with tub-Gal80ts at the permissive temperature of 18°C, n=45–48 per genotype (F) Lrp1 (8397 GD) and Megalin (105071 KK) RNAi driven by 9–137-GAL4 (SPG and PG) with tub-Gal80ts at the restrictive temperature of 31°C, n=45–48 per genotype. (G) Lrp1 (8397 GD) and Megalin (36389 GD) RNAi driven by 9–137-GAL4 (PG and SPG) with tub-Gal80ts at the permissive temperature of 18°C, n=45–48 per genotype (H) Lrp1 (8397 GD) and Megalin (36389 GD) RNAi driven by 9–137-GAL4 (SPG and PG) with tub-Gal80ts at the restrictive temperature of 31°C, n=45–48 per genotype. The same flies were assayed at permissive and restrictive temperatures. Comparisons were with one-way ANOVA, with Holk-Sidak post hoc. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Error bars represent standard error of the mean (SEM).

Figure 3—source data 1

The sleep data of flies with simultaneous knockdown of Lrp1 and Megalin in all glia, perineurial glia , subperineurial glia or in perineurial glia plus subperineurial glia.

https://cdn.elifesciences.org/articles/86336/elife-86336-fig3-data1-v3.xlsx

To target Lrp1 and Megalin in barrier glia at the adult stage and thereby avoid developmental confounds, we coupled drivers specific to these glia with the temperature-sensitive tubulin-Gal80 (tub-Gal80ts) system that suppresses GAL4 expression at 18°C but allows it at 31°C. In these experiments, to verify their role in sleep, we sought to use multiple RNAi lines for these transporters and were able to obtain three lines for Megalin (105071 KK, 36389 GD, and 27242 GD); however, for Lrp1 only one line (8397 GD) had a chromosomal location that facilitated generation of double knockdowns. We combined each of the Megalin RNAi transgenes with the Lrp RNAi transgene and crossed them with the 9–137-GAL4 driver, which expresses in both PG and SPG (DeSalvo et al., 2014), and with tub-Gal80ts. Knockdown of Lrp1 (8397 GD) together with knockdown of Megalin, using either 105071 KK or Megalin 36389 KK, increased total sleep significantly in a temperature-dependent fashion, indicating an adult-specific role of these transporters in barrier glia (Figure 3E, F, G and H and Figure 3—figure supplement 3D). Surprisingly, in these flies, increases were during the day, and in one case (105071 KK) were accompanied by increased bout length and decreased bout number (Figure 3—figure supplement 2D). The activity index was not significantly different from that of controls (Figure 3—figure supplements 2F; 3F). However, Lrp1 (8397 GD) with Megalin 27242 GD did not alter sleep (Figure 3—figure supplement 4), likely due to insufficient knockdown (see below). As expected, most single knockdowns of Lrp1 or Megalin did not alter sleep amount (Figure 3—figure supplement 5A, B and E–J; Figure 3—figure supplement 6H). However, Lrp1 RNAi construct GD 109605 (Figure 3—figure supplement 5C, D) and Megalin RNAi construct GD 105387 (Figure 3—figure supplement 6A, B) individually increased total sleep when expressed by the driver 9–137-GAL4. The GD 109605 construct was not effective at knocking down Lrp1, suggesting that sleep phenotypes are due to off-target effects (Figure 3—figure supplement 7). And the GD 105387 line affected sleep at the permissive temperature of 18°C, when the GAL4 is not expected to be active, as well as at the restrictive temperature of 31°C, raising concerns about specificity.

Speculating that lack of a sleep phenotype with Megalin 27242 GD could be due to insufficient downregulation of Megalin gene expression, we sought to confirm the efficiency of knockdown of Lrp1 and Megalin genes in the fly lines. As qPCR analysis is typically done with head RNA, and these genes are widely expressed, we tested the efficacy of the RNAi by expressing it with the ubiquitously expressed Act-GAL4 driver. Lrp1 RNAi fly lines 8397 GD, 13913 GD, and Megalin 36389 GD, 105387 KK, 105071 KK showed effective downregulation of their target genes (Figure 3—figure supplement 7). The Megalin gene in RNAi fly line 27242 GD was only slightly downregulated (Figure 3—figure supplement 7B), which is consistent with its lack of an effect on sleep (Figure 3—figure supplement 4).

To confirm that sleep alterations were specifically caused by the knockdown of Lrp1 and Megalin in barrier glia, we asked if knockdown of Lrp1 and Megalin in non-barrier glia had any effect on sleep. We used MZ0709-GAL4, NP2222-GAL4, and Eaat-GAL4 drivers to express Lrp1 and Megalin RNAi in ensheathing glia, cortex glia, and astrocyte-like glia, respectively, coupled with the tub-Gal80ts system to restrict knockdown to adults. None of these manipulations affected sleep (Figure 3—figure supplement 8), showing that Lrp1 and Megalin act only in barrier glia to regulate sleep.

Knockdown of Orct and Orct2 in barrier glia increases sleep

The organic cation (Orct) transporters are multi-substrate transporters whose substrates include carnitine (Lahjouji et al., 2001), and perhaps also carnitylated molecules, based on in vitro evidence for Orct2 (Kou et al., 2017). However, like Lrp1 and Megalin (Lrp2), Orct and Orct2 are considered to be complementary (Brankatschk et al., 2014). As with the double knockdown of Lrp1 and Megalin, knockdown of both Orct genes increased total sleep (Figure 4A). Simultaneous knockdown of Orct2 and Orct in all glia with Repo-GAL4 caused daytime and nighttime increase, the latter driven by increases in nighttime sleep bout length (Figure 4—figure supplement 1A, B). Locomotor activity during wake (activity index) did not change (Figure 4—figure supplement 1C). Testing their function in barrier glia showed that knockdown of Orct1 and -2 in either PG or SPG replicated the increased pan-glial sleep phenotype (Figure 4B–D), although this was only true with one line for the SPG (Figure 4D). Night-specific increases were noted with knockdown of Orct2 106681 KK and Orct 6782 GD in PG (Figure 4—figure supplement 1E).

Figure 4 with 6 supplements see all
Sleep amount changes with knockdown of Orct genes in all glia or barrier glia.

Total sleep in female flies with knockdown of (A) Orct (6782 GD) and Orct2 (106681 KK) RNAi driven by Repo-GAL4. n=45–48 per genotype; (B) Orct (6782 GD) and Orct2 (106681 KK) RNAi driven by NP6293-GAL4 (perineurial glia [PG]). n=45–48 per genotype; (C) Orct (6782 GD) and Orct2 (106681 KK) RNAi driven by moody-GAL4 (subperineurial glia [SPG]) n=45–48 per genotype; (D) Orct (6782 GD) and Orct2 (106681 KK) RNAi expressed by Rab9-GAL4 (SPG), n=45–48 per genotype; (E) Orct (6782 GD) and Orct2 (106681 KK) RNAi driven by 9–137-GAL4 (PG and SPG) with tub-Gal80ts at the permissive temperature of 18°C, n=45–48 per genotype; (F) Orct (6782 GD) and Orct2 (106681 KK) RNAi driven by 9–137-GAL4 (SPG and PG) with tub-Gal80ts at the restrictive temperature of 31°C, n=45–48 per genotype; (G) Orct (47133 GD) and Orct2 (106681 KK) RNAi driven by 9–137-GAL4 (PG and SPG) with tub-Gal80ts at the permissive temperature of 18°C, n=45–48 per genotype; (H) Orct (47133 GD) and Orct2 (106681 KK) RNAi driven by 9–137-GAL4 (SPG and PG) with tub-Gal80ts at the restrictive temperature of 31°C, n=45–48 per genotype. Comparisons were with one-way ANOVA, followed by Holk-Sidak post hoc. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Error bars represent standard error of the mean (SEM).

Figure 4—source data 1

The sleep data of fiies that had simultaneous knock down of Orct and Orct2 in all glia or barrier glia.

https://cdn.elifesciences.org/articles/86336/elife-86336-fig4-data1-v3.xlsx

Coupling each of two different Orct RNAis (6782 GD and 47133 GD) with Orct2 106681 KK RNAi and expressing them in adults using the 9–137-GAL4 driver with tub-Gal80ts significantly increased sleep during the day and night (Figure 4E, F, G and H; Figure 4—figure supplement 2 and Figure 4—figure supplement 3D, E). Day sleep bout length was increased in both pairs of knockdowns and night bout length also increased in Orct2 106681 KK coupled with Orct 47133 GD, accompanied by a decrease in sleep bout number (Figure 4—figure supplements 2 and 3E). However, Orct2 106681 KK coupled with Orct 6782 GD did not show changes in sleep bout length at night (Figure 4—figure supplement 2). As with other knockdown lines, activity index was not different from that of controls (Figure 4—figure supplements 2 and 3). Single knockdowns of Orct and Orct2 genes did not change sleep amount (Figure 4—figure supplement 4A, B, E, F, G, H), except for the Orct RNAi 52658 GD construct that increased sleep (Figure 4—figure supplement 4C, D) without changing expression of Ocrt (Figure 4—figure supplement 6), thereby rendering this result unreliable.

Checking the knockdown efficiency of Orct and Orct2 RNAi constructs with the Act-GAL4 driver revealed that Orct RNAi line 6782 GD and Orct2 RNAi line 106681 KK significantly decreased target gene expression (Figure 4—figure supplement 6A B). Although simultaneous knockdown with Orct 47133 GD and Orct2 106681 KK increased sleep, we found that the 47133 GD line only slightly reduced Orct gene expression (Figure 4—figure supplement 6A).

We also used non-barrier-glia drivers MZ0709-GAL4, NP2222-GAL4, and Eaat-GAL4 to express Orct and Orct2 RNAi in ensheathing glia, cortex glia, and astrocyte-like glia, respectively. Simultaneously knocking down Orct and Orct2 in ensheathing glia, cortex glia, and astrocyte-like glia did not change sleep time (Figure 4—figure supplement 5), demonstrating that only knockdown of Orct and Orct2 in barrier glia increases sleep.

Knockdown of Lrp and Orct genes in glia leads to accumulation of acylcarnitines

Given that knockdown of the Lrp and Orct genes in glia parallels effects of Shi expression in terms of increasing total sleep, we asked if it had the same effect on metabolite accumulation in fly heads. Based on the metabolomic results of Repo >Shi1, we chose to specifically assay acylcarnitines through LC-MS analysis. As in the case of the Shi1 experiment, we collected heads from flies in which either Lrp1 and Megalin or Orct 1 and Orct2 were knocked down with Repo-GAL4.

Knockdown of Lrp1 and Megalin in all glia increased abundance in fly heads of the following acylcarnitine species: C3, C4 butyryl, C4-OH isobutyryl, C4 isobutyryl, C16, C18, C20 (Figure 5). Of these, C16 and a different form of C18 were also detected in Repo>Shi1 flies. Longer-chain acylcarnitines, for example, those over C22, were largely undetected in the Lrp experimental samples, perhaps due to differences in methodological sensitivity (they were processed in a different facility from Repo>Shi1 flies). Knockdown of Orct and Orct2 similarly enriched acylcarnitines in fly heads. In particular, acylcarnitine species C16, C16:1, and C20 were increased significantly (Figure 6), with C16 and C16:1 matching results of Repo>Shi1 and C16 also overlapping with Repo>Lrp. Similar to the case of the Lrp samples, longer-chain acylcarnitines over 22 were not detected well. Overall, while the total sets of species affected are not identical, blocking transporters of lipids or carnitines yields similar increases in acylcarntines to those seen with blocking endocytosis in glia, with certain species being affected in all manipulations.

Acylcarnitine levels are increased in Repo>Lrp1+Mgl RNAi fly heads.

Short, medium, and long chain length acylcarnitines from Repo-GAL4>Lrp1+Mgl RNAi fly heads and parental controls. The raw signal from LC/MS is plotted, n=3 samples, of 300 fly heads each. One-way ANOVA, with Holk-Sidak post hoc comparisons. *p<0.05, **p<0.01. Error bars represent standard error of the mean (SEM).

Figure 5—source data 1

Acylcarnitine measurements in fly heads with simultaneous knockdown of Lrp1 and Megalin in all glia.

https://cdn.elifesciences.org/articles/86336/elife-86336-fig5-data1-v3.xlsx
Acylcarnitine levels are increased in Repo>Orct2+Orct RNAi fly heads.

Short, medium, and long chain length acylcarnitines from Repo-GAL4>Orct + Orct2 RNAi fly heads and parental controls. The raw signal from LC/MS is plotted, n=3 samples, of 300 fly heads each. One-way ANOVA, with Holk-Sidak post hoc comparisons. *p<0.05. Error bars represent standard error of the mean (SEM).

Figure 6—source data 1

Acylcarnitine measurements in fly heads with simultaneous knockdown of Orct and Orct2 in all glia.

https://cdn.elifesciences.org/articles/86336/elife-86336-fig6-data1-v3.xlsx

Acylcarnitine feeding has a small effect on sleep

The accumulation of acylcarnitines in the brains of Repo>Shi1 flies or flies with decreased expression of Lrp1 and Megalin or Orct and Orct2 in barrier glia raised the possibility that acylcarnitines contribute to the sleep phenotypes of these flies; in other words that they increase sleep or sleep need. To investigate whether supplementing fly food with acylcarnitines could alter sleep, we fed flies each of three mixes of acylcarnitines of varying chain lengths, Mix 1 (A-141, Sigma), 2 (A-144, Sigma), and 3 (A-145, Sigma). Acylcarnitine Mix 1 contains 10 mg/mL of C0 and C2 in methanol, of which C2 was increased in heads of Repo>Shi1 flies and of short-sleeping mutants fmn, rye, and sssP1 (Bedont et al., 2023). Acylcarnitines in Mix 2 (C4, iC4, C5, iC5, C6, C8, C10, C12, C14, C18) and Mix 3 (C8:1, C10:1, C12:1, C14:1, C14:2, C16:1, C18:1, C18:2) are at 100–200 μg/mL concentration in 90:10 methanol: 1% aqueous formic acid. Mix 2 contains iC4 and C18 that were found increased with the Lrps pan-glial knockdown and in fmn, rye, and sssP1, while Mix 3 has the C16:1, C18:1, and C18:2 species found in Repo>Shi1, Repo>Orcts and in sleep short mutants fmn, rye, and sssP1. No commercial products are available for the very long chain acylcarnitines, such as C20, C22, C24, C24:1, and C26:1, that increase with pan-glial knockdown of Lrps or Orcts or in fmn, rye, and sssP1. Each of the acylcarnitine mixes was added to the fly food (5% sugar and 1.8% agar) of experimental iso31 flies and similar concentrations of methanol or methanol and aqueous formic acid were added to the food of controls. Acylcarnitine Mix 1 was tested at five different concentrations of 2.5 mg/L, 5 mg/L, 10 mg/L, 25 mg/L, and 50 mg/L. As the LC-MS analysis indicated that the concentration of C2 is approximately 100–1000 times higher than that of the medium or long chain acylcarnitines, acylcarnitine Mix 2 and 3 were tested at concentrations of 0.025 mg/L, 0.05 mg/L, 0.10 mg/L, 0.25 mg/L, and 0.5 mg/L (Figure 7). All experimental and control flies were monitored for sleep for 3 days after the feeding.

Figure 7 with 5 supplements see all
Effects of acylcarnitine Mix 1, 2, or 3 on sleep.

Total sleep was evaluated in female iso31 flies fed with mixes of short or long chain length acylcarnitines or control food. (A) Acylcarnitine Mix 1 at 2.5 mg/L, 5 mg/L, 10 mg/L, 25 mg/L, and 50 mg/L concentrations. (B) Acylcarnitine Mix 2 at 0.025 mg/L, 0.05 mg/L, 0.10 mg/L, 0.25 mg/L, and 0.5 mg/L concentrations. (C) Acylcarnitine Mix 3 at 0.025 mg/L, 0.05 mg/L, 0.10 mg/L, 0.25 mg/L, and 0.5 mg/L concentrations. N=16 per condition. Mann-Whitney T test was performed. *p<0.05, **p<0.01. Error bars represent standard error of the mean (SEM).

Figure 7—source data 1

Sleep data of flies fed food containing acylcarnitines Mix 1, 2 or 3 or control solvent.

https://cdn.elifesciences.org/articles/86336/elife-86336-fig7-data1-v3.xlsx

Despite a slight increase in sleep at the 2.5 mg/L concentration, significant changes in sleep were not seen with any concentration of Mix 1 (Figure 7A, Figure 7—figure supplement 1). Low concentrations of acylcarnitine Mix 2 0.025 mg/L and 0.05 mg/L increased daytime sleep amount slightly (Figure 7B, Figure 7—figure supplement 2A, E) and significantly increased nighttime sleep (Figure 7—figure supplement 2B, F), in case of 0.05 mg/L with a decrease in nighttime sleep bout number, suggesting that sleep is more consolidated. The 0.1 mg/L concentration of acylcarnitine Mix 2 increased daytime sleep and sleep bout length, again suggesting more consolidation (Figure 7—figure supplement 2). Low concentrations of acylcarnitine Mix 3—0.025 mg/L, 0.05 mg/L, and 0.1 mg/L—also increased daytime sleep with increased daytime sleep bout length (Figure 7—figure supplement 4A, D, E, H, I and L). The 0.1 mg/L concentration of acylcarnitine Mix 3 increased nighttime sleep as well (Figure 7—figure supplement 4J). Activity index decreased in flies fed low concentrations of Mix 2 or Mix 3 alone, raising the possibility that lower mobility contributes to the increased sleep; however, results of experiments where flies were fed Mix 2 and 3 together alleviated this concern (see below). High concentrations of acylcarnitine Mix 2 or 3 had no effect on sleep (Figure 7C).

Since each of the long chain acylcarnitine mixes, Mix 2 or Mix 3, increased sleep, we asked if the two together had more robust effects on fly sleep. Combinations of Mix 2 and Mix 3 were tested at concentrations of 0.0125 mg/L, 0.025 mg/L, 0.05 mg/L, 0.10 mg/L, 0.25 mg/L, and 0.5 mg/L (Figure 8). As with each mix alone, low concentrations of Mix 2 and 3 together increased sleep amounts (Figure 8A and B), but high concentrations had no effect (Figure 8D and F) except for a slight increase with 0.25 mg/L (Figure 8E).

Figure 8 with 2 supplements see all
Effects of acylcarnitine Mix 2 and 3 on sleep.

Total sleep was evaluated in female iso31flies fed with mixes of long chain length acylcarnitines or control food. Each of acylcarnitine Mix 2 and 3 was added at a concentration of 0.0125 mg/L (A), 0.025 mg/L (B), 0.05 mg/L (C), 0.10 mg/L (D), 0.25 mg/L (E), or 0.5 mg/L (F). n=32 per condition. Mann-Whitney T test was performed. *p<0.05, **p<0.01, ****p<0.0001. Error bars represent standard error of the mean (SEM).

Figure 8—source data 1

Sleep data of flies fed with food containing acylcarnitines Mix 2 and 3 or control solvent.

https://cdn.elifesciences.org/articles/86336/elife-86336-fig8-data1-v3.xlsx

The magnitude of the change in sleep was not noticeably different, as compared to the individual mixes, suggesting that sleep-promoting components in the two mixes are redundant rather than additive. As in the case of the feeding of individual mixes, we also examined sleep architecture—daytime and nighttime sleep bout length and number and activity index (activity per waking minute)—in flies fed Mix 2 and 3 together (Figure 8—figure supplements 1 and 2). In general, changes in day or night sleep were accompanied by increased sleep bout length and decreased bout number, indicating better sleep consolidation. Flies fed Mix 2 and 3 together showed either no change or even an increase in activity index, demonstrating an effect on sleep independent of overall reduced mobility.

Discussion

Our previous work suggested that endocytosis through the BBB is a function of sleep, but the nature of the molecules trafficked remained unknown. Using a dual-pronged approach of a targeted genetic screen and unbiased metabolomic analysis, we report here that the passage of lipids through the BBB is important for sleep. Blocking such transport increases sleep in conjunction with an accumulation of acylcarnitines.

Through a pan-glial RNAi knockdown screen of candidate genes expressed in the fly barrier, we identified molecules that affect daily sleep amount. In initial follow-up experiments, we targeted knockdown to each barrier layer separately, which is subject to the concern that behavioral phenotypes requiring simultaneous knockdown in both barrier populations would be missed. Nevertheless, we consider this scenario to be less probable, as permeability through the populations is quite different, with smaller solutes likely passing through the PG but not the tight barrier of the SPG, lipophilic solutes, or xenobiotics potentially passing through each uninhibited, and larger solute requiring endocytic mechanisms that likely have to work in each population in tandem. Therefore, in most cases of knockdown, transport would either be inhibited by the one barrier population essential for it or would be interrupted by either population. This was indeed the case for the VMAT isolated in the initial screen. The lipid and carnitine transporters we subsequently focused on also produced phenotypes through knockdown in a single barrier layer, although we also tested effects of knockdown in both layers. While there was redundancy at the level of the transporters, the two barrier layers did not compensate for each other.

An additional consideration of preliminarily screening with pan-glial drivers is that if knockdown in multiple glial subtypes has opposing effects on sleep, we may have obscured a role for the barrier cells. Again, we attempted to minimize this risk by primarily selecting genes whose expression is both highly abundant and specifically enriched in the barrier populations, as opposed to the transcriptome dataset of all glial cells (DeSalvo et al., 2014). The assumption is that multifold expression in the barrier populations is indicative of prevailing importance in these cells, although this is a caveat.

The vesicular monoamine transporter (VMAT) has previously been identified as a target of reserpine, which promotes sleep in the fly (Nall and Sehgal, 2013). VMAT mutants exhibit higher baseline sleep, and also lose less sleep than controls when subject to mechanical sleep deprivation. VMAT can traffic multiple monoamines such as dopamine, serotonin, histamine, and octopamine, but no single neuronal population or neurotransmitter system was implicated as responsible for the VMAT sleep phenotype (Nall and Sehgal, 2013). In flies, VMAT exists as two isoforms, VMAT-A, which is expressed in monoaminergic neurons, and VMAT-B, which appears to be specific to PG (DeSalvo et al., 2014), as it is also found in fenestrated glia in the visual system (Romero-Calderón et al., 2008), which are a specialized form of PG (Kremer et al., 2017). It is unknown whether VMAT-B would function similarly in glia as VMAT-A does in neurons. VMAT-B contains an additional cytoplasmic domain, which has been suggested to promote retention in the plasma membrane as opposed to trafficking to vesicles (Greer et al., 2005). VMAT knockdown increased sleep, as did disrupting endocytic trafficking at the barrier (Artiushin et al., 2018). Whole-brain levels of monoamines were not altered in flies that expressed Shi1 in glia, nevertheless it is possible that this gross analysis would not be sensitive to local changes at the barrier. In the visual system glia, VMAT-B may be necessary for uptake of histamine (Romero-Calderón et al., 2008). Interestingly, histamine is known to alter permeability of the BBB in mammals (Lu et al., 2010).

Our metabolomic data indicated that acylcarnitines are elevated in the heads of the long-sleeping Repo>Shi1 flies. Acylcarnitines are transported to mitochondria for fatty acid oxidation, but are also secreted as they are found in plasma in mammals (Schooneman et al., 2013). Given that circulating acylcarnitines can be taken up by cells, we investigated Lrp1/Megalin and Orct/Orct2 as candidate transporters for this uptake. Lrp1 and Megalin are lipoprotein carrier receptors known to function in the fly barrier (Brankatschk et al., 2014) and although knockdown of each one separately in all glia did not affect sleep, reducing expression of both in all glia or barrier glia increased sleep. Likewise, knockdown of Orct and Orct2, which are homologs of the human carnitine transporters (Eraly and Nigam, 2002) and transport carnitine as well as acylcarnitines (Pochini et al., 2004), increases sleep. LC-MS analysis shows that knockdown of these transporters enriches acylcarnitines in fly heads just as blocking endocytosis does, supporting the idea that LRP and ORCT are among the proteins affected by the block in endocytosis. Exactly where the accumulation of acylcarnitines occurs is not known at this time, but we suggest that it is largely extracellular. In addition, given that heads contain brains as well as fat body, we acknowledge that the fat body may contribute to changes in acylcarnitines and other metabolites reported here. It is further possible that the differences in lipid species elevated in head as opposed to body samples may reflect the different outputs of the head and abdominal fat bodies, respectively.

Acylcarnitine accumulation in flies with blocked glial endocytosis or lipid transport could occur as a consequence of the high sleep in these animals. We believe this is unlikely as an increase in acylcarnitines appears to generally occur under conditions of sleep deprivation, that is conditions that would promote sleep. Carnitine conjugation of long chain fatty acids was reported in cortical metabolites of sleep-deprived mice, while short and medium chain fatty acids were reduced (Hinard et al., 2012). Changes in acylcarnitines were also noted in the peripheral blood of sleep-deprived or sleep-restricted humans (Davies et al., 2014; Weljie et al., 2015) as well as over a day:night cycle (Ang et al., 2012; Dallmann et al., 2012). We find too that a common feature of short-sleeping fly mutants, which are models for chronic sleep deprivation, is an increase in acylcarnitines, regardless of the mechanism that causes the sleep loss (Bedont et al., 2023). In the metabolomic analysis reported here, samples were collected at ZT 0–3, and so not after a period of sleep deprivation; however, all the manipulations (knockdown of Orct/Lrps or expression of Shi) increased sleep need, and so may be associated with high acylcarnitines at all times. Thus, acylcarnitines are generally associated with sleepiness rather than sleep. While this does not necessarily mean that they promote sleep, we show here that feeding flies low concentrations of medium and long chain acylcarnitines modestly increases sleep, although the fate of these acylcarnitines or where they act to affect sleep is not known. High concentrations have no effect, either due to some feedback mechanism or perhaps because high levels of the methanol solvent inhibit feeding. Regardless of their role in sleep regulation, we suggest that acylcarnitines are a key marker of sleep need across species, and could be exploited for this purpose. Sleep is likely required to reduce levels of acylcarnitines, perhaps as part of a broader role in metabolic homeostasis.

Materials and methods

Fly stocks

Request a detailed protocol

The initial screen was performed with lab stock drivers:;;Repo-GAL4/TM6c, Sb and UAS-Dicer; RepoGeneSwitch. SPG driver Moody-GAL4 and surface driver 9–137-GAL4 were shared by Roland Bainton, while PG driver NP6293-GAL4 was a gift of Marc Freeman.;;UAS-20xShi.ts1 (referred to as UAS-20xShi1) was shared by Gerald Rubin. RNAi lines were ordered from VDRC (KK and GD collection) and Bloomington (TRiP collection) stock centers, with the stock number provided in the below table.

Iso31Laboratory Stocks
w; 9–137-GAL4; tub-Gal80tsLaboratory Stocks
w; MZ0708-GAL4; tub-Gal80tsLaboratory Stocks
w; Eaat1-GAL4; tub-Gal80tsLaboratory Stocks
w; tub-Gal80ts; NP222-GAL4Laboratory Stocks
y[1] w[*]; PBac{y[+mDint2] w[+mC]=I-SceI(FRT.Rab9-GAL4.ATG(loxP.3xP3-RFP))}VK00033/TM3, Sb[1]Bloomington Stock Center51587
w;; Repo-GAL4, 6x crossed to IsoBloomington Stock Center7415
y[1] v[1]; P{y[+t7.7] v[+t1.8]=TRiP.HMC02346}attP2Bloomington Stock Center44471
y[1] sc[*] v[1] sev[21]; P{y[+t7.7] v[+t1.8]=TRiP.HMC03067}attP2Bloomington Stock Center50666
y[1] v[1]; P{y[+t7.7] v[+t1.8]=TRiP.JF01350}attP2Bloomington Stock Center31381
y[1] v[1]; P{y[+t7.7] v[+t1.8]=TRiP.JF01628}attP2Bloomington Stock Center31151
y[1] sc[*] v[1] sev[21]; P{y[+t7.7] v[+t1.8]=TRiP.HMC05119}attP40Bloomington Stock Center60125
y[1] sc[*] v[1] sev[21]; P{y[+t7.7] v[+t1.8]=TRiP.HMC04670}attP40Bloomington Stock Center57583
Orct GDVDRC52658
Orct2 GDVDRC1176
Orct GDVDRC47133
Orct2 KKVDRC106681
Orct GDVDRC6782
Megalin KKVDRC105071
Lrp1 GDVDRC8397
Lrp1 KKVDRC109605
Lrp1 GDVDRC13913
LRP1 GDVDRC3710
Megalin GDVDRC27242
Megalin GDVDRC105387
Megalin GDVDRC36389
CG3036VDRC108500
CG3168VDRC48011
CG4797VDRC10598
CG3897VDRC101083
L(2)80717 GDVDRC11117
CG6126 GDVDRC7326
Lsd-2 KKVDRC102269
CG16700 KKVDRC110058
Mct1 KKVDRC106773
JhI-21 KKVDRC108509
Mnd KKVDRC110217
Lrp1 KKVDRC106364
CG5687 GDVDRC33262
CG6386 KKVDRC108502
CG4462 KKVDRC105566
JhI-21 KKVDRC108509
Mnd KKVDRC110217
Lrp1 KKVDRC106364
CG5687 GDVDRC33262
CG6386 KKVDRC108502
CG4462 KKVDRC105566
CG4462 GDVDRC6786
Npc2b GDVDRC16652
Catsup KKVDRC100095
CG8036 KKVDRC105633
Arc1 KKVDRC109141
Vinc KKVDRC105956
Ctp KKVDRC109084
Vha16 KKVDRC104490
Blos2 KKVDRC107449
Capu KKVDRC110404
Snapin GDVDRC28145
Nuf KKVDRC104172
Moe KKVDRC110654
Cln7 kkVDRC109291
Ggamma30A KKVDRC102706
Nplp3 KKVDRC105584
InR GDVDRC992
Sdr KKVDRC105549
Grip KKVDRC103551
Olf86-m GDVDRC17657
Wgn GDVDRC9152
sesB KKVDRC104576
Acon GDVDRC11767
Fer1HCH KKVDRC102406
Fer2LCH KKVDRC106960
MtnA KKVDRC105011
Tctp GDVDRC45532
Neu3 KKVDRC102041
Ppn KKVDRC108005
Stall KKVDRC104737
CG10433 GDVDRC41162
Arouser KKVDRC105755
Gste7 KKVDRC104977
DnaJ-1 KKVDRC104618
Argk GDVDRC34037
Santa-maria GDVDRC33153
14-4-3 E KKVDRC108129
Cyp6a20 GDVDRC3313
Pld KKVDRC106137
CG10126 KKVDRC44104
CG10184 KKVDRC104488
CG7800 KKVDRC100221
AdamTS-A GDVDRC33347
Hnf-4 GDVDRC12692
CG9743 KKVDRC108185
CG7461 KKVDRC110220

Behavior

Request a detailed protocol

Flies were crossed and raised on standard cornmeal-molasses medium in bottles under 12:12 hr light:dark. Offspring were entrained at 25°C, in LD12:12 conditions until at least 6 days post-eclosion, before age-matched flies, which were group-housed in bottles, were used in sleep assays. Mated females were loaded into glass locomotor tubes with 2% agar 5% sugar. Sleep was quantified by the Drosophila Activity Monitor (DAM) system, using the established definition of a minimum 5 min of inactivity. Data was analyzed in PySolo (Gilestro and Cirelli, 2009).

For the sleep experiments using 9–137-GAL4, MZ0709-GAL4, NP2222-GAL4, or Eaat-GAL4 coupled with tub-Gal80ts, the same group flies were monitored initially at 18°C for 3 days and then shifted to 31°C for 2 days and shifted back to 18°C again.

Metabolomics

Request a detailed protocol

Approximately 7-day-old Repo>Shi1 flies were maintained at 1°C for 2 days to impair endocytosis in all glia. At ZT 0–3, whole flies were quickly frozen in Falcon tubes chilled on dry ice, and placed at –80°C. Each tube contained 50 flies. Heads were subsequently removed from the body by briefly vortexing the tube. Heads were then separated from the rest of the body by an array of copper sieves, whose housing was buried in dry ice to keep the preparation cool. For each sample, 200 fly heads, of equal parts from males and females, were collected in 1.5 mL tubes which were quickly refrozen. Samples were shipped on dry ice to Metabolon, Inc, where they were assessed by LC-MS (Evans et al., 2009). For metabolomic analysis of Lrp and Orct knockdown fly lines, each sample contained 300 fly heads, of equal parts from males and females. Samples were processed by LC-MS at the Penn Metabolomics Core.

For the comparison of acylcarnitine levels in the bodies, 50 bodies of equal numbers from males and females were collected and quickly frozen in Eppendorf tubes chilled on dry ice, and placed at –80°C. Samples were processed by LC-MS at the Penn Metabolomics Core.

qPCR validation of Lrps or Orcts RNAis

Request a detailed protocol

To determine the knockdown efficiency of target transcripts with Lrp and Orct RNAis, the Lrp and Orct RNAi flies were crossed with Act-GAL4 driver to express RNAi ubiquitously. RNA was extracted from experimental and control fly lines by using the SV total RNA extraction kit and its protocol (Z3105, Promega). The RNAs were resuspended in nuclease-free water about 300 ng/μL. cDNA libraries of RNAi fly lines and control iso31 were synthesized by using the High Capacity cDNA Reverse Transcription Kit and its protocol (4368814, Applied Biosystems). cDNAs from Lrps and Orcts RNAi and iso31 were amplified using SYBR Green PCR mix (4364346, Lifetech) on an Applied Biosystems ViiA7 qPCR machine. Primers are listed below. We calculated relative transcript levels by ddCT.

Lrp1 forwardTGCTCCCTGCTCTTCAGTTG
Lrp1 reverseCTCAGTGAAAAATTGCTGCGG
Megalin forwardCGCCATTCACAGCATCACAG
Megalin reverseAGGTTGCTGATGGCCACAAG
Orct forwardACTTCGGGAGTCTTCCTGGTT
Orct reverseGCCGTGAGCATAAAGCCCA
Orct2 forwardGAGTGGAACCTGGTTTGTGG
Orct2 reverseCCGTACTTATCGCTCAGTTGAC

Acylcarnitine mixture feeding

Request a detailed protocol

Three kinds of acylcarnitine mixtures Mix 1, Mix 2, and Mix 3 were ordered from Sigma (A-141, A-144, and A-145, Sigma). These acylcarnitine mixtures, which used methanol or methanol plus formic acid as solvents, were added individually into the fly food (5% sugar and 1.8% agar) and the same concentrations of methanol or methanol and aqueous formic acid were added into control food. Iso31 flies were fed five concentrations of acylcarnitine Mix 1—2.5 mg/L, 5 mg/L, 10 mg/L, 25 mg/L, and 50 mg/L. Mix 2 or Mix 3 was added into food at concentrations of 0.025 mg/L, 0.05 mg/L, 0.10 mg/L, 0.25 mg/L, and 0.5 mg/L, respectively. For feeding of acylcarnitine mixtures of Mix 2 and Mix 3, each of Mix 2 and Mix 3 was added at a concentration of 0.0125 mg/L, 0.025 mg/L, 0.05 mg/L, 0.10 mg/L, 0.25 mg/L, and 0.5 mg/L, as done for separate feeding of Mix 2 or Mix 3. For the sleep assay, ~7-day-old iso31 flies were housed individually in glass tubes containing acylcarnitine or methanol or methanol and aqueous formic acid control food in Percival incubators. As above, sleep was determined based upon activity recorded with the Trikinetics DAM system (http://www.trikinetics.com/). All flies were maintained in a 12 hr:12 hr light:dark cycle and behavior was monitored for 5 days, of which the last 3 were analyzed for sleep. Pysolo (http://www.pysolo.net) (Gilestro and Cirelli, 2009) was used to analyze and plot sleep patterns.

Statistics

For both behavioral and acylcarnitine metabolomics results, the experimental group was compared to two parental controls by one-way ANOVA with Holm-Sidak post hoc tests (in addition to specific tests mentioned in figure legends) with GraphPad Prism software. For the initial comparisons of metabolomic data, Metabolon performed Welch’s t-tests (Table 1 and Supplementary file 1) on scaled signal data for each metabolite, to compare different conditions. Raw signal was scaled so that the median would be equal to 1, using all samples that had been concurrently run. Missing values were filled in with the lowest value of run samples for that metabolite. Additional details of statistics tests are listed in the figure legends. For the qPCR results, the experimental group was compared to control iso31 by Mann-Whitney T test.

Data availability

All data generated or analysed during this study are included in the manuscript and supporting file. Source data files have been provided for Figures 1, 2, 3, 4, 5, 6, 7, 8 and Figure 1—figure supplement 1, Figure 2—figure supplement 1, Figure 3—figure supplement 1, 3 ,4, 5 ,6, 7, 8. Figure 4—figure supplement 1, 2, 3, 4, 5, 6.

References

Decision letter

  1. Mani Ramaswami
    Reviewing Editor; Trinity College Dublin, Ireland
  2. K VijayRaghavan
    Senior Editor; National Centre for Biological Sciences, Tata Institute of Fundamental Research, India

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "Modulation of sleep by trafficking of lipids through the Drosophila blood brain barrier" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Comments to the Authors:

We are sorry that, after consultation with the reviewers, we have decided that this work cannot be considered further for publication by eLife. However, we would be willing to consider an extensively revised manuscript if you choose to address the reviewers' concerns and resubmit to eLife as a new submission.

Specifically, as you will see from the comments below, while all of the reviewers found the question and approach of this study to be very interesting, all felt that several additional experiments including routine controls and procedural enhancements as well as careful rewriting would be required. Particularly critical is to attempt experiments to establish causality of acylcarnitine build up and sleep regulation (e.g through acylcarnitine microinjection?). A subset of other issues raised in the reviewer discussions included the need for: (1) more complete description of used methods used; (2) description of all drivers as well as transgenic constructs used, with references as well as validation of driver expression via UAS-GFP if not previously described (primarily to more clearly implicate barrier glia as against other glial cells). (3) validation of RNAi efficacy via qRT (repo>RNAi and head or brain mRNA levels); (4) ideally, key experiments to employ more than one RNAi transgene to ensure that effects are not due to off-target effects, while acknowledging that sometimes these genetics are difficult (e.g. with TRiP lines).

Reviewer #1 (Recommendations for the authors):

Artiushin et al. use a combination of metabolomics and behavioral genetics to detail the relationship between sleepiness and lipid transport in Drosophila. In previous studies, the Sehgal lab found that disrupting endocytosis in glia of the fly blood-brain barrier is sufficient to increase sleep. Here, the authors sought to explore the mechanism(s) behind this phenomenon by measuring metabolites in the heads of Repo>shi1 flies. LC-MS found elevations in a variety of acylcarnitines in the heads of long-sleeping Repo>shi1 flies, then behavioral experiments found that RNAi-mediated knock-down of two lipid-transport related genes (Lrp1 and Megalin) or two carnitine transporters (Orct and Orct2) are sufficient to significantly increase sleep time. Finally, LC-MS measurements in each long-sleeping RNAi-knock down genotype (Repo>Lrp1RNAi, MegalinRNAi and Repo>OrctRNAi, Orct2RNAi) also showed increased levels of acylcarnitine species. Together, these studies nicely parallel findings of heightened acylcarnitine levels in the blood of sleep-deprived humans, indicating that the fly provides a tractable model for understanding the mechanistic link between fatty acid metabolism and sleep.

While the authors' interpretations are consistent with their findings, the addition of a few key experiments, new analyses of behavioral data, and text edits could strengthen the manuscript.

Main comments:

The authors find sleep effects of manipulating fatty acid transport-related pathways in BBB glia, but it is not clear to me whether the acylcarnitines are moving into the brain or out of it. These possibilities each indicate different (and interesting) possibilities for sleep regulation and/or function. Along similar lines, the authors could test whether acylcarnitine levels are increased in the bodies of sleep deprived flies – this finding might help to distinguish whether these are sleep-promoting signals generated in the periphery or a metabolite that needs to be cleared from the brain.

If the sample material for LC-MS experiments came from whole heads, then it is possible that the head fat bodies contributed to the fatty acid changes shown in Figures 1, 5 & 6. The authors should mention and discuss this possibility in the manuscript.

Do the sleep changes in Figures2-4 alter sleep during the day or night (or both)? Do these RNAis alter sleep architecture or waking locomotor activity?

The authors use Gal80TS to restrict the timing of Mgl/Lrp1 RNAi (Figure 3B-C) and Orct/Orct2 RNAi (Figure 4B-C). A clearer description of the temperature protocols should be included. Do the 18°C and 31°C represent different groups of flies that were tested in parallel? Or do they show the same flies before and after a temperature increase? In addition, the flies housed at 18°C (when RNAi expression should be blocked) show similar behavioral phenotypes as the 31°C condition and as flies expressing the same RNAis without Gal80TS (see panels 3A & 4A, respectively). These data indicate either that the Gal80 repressor is not effectively blocking RNAi expression or that the sleep effects observed here cannot be solely attributed to the RNAi knock-downs.

Does supplementing fly food with acylcarnitines alter sleep? If so, are flies expressing Mgl/Lrp RNAi or Orct/Orct2 RNAi in perineural glia sensitized to dietary acylcarnitines?

Do all error bars indicate SEM as listed in the legends? Several of the behavioral plots appear to show overlapping error bars across comparisons that are designated as statistically significant. See Figures 3B, 3C, 4B, 4C and Figure 3 – —figure supplement 1.

Reviewer #2 (Recommendations for the authors):

Here, Artiushin et al. perform metabolomics on the heads transgenic mutants that exhibit long sleep due to inhibition of glial endocytosis (via repo driving shibire) compared to genetic controls. They identified multiple changes in metabolites; the most consistent and significant change was increased acylcarnitine levels. The authors performed a candidate screen by driving RNAi-mediated inhibition of 50 glia-enriched genes with pan-glial or sub-glial expression drivers. Knockdown of lipid transporter proteins in barrier glia led to increased acylcarnitines and increased sleep, suggesting that increased acylcarnitine levels may increase sleep. Previous results had shown that barrier glia and endocytosis were involved in sleep regulation; these data support the hypothesis that lipid metabolites, transported by barrier glia into the brain, mediate sleep regulation. Specifically, they point to the hypothesis that increased acylcarnitine levels in the brain are correlated with induction of sleep.

The conclusions of this paper are mostly well supported but some specific aspects need to be clarified or tested.

1) The n are low for several behavioral experiments and the number of trials is not clear for any experiments. Sleep and behavior are notoriously inconsistent from trial to trial. Experiments should be repeated with 16-32 flies per trial with 3 trials per experiment to demonstrate robustness of results.

2) As negative controls for subglial drivers that drive RNAi expression and affect sleep behavior, the authors should also test a non-barrier sub-glial expression driver such as eye glial drivers for effects on sleep behavior.

3) Are there other methods of increasing acylcarnitine levels in the body or brain that will increase sleep?

4) What are acylcarnitine levels for short-sleeping mutants?

This reviewer appreciates all the data and statistics listed for metabolomics. These data will be invaluable for other researchers in the field. Nevertheless, the precise n and trial number for each experiment should be listed.

Reviewer #3 (Recommendations for the authors):

This manuscript is a follow up to a previous publication from the group that linked blood-brain-barrier function to sleep in Drosophila. In this new report, they document changes in metabolites that occurs when the fly BBB is disrupted, with acylcarnitines accumulating in particular. Finally, using RNAi, they show that disruption of candidate transporters in the BBB alter sleep.

Overall, the manuscript provides promising data with the potential to be very interesting; however, the current version of this manuscript is hastily assembled. For example, the introduction refers somewhat vaguely to this group's previous paper, of which the current work is an extension, without really giving the reader much context of the state of the field. Although familiar with that previous paper, I had to refer back to it at several points when reading the current document to make full sense of the current manuscript. Particularly poor are the documentation of the methods; an expert in the field would not be able to replicate the findings based on the methods. For just one example-unless I am mistaken (and I misread Flybase), the Shi.ts1 manipulation used to impair the BBB is a temperature sensitive allele, yet there is no mention of samples being placed in restrictive or permissive temperatures. They do an RNAi screen in the middle of this paper, but that screen and its methods are not described. Not even all the fly stocks that appear to be used in the manuscript appear listed in the Methods.

More complete documentation of the experiments being performed would make it easier to properly evaluate the figures and conclusions.

The paper needs a major rewrite of the methods. Other areas that need to be addressed:

In Table 1 (and Figure 1), many of the metabolites labelled as interesting appear to be upregulated to quite a degree when comparing the two control strains, which makes it hard to judge whether these metabolic changes are meaningful or not. They appear to use a t-test for each metabolite, which further confounds the interpretation-some kind of false discovery rate correction should be applied to such large lists of comparisons.

On a related note, the manuscript could do a better job of guiding the reader to how the genetic manipulations are linked to the metabolite profile deemed important/different in the first figures. For example, comparing the results from Figure 5 and 6, it would appear that only C16 and C16.1 are showing differences that are found in Figure 1, but I could only find this upon careful examination myself when the authors should really do a better job to help us make sense of this data. Similarly, Figure 2's results don't seem in any way connected to the metabolite profiles, so I cannot understand why this data is included here.

Other aspects that I found difficult to assess:

– The RNAi screen, were more than one RNAi aimed at the same gene to confirm the effects?

– The raw values for metabolites are labelled as "area under the curve" in Figure 1 (unclear what that means) and plotted as μM in Figure 5.

– The n's for each experiment appear to be 13-16 per genotype. That sounds like a single tracking experiment to me. Were any of these experiments done as independent biological replicates (i.e. repeated on separate days?)

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Modulation of sleep by trafficking of lipids through the Drosophila blood brain barrier" for further consideration by eLife. Your revised article has been evaluated by K VijayRaghavan (Senior Editor) and a Reviewing Editor.

The reviewers have discussed their reviews with one another, and the Reviewing Editor requests you to consider responding appropriately to the remaining minor queries and comments.

Reviewer #1 (Recommendations for the authors):

Specific comments:

1. Does Figure 1 represent data from the same experiments described in Table 1? It's not clear to me whether Figure 1 shows more detail from the same data points or represents additional replicates. Also, the legend for Figure 1 describes the use of T-tests for significance – these should be analyzed using ANOVA instead with post hoc comparisons for the significance of each fatty acid.

2. The authors list the contents of each acylcarnitine type that is contained within the mixtures that are fed to flies, but a clearer discussion in the text that compares these lists with the lipids that were significantly changed in figure 1 might help the reader.

3. Is sleep architecture changed in flies that sleep more with acylcarnitine feeding?

Reviewer #2 (Recommendations for the authors):

The authors have addressed all of my concerns from the first version.

Reviewer #3 (Recommendations for the authors):

The authors did great work in addressing the majority of prior reviewer comments, specifically by clarifying their Materials and methods and adding supplementary experiments that increase confidence in their findings. The authors find that many different types of gene manipulations in the blood-brain barrier are associated with both long sleep and high acylcarnitine accumulation in the brain; the number of these experiments strongly supports a relationship between these three events. Precisely how these different gene manipulations in the BBB lead to acylcarnitine accumulation in the brain and how/where/why acylcarnitine accumulation in the brain might cause long sleep (as opposed to resulting from long sleep) remain unclear. Thus, while the authors present sufficient and exciting evidence that sleep, BBB function, and acylcarnitines are related, due to the lack of a cohesive model or mechanism, the authors are right to be cautious in making more specific claims about the causal nature of these relationships.

This manuscript was clearer and more well-written than the previous version. This reviewer did not go through with a fine-tooth comb to examine whether claims of causality were overstated; the authors should do this.

I include here a few points that could be mentioned in the discussion, results, or methods sections as caveats or clarifying points:

– Is there any evidence in the literature that acylcarnitines are removed from the brain via the BBB, as would be suggested by acylcarnitine accumulation after the loss of endocytosis or transporter function in BBB? Is this the authors' hypothesis?

– The feeding experiments, though the effects were mild, go one step toward addressing the role of acylcarnitines in triggering sleep except that it's not clear if eating acylcarnitines leads to acylcarnitine accumulation in the brain or if this is a separate effect.

– Another detail that the authors could add: the time of day of collection for metabolomics experiments. The authors might comment in the discussion on how this timing might relate to the mechanism.

https://doi.org/10.7554/eLife.86336.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Comments to the Authors:

We are sorry that, after consultation with the reviewers, we have decided that this work cannot be considered further for publication by eLife. However, we would be willing to consider an extensively revised manuscript if you choose to address the reviewers' concerns and resubmit to eLife as a new submission.

Specifically, as you will see from the comments below, while all of the reviewers found the question and approach of this study to be very interesting, all felt that several additional experiments including routine controls and procedural enhancements as well as careful rewriting would be required. Particularly critical is to attempt experiments to establish causality of acylcarnitine build up and sleep regulation (e.g through acylcarnitine microinjection?). A subset of other issues raised in the reviewer discussions included the need for: (1) more complete description of used methods used; (2) description of all drivers as well as transgenic constructs used, with references as well as validation of driver expression via UAS-GFP if not previously described (primarily to more clearly implicate barrier glia as against other glial cells). (3) validation of RNAi efficacy via qRT (repo>RNAi and head or brain mRNA levels); (4) ideally, key experiments to employ more than one RNAi transgene to ensure that effects are not due to off-target effects, while acknowledging that sometimes these genetics are difficult (e.g. with TRiP lines).

We have included (1) more complete methods; (2) description of drivers, including references for GFP expression of these; (3) validation of RNAi efficacy; (4) use of more than one RNAi transgene wherever possible. We have also conducted other additional experiments recommended by reviewers, as indicated below.

Reviewer #1 (Recommendations for the authors):

Artiushin et al. use a combination of metabolomics and behavioral genetics to detail the relationship between sleepiness and lipid transport in Drosophila. In previous studies, the Sehgal lab found that disrupting endocytosis in glia of the fly blood-brain barrier is sufficient to increase sleep. Here, the authors sought to explore the mechanism(s) behind this phenomenon by measuring metabolites in the heads of Repo>shi1 flies. LC-MS found elevations in a variety of acylcarnitines in the heads of long-sleeping Repo>shi1 flies, then behavioral experiments found that RNAi-mediated knock-down of two lipid-transport related genes (Lrp1 and Megalin) or two carnitine transporters (Orct and Orct2) are sufficient to significantly increase sleep time. Finally, LC-MS measurements in each long-sleeping RNAi-knock down genotype (Repo>Lrp1RNAi, MegalinRNAi and Repo>OrctRNAi, Orct2RNAi) also showed increased levels of acylcarnitine species. Together, these studies nicely parallel findings of heightened acylcarnitine levels in the blood of sleep-deprived humans, indicating that the fly provides a tractable model for understanding the mechanistic link between fatty acid metabolism and sleep.

While the authors' interpretations are consistent with their findings, the addition of a few key experiments, new analyses of behavioral data, and text edits could strengthen the manuscript.

Main comments:

The authors find sleep effects of manipulating fatty acid transport-related pathways in BBB glia, but it is not clear to me whether the acylcarnitines are moving into the brain or out of it. These possibilities each indicate different (and interesting) possibilities for sleep regulation and/or function. Along similar lines, the authors could test whether acylcarnitine levels are increased in the bodies of sleep deprived flies – this finding might help to distinguish whether these are sleep-promoting signals generated in the periphery or a metabolite that needs to be cleared from the brain.

This is a good idea, thank you. We assume the reviewer meant bodies of RepoGal4>Shi1, which were the ones analyzed here, and not sleep-deprived flies. We collected bodies of RepoGal4>Shi1, flies and controls and report here that while they show changes in lipid content, the changes are different from those seen in heads. Please see the results in Figure 1- supplement Figure 1.

If the sample material for LC-MS experiments came from whole heads, then it is possible that the head fat bodies contributed to the fatty acid changes shown in Figures 1, 5 & 6. The authors should mention and discuss this possibility in the manuscript.

Good point. We have included discussion of this possibility in our revised manuscript.

Do the sleep changes in Figures2-4 alter sleep during the day or night (or both)? Do these RNAis alter sleep architecture or waking locomotor activity?

Day and night changes in sleep are now shown in Figures2-4. We have also characterized sleep architecture for all behavioral assays and now show this, along with activity index (waking locomotor activity), in Figure 2 and in several supplemental figures for Figures3 and 4. None of the RNAis altered waking locomotor activity.

The authors use Gal80TS to restrict the timing of Mgl/Lrp1 RNAi (Figure 3B-C) and Orct/Orct2 RNAi (Figure 4B-C). A clearer description of the temperature protocols should be included. Do the 18°C and 31°C represent different groups of flies that were tested in parallel? Or do they show the same flies before and after a temperature increase? In addition, the flies housed at 18°C (when RNAi expression should be blocked) show similar behavioral phenotypes as the 31°C condition and as flies expressing the same RNAis without Gal80TS (see panels 3A & 4A, respectively). These data indicate either that the Gal80 repressor is not effectively blocking RNAi expression or that the sleep effects observed here cannot be solely attributed to the RNAi knock-downs.

A description of the temperature protocol has been added to the Methods and figure legend in the revised MS. The same flies were assayed before and after a temperature increase (Figure3 &4). We have done additional repeats of sleep tests and, as shown in Figures 3 and 4, the data indicate that the Gal80 repressor is effectively blocking RNAi expression.

Does supplementing fly food with acylcarnitines alter sleep? If so, are flies expressing Mgl/Lrp RNAi or Orct/Orct2 RNAi in perineural glia sensitized to dietary acylcarnitines?

To address the reviewer’s question, we fed iso31 flies mixes of acylcarnitines of different chain lengths and monitored sleep behavior. We found that acylcarnitines mixture 1 (C0 and C2) did not alter sleep amount, but low concentrations of acylcarnitine mixtures 2 and 3, which are composed of medium and long chain acylcarnitines, produced a small increase in sleep. Given that sleep is already increased with knockdown of Mgl/Lrp or Orct/Orct2 in perineural glia, we suspect that it would be difficult to detect a further increase with acylcarnitine feeding. Fly sleep easily gets to ceiling levels.

Do all error bars indicate SEM as listed in the legends? Several of the behavioral plots appear to show overlapping error bars across comparisons that are designated as statistically significant. See Figures 3B, 3C, 4B, 4C and Figure 3 – —figure supplement 1.

Sorry about this. Previous figures actually plotted SD in some cases, but we have now fixed this so error bars show SEM, as indicated in the legends of the revised MS.

Reviewer #2 (Recommendations for the authors):

Here, Artiushin et al. perform metabolomics on the heads transgenic mutants that exhibit long sleep due to inhibition of glial endocytosis (via repo driving shibire) compared to genetic controls. They identified multiple changes in metabolites; the most consistent and significant change was increased acylcarnitine levels. The authors performed a candidate screen by driving RNAi-mediated inhibition of 50 glia-enriched genes with pan-glial or sub-glial expression drivers. Knockdown of lipid transporter proteins in barrier glia led to increased acylcarnitines and increased sleep, suggesting that increased acylcarnitine levels may increase sleep. Previous results had shown that barrier glia and endocytosis were involved in sleep regulation; these data support the hypothesis that lipid metabolites, transported by barrier glia into the brain, mediate sleep regulation. Specifically, they point to the hypothesis that increased acylcarnitine levels in the brain are correlated with induction of sleep.

The conclusions of this paper are mostly well supported but some specific aspects need to be clarified or tested.

1) The n are low for several behavioral experiments and the number of trials is not clear for any experiments. Sleep and behavior are notoriously inconsistent from trial to trial. Experiments should be repeated with 16-32 flies per trial with 3 trials per experiment to demonstrate robustness of results.

All the experiments have been repeated at least 3 times and additional numbers have been added.

2) As negative controls for subglial drivers that drive RNAi expression and affect sleep behavior, the authors should also test a non-barrier sub-glial expression driver such as eye glial drivers for effects on sleep behavior.

As suggested by the reviewer, we used non-barrier glial drivers MZ0709-GAL4, NP2222-GAL4 and Eaat-GAL4 with tubgal80ts to express Lrps or Orcts RNAi in ensheathing glia, cortex glia and astrocyte-like glia respectively. None of these affected sleep (Figure 3—figure supplement 5 and Figure 4—figure supplement 2).

3) Are there other methods of increasing acylcarnitine levels in the body or brain that will increase sleep?

We used acylcarnitines as a food supplement for iso31 and report increases in sleep with some of them (Figure 1—figure supplement Figure 2). We also tried to inject acylcarnitines into fly heads but we could not control the amount and distribution of the injected material.

4) What are acylcarnitine levels for short-sleeping mutants?

We have conducted metabolomic analysis of three short-sleeping mutants Fmn, Rye and SssP1 and found that acylcarnitines are increased in these (Bedont et al., BioRxiv, also under consideration). As we believe these mutants to be chronically sleep-deprived, i.e. they have the need to sleep but are unable to implement it, these data support our hypothesis that acylcarnitines reflect high sleep need.

This reviewer appreciates all the data and statistics listed for metabolomics. These data will be invaluable for other researchers in the field. Nevertheless, the precise n and trial number for each experiment should be listed.

All the experiments have been repeated at least 3 times and new data added.

Reviewer #3 (Recommendations for the authors):

This manuscript is a follow up to a previous publication from the group that linked blood-brain-barrier function to sleep in Drosophila. In this new report, they document changes in metabolites that occurs when the fly BBB is disrupted, with acylcarnitines accumulating in particular. Finally, using RNAi, they show that disruption of candidate transporters in the BBB alter sleep.

Overall, the manuscript provides promising data with the potential to be very interesting; however, the current version of this manuscript is hastily assembled. For example, the introduction refers somewhat vaguely to this group's previous paper, of which the current work is an extension, without really giving the reader much context of the state of the field. Although familiar with that previous paper, I had to refer back to it at several points when reading the current document to make full sense of the current manuscript. Particularly poor are the documentation of the methods; an expert in the field would not be able to replicate the findings based on the methods. For just one example-unless I am mistaken (and I misread Flybase), the Shi.ts1 manipulation used to impair the BBB is a temperature sensitive allele, yet there is no mention of samples being placed in restrictive or permissive temperatures. They do an RNAi screen in the middle of this paper, but that screen and its methods are not described. Not even all the fly stocks that appear to be used in the manuscript appear listed in the Methods.

More complete documentation of the experiments being performed would make it easier to properly evaluate the figures and conclusions.

We apologize for the omissions. As noted, we have conducted additional experiments and also extensively re-written the manuscript to address these concerns. In particular, we now indicate that we maintained shibire flies for metabolomics at 30℃, even though sleep phenotypes are evident at permissive temperatures also (Artiushin et al., 2018) and we include details of the screen as well as a list of fly stocks in Methods.

The paper needs a major rewrite of the methods. Other areas that need to be addressed:

The methods section has been rewritten in the revised MS.

In Table 1 (and Figure 1), many of the metabolites labelled as interesting appear to be upregulated to quite a degree when comparing the two control strains, which makes it hard to judge whether these metabolic changes are meaningful or not. They appear to use a t-test for each metabolite, which further confounds the interpretation-some kind of false discovery rate correction should be applied to such large lists of comparisons.

Short, medium, long and very long chain acylcarninites are shown in Table 1 and Figure 1. We have used one-way ANOVA with Holk-Sidak post-hoc comparisons to compare these in experimental and control groups in the revised manuscript (Figure 1).

On a related note, the manuscript could do a better job of guiding the reader to how the genetic manipulations are linked to the metabolite profile deemed important/different in the first figures. For example, comparing the results from Figure 5 and 6, it would appear that only C16 and C16.1 are showing differences that are found in Figure 1, but I could only find this upon careful examination myself when the authors should really do a better job to help us make sense of this data. Similarly, Figure 2's results don't seem in any way connected to the metabolite profiles, so I cannot understand why this data is included here.

We apologize for this oversight, and have added comparison of Figure 5-6 data to Figure 1. Regarding Figure 2, we acknowledge that these data are not connected to metabolite profiles, but they were important findings of our transporter screen and we feel they should be reported. If the reviewer feels these data should be in the supplement, we can move them.

Other aspects that I found difficult to assess:

– The RNAi screen, were more than one RNAi aimed at the same gene to confirm the effects?

The revised manuscript reports the use of multiple RNAis for Lrp1/Mgl and Orct/Orct2. We have now also checked the knockdown efficiency of RNAi lines by qPCR (Figure 3—figure supplement 2 and Figure 4—figure supplement 3).

– The raw values for metabolites are labelled as "area under the curve" in Figure 1 (unclear what that means) and plotted as μM in Figure 5.

The metabolite data were from two different facilities (Metabolon and The Penn Metabolomics Core). They used different methods, ’area under the curve’ by Metabolon and μM by the Penn Metabolomics Core to show raw values.

– The n's for each experiment appear to be 13-16 per genotype. That sounds like a single tracking experiment to me. Were any of these experiments done as independent biological replicates (i.e. repeated on separate days?)

We show additional data in the revised MS. The sleep data are from at least 3 independent trials and about 48 flies per genotype.

[Editors’ note: what follows is the authors’ response to the second round of review.]

Reviewer #1 (Recommendations for the authors):

Specific comments:

1. Does Figure 1 represent data from the same experiments described in Table 1? It's not clear to me whether Figure 1 shows more detail from the same data points or represents additional replicates. Also, the legend for Figure 1 describes the use of T-tests for significance – these should be analyzed using ANOVA instead with post hoc comparisons for the significance of each fatty acid.

Yes. Figure 1 represents data from the same experiments described in Table 1. We did indeed use one-way ANOVA with Holk-Sidak post-hoc comparisons, and not T-tests, for significance in figure 1. Table 1 compares the experimental to each of the parental controls separately and so uses a T-test.

2. The authors list the contents of each acylcarnitine type that is contained within the mixtures that are fed to flies, but a clearer discussion in the text that compares these lists with the lipids that were significantly changed in figure 1 might help the reader.

We have added a comparison of acylcarnitine mixtures (from Σ) and the ones that are significantly changed in figure1.

3. Is sleep architecture changed in flies that sleep more with acylcarnitine feeding?

We have added the analysis of sleep architecture in flies fed acylcarnitines, in multiple additional supplemental figures.

Reviewer #3 (Recommendations for the authors):

The authors did great work in addressing the majority of prior reviewer comments, specifically by clarifying their Materials and methods and adding supplementary experiments that increase confidence in their findings. The authors find that many different types of gene manipulations in the blood-brain barrier are associated with both long sleep and high acylcarnitine accumulation in the brain; the number of these experiments strongly supports a relationship between these three events. Precisely how these different gene manipulations in the BBB lead to acylcarnitine accumulation in the brain and how/where/why acylcarnitine accumulation in the brain might cause long sleep (as opposed to resulting from long sleep) remain unclear. Thus, while the authors present sufficient and exciting evidence that sleep, BBB function, and acylcarnitines are related, due to the lack of a cohesive model or mechanism, the authors are right to be cautious in making more specific claims about the causal nature of these relationships.

This manuscript was clearer and more well-written than the previous version. This reviewer did not go through with a fine-tooth comb to examine whether claims of causality were overstated; the authors should do this.

I include here a few points that could be mentioned in the discussion, results, or methods sections as caveats or clarifying points:

– Is there any evidence in the literature that acylcarnitines are removed from the brain via the BBB, as would be suggested by acylcarnitine accumulation after the loss of endocytosis or transporter function in BBB? Is this the authors' hypothesis?

This is an interesting point. We were not able to find anything in the literature that says acylcarnitines are removed from the brain via the BBB. It is possible that this uptake is specific for glial cells (Drosophila BBB is composed entirely of glia). Please note too that although the role of the BBB is supported by the sleep phenotypes we get with knockdown of lipid/carnitine transporters specifically in the BBB, the metabolomic analysis was conducted with manipulations of all glia.

– The feeding experiments, though the effects were mild, go one step toward addressing the role of acylcarnitines in triggering sleep except that it's not clear if eating acylcarnitines leads to acylcarnitine accumulation in the brain or if this is a separate effect.

It is true that at this point we do not know if the fed acylcarnitines accumulate in the brain. We have acknowledged this point in the manuscript.

– Another detail that the authors could add: the time of day of collection for metabolomics experiments. The authors might comment in the discussion on how this timing might relate to the mechanism.

The flies were collected at ZT 0-3 for the metabolomics analysis. We have added discussion of how the timing relates to the results.

https://doi.org/10.7554/eLife.86336.sa2

Article and author information

Author details

  1. Fu Li

    Howard Hughes Medical Institute and Chronobiology and Sleep Institute, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Perelman School of Medicine, University of Pennsylvania, Philadelphia, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft, Writing – review and editing
    Contributed equally with
    Gregory Artiushin
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0879-8008
  2. Gregory Artiushin

    Howard Hughes Medical Institute and Chronobiology and Sleep Institute, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Perelman School of Medicine, University of Pennsylvania, Philadelphia, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft
    Contributed equally with
    Fu Li
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1615-3012
  3. Amita Sehgal

    Howard Hughes Medical Institute and Chronobiology and Sleep Institute, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Perelman School of Medicine, University of Pennsylvania, Philadelphia, United States
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Project administration, Writing – review and editing
    For correspondence
    amita@pennmedicine.upenn.edu
    Competing interests
    Reviewing editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7354-9641

Funding

National Institute of Diabetes and Digestive and Kidney Diseases (R01DK120757)

  • Amita Sehgal

Howard Hughes Medical Institute

  • Amita Sehgal

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Dr. Chris Petucci and the Penn Metabolomics Core for providing measurements of acylcarnitines in fly heads. We thank Zhifeng Yue for her support with the maintenance of fly strains. The work was supported by HHMI and by R01DK120757.

Senior Editor

  1. K VijayRaghavan, National Centre for Biological Sciences, Tata Institute of Fundamental Research, India

Reviewing Editor

  1. Mani Ramaswami, Trinity College Dublin, Ireland

Version history

  1. Preprint posted: February 19, 2022 (view preprint)
  2. Received: January 21, 2023
  3. Accepted: April 11, 2023
  4. Accepted Manuscript published: May 4, 2023 (version 1)
  5. Accepted Manuscript updated: May 5, 2023 (version 2)
  6. Version of Record published: May 23, 2023 (version 3)

Copyright

© 2023, Li, Artiushin et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,204
    Page views
  • 205
    Downloads
  • 2
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Fu Li
  2. Gregory Artiushin
  3. Amita Sehgal
(2023)
Modulation of sleep by trafficking of lipids through the Drosophila blood-brain barrier
eLife 12:e86336.
https://doi.org/10.7554/eLife.86336

Share this article

https://doi.org/10.7554/eLife.86336

Further reading

    1. Neuroscience
    E Nicholas Petersen, Mahmud Arif Pavel ... Scott B Hansen
    Research Article

    Rapid conversion of force into a biological signal enables living cells to respond to mechanical forces in their environment. The force is believed to initially affect the plasma membrane and then alter the behavior of membrane proteins. Phospholipase D2 (PLD2) is a mechanosensitive enzyme that is regulated by a structured membrane-lipid site comprised of cholesterol and saturated ganglioside (GM1). Here we show stretch activation of TWIK-related K+ channel (TREK-1) is mechanically evoked by PLD2 and spatial patterning involving ordered GM1 and 4,5-bisphosphate (PIP2) clusters in mammalian cells. First, mechanical force deforms the ordered lipids, which disrupts the interaction of PLD2 with the GM1 lipids and allows a complex of TREK-1 and PLD2 to associate with PIP2 clusters. The association with PIP2 activates the enzyme, which produces the second messenger phosphatidic acid (PA) that gates the channel. Co-expression of catalytically inactive PLD2 inhibits TREK-1 stretch currents in a biological membrane. Cellular uptake of cholesterol inhibits TREK-1 currents in culture and depletion of cholesterol from astrocytes releases TREK-1 from GM1 lipids in mouse brain. Depletion of the PLD2 ortholog in flies results in hypersensitivity to mechanical force. We conclude PLD2 mechanosensitivity combines with TREK-1 ion permeability to elicit a mechanically evoked response.

    1. Developmental Biology
    2. Neuroscience
    Athina Keramidioti, Sandra Schneid ... Charles N David
    Research Article

    The Hydra nervous system is the paradigm of a ‘simple nerve net’. Nerve cells in Hydra, as in many cnidarian polyps, are organized in a nerve net extending throughout the body column. This nerve net is required for control of spontaneous behavior: elimination of nerve cells leads to polyps that do not move and are incapable of capturing and ingesting prey (Campbell, 1976). We have re-examined the structure of the Hydra nerve net by immunostaining fixed polyps with a novel antibody that stains all nerve cells in Hydra. Confocal imaging shows that there are two distinct nerve nets, one in the ectoderm and one in the endoderm, with the unexpected absence of nerve cells in the endoderm of the tentacles. The nerve nets in the ectoderm and endoderm do not contact each other. High-resolution TEM (transmission electron microscopy) and serial block face SEM (scanning electron microscopy) show that the nerve nets consist of bundles of parallel overlapping neurites. Results from transgenic lines show that neurite bundles include different neural circuits and hence that neurites in bundles require circuit-specific recognition. Nerve cell-specific innexins indicate that gap junctions can provide this specificity. The occurrence of bundles of neurites supports a model for continuous growth and differentiation of the nerve net by lateral addition of new nerve cells to the existing net. This model was confirmed by tracking newly differentiated nerve cells.