Zohar Abergel1, Maayan Shaked1, Virendra Shukla1, Zheng-Xing Wu2, Einav Gross1. 1. Department of Biochemistry and Molecular Biology, Faculty of Medicine, IMRIC, The Hebrew University of Jerusalem, Jerusalem, Israel. 2. Key Laboratory of Molecular Biophysics of Ministry of Education, Department of Biophysics and Molecular Physiology, College of Life Science and Technology, Institute of Biophysics and Biochemistry, Huazhong University of Science and Technology, Wuhan, P.R. China.
Abstract
Among the fascinating adaptations to limiting oxygen conditions (hypoxia) is the suppression of food intake and weight loss. In humans, this phenomenon is called high-altitude anorexia and is observed in people suffering from acute mountain syndrome. The high-altitude anorexia appears to be conserved in evolution and has been seen in species across the animal kingdom. However, the mechanism underlying the recovery of eating behavior after hypoxia is still not known. Here, we show that the phosphatidylinositol transfer protein PITP-1 is essential for the fast recovery of eating behavior after hypoxia in the nematode Caenorhabditis elegans. Unlike the neuroglobin GLB-5 that accelerates the recovery of eating behavior through its function in the oxygen (O2 )-sensing neurons, PITP-1 appears to act downstream, in neurons that express the mod-1 serotonin receptor. Indeed, pitp-1 mutants display wild-type-like O2 -evoked-calcium responses in the URX O2 -sensing neuron. Intriguingly, loss-of-function of protein kinase C 1 (PKC-1) rescues pitp-1 mutants' recovery after hypoxia. Increased diacylglycerol (DAG), which activates PKC-1, attenuates the recovery of wild-type worms. Together, these data suggest that PITP-1 enables rapid recovery of eating behavior after hypoxia by limiting DAG's availability, thereby limiting PKC activity in mod-1-expressing neurons.
Among the fascinating adaptations to limiting oxygen conditions (hypoxia) is the suppression of food intake and weight loss. In humans, this phenomenon is called high-altitude anorexia and is observed in people suffering from acute mountain syndrome. The high-altitude anorexia appears to be conserved in evolution and has been seen in species across the animal kingdom. However, the mechanism underlying the recovery of eating behavior after hypoxia is still not known. Here, we show that the phosphatidylinositol transfer protein PITP-1 is essential for the fast recovery of eating behavior after hypoxia in the nematode Caenorhabditis elegans. Unlike the neuroglobin GLB-5 that accelerates the recovery of eating behavior through its function in the oxygen (O2 )-sensing neurons, PITP-1 appears to act downstream, in neurons that express the mod-1serotonin receptor. Indeed, pitp-1 mutants display wild-type-like O2 -evoked-calcium responses in the URX O2 -sensing neuron. Intriguingly, loss-of-function of protein kinase C 1 (PKC-1) rescues pitp-1 mutants' recovery after hypoxia. Increased diacylglycerol (DAG), which activates PKC-1, attenuates the recovery of wild-type worms. Together, these data suggest that PITP-1 enables rapid recovery of eating behavior after hypoxia by limiting DAG's availability, thereby limiting PKC activity in mod-1-expressing neurons.
acute mountain syndromeelegansCaenorhabditis elegansdiacylglycerolglobininositol 1,4,5‐trisphosphateneuropeptide receptor 1phosphatidylinositol transfer proteinphospholipase C betaphorbol 12‐myristate 13‐acetateprotein kinase C 1phosphatidylinositolphosphatidic acidsoluble guanylate cyclases
INTRODUCTION
Eating behavior influences our food choice, meal timing, and quantity, and therefore, determines our health and disease state.
The factors affecting eating behavior are numerous and diverse and include internal and external cues, one of which being oxygen (O2). O2 is essential for energy production in all aerobic animals, and therefore, insufficient O2 supply (hypoxia) is a significant stress condition in most animals, including humans. Among the fascinating responses to hypoxia is the suppression of food intake and, consequently, body mass loss. This phenomenon is called high‐altitude anorexia (or hypoxia‐anorexia) and is observed in people suffering from acute mountain syndrome (AMS).
AMS occurs in response to a rapid exposure to moderate or high altitudes (above 4000 meters), where the partial O2 pressure is low. Results from Operation Everest III (Comex‐’97) demonstrated that in a controlled environment that mimicked the decrease in O2 concentration while climbing on Everest at 5000 to 8000 meters, hypoxia was the stimulus that changed the eating behavior of volunteers and decreased food intake due to a loss in appetite.
Intriguingly, the hypoxic‐anorexia phenomenon appears to be conserved in evolution and has been observed in crabs, flies, gastropods, fish, rodents,
and in the nematode Caenorhabditis elegans (C. elegans).
,
The hypoxic‐anorexia effect is reversible. Indeed, data from Comex‐’97 show that upon returning to ambient O2‐levels, the volunteers’ body mass increases.
However, the mechanism underlying the recovery of eating behavior after hypoxia is still poorly understood.The eating behavior of C. elegans is modulated by neuronal O2‐sensors and neuropeptide signaling
,
. Animals bearing the dominant neuropeptide G‐protein‐coupled receptor npr‐1(215V) allele (e.g., N2 laboratory strain) move slowly on bacteria (the C. elegans food source) and do not clump together on the bacterial lawn border at 21% O2
By contrast, animals bearing either the recessive npr‐1(215F) allele (found in many wild C. elegans isolates. e.g., the Hawaii strain CB4856) or the strong loss‐of‐function allele npr‐1(ad609), hereafter referred to as npr‐1(−) worms), move fast on food and clump together on the bacterial lawn border (bordering eating behavior) at 21% O2. The bordering behavior is dependent on the activity of the atypical soluble guanylate cyclases (sGCs) O2‐sensors GCY‐35 and GCY‐36.
,
Upon O2 binding, the GCY‐35/GCY‐36 functional complex is thought to generate cGMP that triggers the opening of the TAX‐2/TAX‐4 cyclic nucleotide‐gated channel complex. The opening of TAX‐2/TAX‐4 facilitates calcium (Ca2+) entry, and thus, the depolarization of the O2‐sensing neurons AQR, PQR, and URX. This cascade activates an escape response from hyperoxia (O2 > 12%,
), which is suppressed by the activity of NPR‐1(215V) in the presence of food. Intriguingly, the clumping of worms on the bacteria lawn border appears to decrease the O2 level that the worm experience. Indeed, a previous study demonstrated that the O2 level inside the clump ranges between 6.4% and 13.7% O2,
and this is the range of concentrations to which the worm is attracted.Hypoxia (1% O2) suppresses the bordering eating behavior and results in worms accumulating outside the bacterial lawn border (Figure 1A). This food‐leaving behavior is reversible upon returning to 21% O2 and is regulated by the neuroglobin GLB‐5.
,
FIGURE 1
PITP‐1 is essential for the fast recovery of bordering behavior after hypoxia. A, Hypoxia inhibits the bordering behavior in a reversible way. Left panel, hypoxia‐induced food‐leaving behavior displayed by glb‐5(+); npr‐1(−) worms after 16 hours incubation in hypoxia (1% O2). Worms resume robust bordering eating behavior after 30 minutes exposure to 21% O2 (right panel). In the right panel, the left arrowhead indicates the accumulation of worms on the bacterial lawn border (i.e., bordering). The right arrowhead indicates a region where the bacterial lawn border in readily observed. Scale bar: 1 mm. B, A schematic diagram of the pitp‐1 gene highlighting the splice site mutation in the heb2 allele. C and D, Bar graphs presenting the recovery of bordering after 24 hours in hypoxia. The bordering index percentage is the fraction of worms located on the bacterial lawn border divided by the total number of worms in the plate, multiplied by 100. Asterisks indicate significance for comparisons with glb‐5(+); npr‐1(−) animals after 24 hours at 1% O2. Data represent the average of at least six independent experiments. Two‐way ANOVA with Bonferroni posttest. ****P < .0001, NS, nonsignificant. Error bars represent SEM
PITP‐1 is essential for the fast recovery of bordering behavior after hypoxia. A, Hypoxia inhibits the bordering behavior in a reversible way. Left panel, hypoxia‐induced food‐leaving behavior displayed by glb‐5(+); npr‐1(−) worms after 16 hours incubation in hypoxia (1% O2). Worms resume robust bordering eating behavior after 30 minutes exposure to 21% O2 (right panel). In the right panel, the left arrowhead indicates the accumulation of worms on the bacterial lawn border (i.e., bordering). The right arrowhead indicates a region where the bacterial lawn border in readily observed. Scale bar: 1 mm. B, A schematic diagram of the pitp‐1 gene highlighting the splice site mutation in the heb2 allele. C and D, Bar graphs presenting the recovery of bordering after 24 hours in hypoxia. The bordering index percentage is the fraction of worms located on the bacterial lawn border divided by the total number of worms in the plate, multiplied by 100. Asterisks indicate significance for comparisons with glb‐5(+); npr‐1(−) animals after 24 hours at 1% O2. Data represent the average of at least six independent experiments. Two‐way ANOVA with Bonferroni posttest. ****P < .0001, NS, nonsignificant. Error bars represent SEMSimilar to npr‐1, glb‐5 is a polymorphic gene that affects the worm bordering behavior and foraging speed.
In the case of the npr‐1‐gene, N2 worms bear the dominant npr‐1(215V) gain‐of‐function allele and the Hawaii strain CB4856 bears the recessive npr‐1(25F) allele. In contrast, with the glb‐5 gene, N2 worms bear the Bristol allele, hereafter referred to as glb‐5(−), which encodes a truncated neuroglobin that appears to be nonfunctional.
By contrast, CB4856 worms bear the Hawaii allele of glb‐5, hereafter referred to as glb‐5(+), which encodes a full‐length functional neuroglobin that appears to represent the ancestral glb‐5 allele, as it is found in most C. elegans wild‐isolates.
GLB‐5(+) fine‐tunes the O2‐responses of worms close to 21% O2. Indeed, npr‐1(−) worms that bear the glb‐5(+) allele (hereafter referred to as glb‐5(+);npr‐1(−) worms) are more sensitive to subtle changes in O2 level and sharply decrease their foraging speed in response to a 21% to 17% O2 shift, whereas npr‐1(−) worms (bearing the glb‐5(−) allele) display only a mild decrease in response to this stimulus.
Like the mammalian cytoglobin
and neuroglobin,
GLB‐5 is a hexacoordinated globin.
Hypotheses for hexacoordinated globins' in vivo functions include regulation of cellular signaling
,
,
and protection against oxidative damage.
,GLB‐5 controls worms' speed and recovery from hypoxia through its activity in the O2‐sensing neurons AQR, PQR, and URX.
The recovery of the bordering behavior in glb‐5(+); npr‐1(−) worms, is rapid and occurs within minutes after returning to 21% O2.
By contrast, the recovery of npr‐1(−) worms bearing the glb‐5(−) allele is slow and is completed only after 4 hours. This significant difference in recovery time has facilitated the isolation of several suppressor mutations that attenuate bordering behavior recovery by suppressing GLB‐5 activity in the O2‐sensing neurons. That is, glb‐5(+); npr‐1(−) animals bearing these mutations do not slowdown in response to a 21% to 17% O2 shift and display high URX‐Ca2+ levels at both 14% and 17%% O2.
,
Here, we discover a new type of suppressor mutation in the gene encoding for the phosphatidylinositol (PtdIn) transfer protein PITP‐1, which attenuates the recovery of bordering eating behavior without affecting the function of GLB‐5 in the O2‐sensing neurons. Our data suggest that PITP‐1 facilitates fast recovery of eating behavior after hypoxia by restricting the activation of protein kinase C 1 (PKC‐1) by diacylglycerol in mod‐1‐expressing neurons. This function is distinct from the role of PITP‐1 in salt chemotaxis.
To generate synchronized worms, we collected eggs from gravid hermaphrodites using hypochlorite solution.
In brief, we collected gravid hermaphrodites into 15 mL tubes by washing the NGM plates three times with M9 buffer (22 mM KH2PO4, 42 mM Na2HPO4, 86 mM NaCl, and 1 mM MgSO4). We centrifuge the tubes for 1 min (1690 g, 1min), and removed the supernatant until 1 mL of volume remained. Then, we added 1 ml of hypochlorite solution (0.5 N NaOH, 1.25% NaOCl) to each tube and inverted it five times. To assist the release of embryos, we aspirated the worm suspensions (back and forth several times) using a syringe with a 21‐gauge needle, for 3 minutes. Then, we sedimented the embryos by centrifugation (1690 g for 2 minutes) and removed most of the hypochlorite solution. Each tube was washed three times with 5 mL of M9 buffer. Next, we removed most of the M9 buffer (without disturbing the embryos pellet) and added 2 mL of fresh M9 buffer to each tube. We rotated the tubes for 16 hours at room temperature (RT, 21ºC). The hatched L1 larvae were collected by centrifugation (1690 g for 3 minutes), counted, and placed into NGM plates seeded with OP50 E coli bacteria and grown until the desired developmental stage.
Bordering behavior assays
Bordering experiments were performed, as described previously.
In brief, 40 L4 hermaphrodites, grown in 21% O2/21°C, were transferred into a 3.5 cm NGM plate that was seeded 2 days before with 50 µL OP50 bacteria (OD600 ~ 0.6). The assay plates were put in 1% O2 (Coy hypoxia chamber, COY Lab Products Inc, Grass Lake, MI, USA, RT) for 24 hours, and then, brought back to 21% O2 (control plates were maintained in 21% O2 at 21°C). We calculated the bordering index (BI) after 30 minutes recovery at 21% O2. BI (%) represents the fraction of worms found on the bacterial lawn border divided by the total amount of worms on the plate, multiplied by 100%.
Speed measurements
All speed measurements were done in the presence of food as described in detail in Abergel et al.
In brief, 2 days before the experiment, we put 20 µL OP50 bacteria (OD600 ~ 0.6) on each low‐NGM plate (0.13 g/L bacto‐peptone) and incubated the plates at RT For each speed experiment, we trapped 8‐10 young hermaphrodites inside 500 µm deep rectangular PDMS (12 mm wide, 0.5 mm deep, and 17 mm long). We used a PHD 2000 syringe pump (Harvard Apparatus) to deliver humidified gases to the microfluidic chambers at a flow rate of 0.5 mL/min. We used Teflon valves, controlled by a ValveBank Controller (Automate Scientific), to rapidly switch between different O2/N2 gas mixtures. We recorded the movies using a Q‐Imaging MicroPublisher 5.0 RTV Microscope Camera (QImaging, RHos) mounted onto an Olympus SZ61 stereomicroscope (Olympus). The videos were taken at 0.5 frames/s. We used custom MATLAB software (The MathWorks)
for video analysis. Notably, in our analysis, we did not track worms that leave the bacterial lawn, and thus, only calculated the speed of worms on food.
Salt chemotaxis
Salt (NaCl) chemotaxis assays were performed as previously described,
with some modifications. In brief, a day before the experiment, we generated a NaCl gradient on a 9‐cm chemotaxis assay plate containing 10 mL of 2% agar, 5 mM potassium phosphate (pH 6.0), 1 mM CaCl2, and 1 mM MgSO4. To generate the gradient, we placed a 1.2 cm‐diameter agar plug, containing 100 mM of NaCl, 1 centimeter from the edge of the plate. We put one microliter of 0.5 M NaN3 to the peak of NaCl gradient (the NaCl agar plug) and the equivalent place on the opposite side of the plate. We placed ~200 synchronized young adult hermaphrodites at the center of the assay plate. After 30 minutes, we counted the worms at the NaCl peak and the control spot (1 cm radius from the NaN3 drop). A chemotaxis index was calculated as followed: [# of animals at the NaCl peak)—(# of animals at the control spot)]/[(total # of animals)—(# of animals at the center of the plate)]. Each data point represents at least six biological repeats.
Phorbol 12‐myristate 13‐acetate treatment
Two days before the experiment, we prepared NGM plates containing 100 µg/mL of ampicillin and 0.1 µg/mL of phorbol 12‐myristate 13‐acetate (PMA) (dissolved in DMSO, p8139 Sigma Aldrich) or 0.001998% of DMSO as a control. After 24 hours at room temperature (RT), we seeded the plates with 50 µL ampicillin‐resistant‐OP50 bacteria (OD600 ~ 0.6) and let them dry for 24 hours at RT. At the day of the experiment, we placed 40 L4 staged hermaphrodites on the plates and transferred them to 1% O2 (Coy hypoxia chamber, COY Lab Products Inc, Grass Lake, MI, USA, RT) for 24 hours; control plates remained at 21% O2. Bordering recovery was measured as described above.
Ethyl methanesulfonate mutagenesis screen
The ethyl methanesulfonate (EMS) screen was performed as described in detail in Abergel et al.
Mapping
We mapped the heb2 mutation to a 1.6 Mb genomic interval (on chromosomes III) using SNPs (single nucleotide polymorphisms).
The list of additional SNPs primers we used is presented in Table 1. The molecular identity of heb2 was determined by next‐generation sequencing (NGS) at the Technion Genome Center.
TABLE 1
SNP primers, location, band sizes (observed after restriction enzyme digestion) and the respective restriction enzyme. The primer sequences are displayed from 5′ to 3′
Genetic location
Genomic location
Clone
N2 digest
CB4856 digest
Enzyme
Primers
III, ‐11
2629011‐2631354
Y71H2B
355
174, 181
HpyCH4V
F: CCGCGGTGTTCTCTTTGAAATTCTC
R: GCTCCTTACAGCTTTCAGATCCTC
III, ‐10.1
2691329‐2835114
Y71H2AM
418
27, 124, 267
HphI
F: GATCTGGCAGCTCTGGCGG
R: GTGCAAGATCCATCGATTCAACG
III, ‐9
2850762‐2893560
F56F11a
251, 304
555
AccI
F: GGGTCACTTGAAAGGTCTTAACAA
R: TTTCCCGCAATTCGGATGTGTG
III, ‐8.9
2891762‐2892784
F56F11
236, 332
568
AvaII
F: CAGAGCTTCAAGTTTATGGAGGCA
R: GTACCTCCAAAGTACGCAAACACC
III, ‐7.9
3052479‐3100358
H06I04
513
234, 279
SpeI
F: CTGTAAATCTCAATATCCAAGTGGC
R: CATGCTTGACTGACCCTAGTTC
III, ‐6.6
3452186‐3486946
C34C12
54, 96, 182
332
NlaIII
F: GCAACCGGTCGCCTTCATAC
R: TTGTTAACTATGATGCATCGAGGTG
III, ‐5.4
3558492‐3596862
C48G5
641
259, 382
BglII
F: GGGGCTTCTTCATTGCGCTCAC
R: TTTTCGACGGATCCAGCCATTTCTG
III, ‐3.8
4112743‐4142803
C30D11
117, 473
590
RsaI
F: CATCGAGAAGGACAAGTCGACTA
R: GGCTCCAAGGCTGCTGCA
SNP primers, location, band sizes (observed after restriction enzyme digestion) and the respective restriction enzyme. The primer sequences are displayed from 5′ to 3′
Molecular biology
Genotyping: The glb‐5(Bri) duplication was followed by examining the PCR products amplified using primers that flank the duplication in this allele.
The npr‐1(215V) and npr‐1(ad609) alleles were followed by amplifying the interval containing the mutated allele, and MwoI digestion that distinguishes between the 215V and ad609 alleles. The pitp‐1(tm1500), pkc‐1 (ok563), and the dgk‐1(ok1462) deletion were followed by PCR using primers that flank the deletion region.
Transgenes
All rescuing plasmids were generated by either restriction enzyme
or Gibson assembly cloning
using modified polycistronic mCherry pPD95.75 expression vector, as described previously.For the pitp‐1(genomic) rescuing construct [Ppitp‐1::pitp‐1(gDNA)::polycismCherry], we amplified the pitp‐1 promoter (~4.2 Kb), and genomic region (~8.2 kb) using N2 genomic DNA. The promoter region was amplified with 5′ GTTATCCGCTCCAATCCATCCTTCCAT 3′ and 5′ CCTCTTCTTTTTCGACTCGACAACCCAG 3′ and inserted using BspEI and AvrII. The genomic sequence of pitp‐1 was cloned in two steps. The first genomic part was amplified using 5′ ATGCTGATCAAAGAATACCGTATCCTGTTAC 3′ and 5′ CAGAATTTTGGTACTCACTGTCATCTTG 3′ and inserted using AvrII and PacI. The second part was amplified using 5′ AAAAGTTTGTAAAAACTGATAAAATTTTAGATTCTCCT 3′ and 5′ CTACCGACGATCTTTTTCATTTTCG 3′ inserted using PacI and NotI. To create the pitp‐1 cDNA (~3.1 kb) [Ppitp‐1::pitp‐1(cDNA)::polycismCherry] rescuing construct, the genomic part of the pitp‐1 gene was replaced with pitp‐1 cDNA. The coding region of pitp‐1 was amplified using 5' ATGCTGATCAAAGAATACCGTATCCTGTTAC 3′ and 5′ CTACCGACGATCTTTTTCATTTTCGAATTTTCC 3' and inserted using AvrII and NotI. To express the pitp‐1 cDNA in the oxygen sensing neurons we cut the gcy‐37 and glb‐5 promoters regions from an existing plasmids using BspEI and AvrII, these promoters regions (~1.3 kb, ~3.1 kb, respectively) replace the pitp‐1 promoter, and thus, generate the [Pgcy‐37::pitp‐1(cDNA)::polycismCherry] and [Pglb‐5::pitp‐1(cDNA)::polycismCherry] expression constructs. To express the pitp‐1 cDNA in other specific neurons, we amplified the different promoters using N2 genomic DNA and replaced the promoter region of pitp‐1. To clone the promoters, we introduced a NgoMIV restriction site to the plasmid backbone (using QuikChange II XL, Agilent Technologies). flp‐6 promoter region (~2.8 kb) was amplified with 5′ CAATTGGATAGGAAATGTCACACGAC 3′ and 5′ ATTCTGGAATAATCATATTGTTTTC 3′ and inserted using NgoMIV and AvrII; [Pflp‐6::pitp‐1(cDNA)::polycismCherry]. ceh‐36 promoter region (~3.1 kb) was amplified with 5′ GAACTCCCGCAGAATGCCAACTTG 3′ and 5′ TGTGCATGCGGGGGCAGGCG 3′ and inserted using NgoMIV and AvrII; [Pceh‐36::pitp‐1(cDNA)::polycismCherry]. rgef‐1 promoter region(~3.6kb) was amplified with 5′ CATCGAATGCAAAAGTCGCAACCTC 3′ and 5′ CGTCGTCGTCGTCGATGCCGTC 3' and inserted using NgoMIV and AvrII; [Prgef‐1::pitp‐1(cDNA)::polycismCherry]. The tax‐2 promoter region (~1.8 kb upstream from ATG+ the first 215 amino acid) was amplified with 5′ CAATGACATCAAACTCGGTCGATGC 3′ and 5′ ATGAAATGTTCCTTTCAATGAAAACCCAAAGC 3′ and inserted using NgoMIV and AvrII; [Ptax‐2::pitp‐1(cDNA)::polycismCherry]. The ocr‐2 promoter region (~3.8kb upstream from ATG+ the first 98 amino acid) was amplified with 5′ GTATGGGAATGTCAGCATGGGAACG 3′ and 5′ CTTTCGTGAGAATGCTCCAAGTCGTTC 3′ and inserted using NgoMIV and AvrII; [Pocr‐2::pitp‐1(cDNA)::polycismCherry]. The gpa‐14 promoter region (~3.3 kb) was amplified with 5' GGTTTCGATGATGTGGTACTTTGGC 3′ and 5′ AAGCTCTCTCAGTTTTGCTTCAAGCTC 3' and inserted using NgoMIV and AvrII; [Pgpa‐14::pitp‐1(cDNA)::polycismCherry]. The cng‐3 promoter region (~760 bp) was amplified with 5′ CACCTCGTGGAATACGGACCTGAAA 3′ and 5′ GTATTGGTGTCAAGAAGGGGGTAG 3′ and inserted using NgoMIV and AvrII; [Pcng‐3::pitp‐1(cDNA)::polycismCherry]. The mod‐1 promoter region (5.5 kb) was cloned in two steps. The first genomic part was amplified using 5′ GTGTTTTCTCGTAATCCCGTTGTACTTCATTG 3′ and 5′ GTCGACAATGCATAAAGGTTCAGTTGCAAG 3′ with SalI restriction enzymes recognize sequence and inserted using NgoMIV and AvrII. The second part was amplified using 5′ ACGTCAAAATCGAATCATTGCTTGGTCTGG 3′ and 5′ AATTTTCTTTCACCGCATTGGCACCTGGAG' inserted using SalI and AvrII; [Pmod‐1:pitp‐1(cDNA) polycismCherry]. The ttx‐7 promoter region (~0.6 kb) was amplified with 5′ CCAATTGTTAACCTTGAGTATGTCTTCAC 3′ and 5′ CTAAAACTTAAAATCAATGAATAAAGAGCTGAAG 3' and inserted using NgoMIV and AvrII; [Pttx‐7::pitp‐1(cDNA)::polycismCherry]. The constructs were injected into pitp‐1(tm1500);glb‐5(Haw);npr‐1(ad609) worms (0.5‐2.5 ng/µL) with the PF15E11.1::GFP co‐injection marker (47.5‐49.5 ng/µL).For the pkc‐1 cDNA rescuing constructs, we cut the pitp‐1 cDNA under rgef‐1 and mod‐1 promoters with AvrII and NotI and inserted the pkc‐1 cDNA with AvrII and NotI; [Prgef‐1:pkc‐1(cDNA)::polycismCherry] and [Pmod‐1‐1:pkc‐1(cDNA)::polycismCherry], respectively. The pkc‐1 cDNA (isoform c) region (~2.3 kb) was amplified with 5′ ATGAAATTCTTCAGTAGTCGGACAATATCATCTG 3′ and 5′ TTAGTAGGTAAAATGCGGATTGATAAATGAAAAACC 3′. The constructs were injected into pitp‐1(tm1500);glb‐5(Haw) pkc‐1(ok563);npr‐1(ad609) worms (2.5‐25 ng/µL) with the PF15E11.1::GFP co‐injection marker (25‐47.5 ng/µL).We used Gibson assembly cloning to generate the [Punc‐25::pkc‐1C::polycismCherry], [Pmod‐1::pkc‐1b(gf)::polycismCherry], and [Punc‐25::pitp‐1 (cDNA) polycismCherry] expression constructs. The three constructs share the polycistronic mCherry pPD95.75 expression vector that was amplified with 5′ CCACGATGCCTGTAGCAATGGCAAC 3′ and 5′ ATTTCATTTCCAAGTTGTTAGCGTATCCATCG 3' (~1.6 kb) and with 5′ GTTGCCATTGCTACAGGCATCGTGG 3' and 5' GCTGTCTCATCCTACTTTCACCTAG 3' (~2.8 kb).For the [Pmod‐1::pkc‐1b(gf)::polycismCherry] rescuing construct, we amplified mod‐1 promoter (~5.5 kb) with 5′ CGATGGATACGCTAACAACTTGGAAATGAAATGTGTTTTCTCGTAATCCCGTTGTACTTC 3′ and 5′ CATATTCATCGTATACCAAACAATCAAAACTCATTATATTTTAATTTTCTTTCACCGCATTGGCACCTGG 3' and also amplified pkc‐1b (~1.9 kb
) using wild‐type N2 cDNA as a template with 5′ CCAGGTGCCAATGCGGTGAAAGAAAATTAAAATATAATGAGTTTTGATTGTTTGGTATACGATGAATATG 3′ and 5′ CTAGGTGAAAGTAGGATGAGACAGCGTCGACTTAGTAGGTAAAATGCGGATTGATAAATGAAAAAC 3′ primer pair. We used Q5 Site‐Directed Mutagenesis Kit (NEB) to introduce the A160E gain‐of‐function mutation into pkc‐1 using 5′ AGACGTGGTGagATGCGACGGA 3′ and 5′ TTGTCGATCATTGAAAGCATTTG 3′ primers. The resulting expression construct was injected into glb‐5 (Haw); npr‐1 (ad609) worms (25‐100 ng/µL) with the PF15E11.1::GFP co‐injection marker (25‐100 ng/µL).To generate the GABAergic pitp‐1 rescue construct, we amplified the promoter region of unc‐25 (~2 kb) with 5′ cgatggatacgctaacaacttggaaatgaaatCATCATGCCCGCTGCTGTTATCAG 3′ and 5' GTAACAGGATACGGTATTCTTTGATCAGCATtatattttTTTTGGCGGTGAACTGAGCTTTTCC 3′ and pitp‐1 (~3.1 kb) amplified with 5′ GGAAAAGCTCAGTTCACCGCCAAAAaaaatataATGCTGATCAAAGAATACCGTATCCTGTTAC 3′ and 5′ ctaggtgaaagtaggatgagacagcgtcgacCTACCGACGATCTTTTTCATTTTCGAATTTTCC 3′. This construct was injected into pitp‐1 (tm 1500) III (Haw); npr‐1 (ad609) worms (0.1‐1 ng/µL) with the PF15E11.1::GFP co‐injection marker (50 ng/µL).To generate the GABAergic pkc‐1 rescue construct, we amplified the promoter region of unc‐25 (~2 kb) with 5′ cgatggatacgctaacaacttggaaatgaaatCATCATGCCCGCTGCTGTTATCAG 3′ and 5′ CAGATGATATTGTCCGACTACTGAAGAATTTCATtatattttTTTTGGCGGTGAACTGAGCTTTTCC 3' and also amplified pkc‐1c (~2.3 kb) with 5′ GGAAAAGCTCAGTTCACCGCCAAAAaaaatataATGAAATTCTTCAGTAGTCGGACAATATCATCTG and 5′ ctaggtgaaagtaggatgagacagcgtcgacTTAGTAGGTAAAATGCGGATTGATAAATGAAAAACC 3′. The resulted expression construct was injected into pitp‐1 (tm 1500); glb‐5 (Haw) pkc‐1 (ok653); npr‐1 (ad609) worms (12.5‐50 ng/µL) with the PF15E11.1::GFP co‐injection marker (12.5‐50 ng/µL) (Table 2).
TABLE 2
Primers used for deletion strains
Forward (F) and Reverse (R) sequences
pitp‐1(tm1500)
F: CATGGAGATTTCAGCCCGGACAACCC4534
5105R: GAGATCATACAATAACATCATTTTGCAGGGC
R: CTGAAAGCTGAGTCTCGATGCTCCAAACTG6033
pkc‐1(ok563)
F: GACTCGGCGTTGATTAATGAGCTTTCTCG
F1: GACCTGAAAGGATAGAAATGGTGATTGGCTTG
R: GATCAAACGCTATTCGGGTATTCGCAACG
dgk‐1(ok1462)
F: CCTCCGGATTATTCATCGTCTGTTCG
R: GACGTTGAACCACTCCTTGTGCTAG
R1: CTATGGCGAGAAGTCACACCACATTG
Primers used for deletion strainsImaging strains: To image pitp‐1 neurons localization we used the modified plasmid pEnter (95.75 MCS) GFP. We amplified the pitp‐1 promoter (~4.2 Kb) using N2 genomic DNA. The promoter was inserted with NheI and XbaI. The constructs were injected into pitp‐1(tm1500);glb‐5(Haw);npr‐1(ad609) worms. To validate the expression of pitp‐1 in the O2‐sensing neurons, we crossed this strain with the AX1846 strain, which expressed mCherry under the promoter region of glb‐5.
The Ca2+ imaging strains were generated as described previously.
,
The Pgcy‐37::YC2.60 extrachromosomal array was introduced to pitp‐1(tm1500);glb‐5(Haw);npr‐1(ad609) worms by crossing with the EVG159 parental strain.
Ca2+ imaging
We performed the Ca2+ imaging experiments as described in.
We recorded the Ca2+ responses at two frames per second, using an Olympus IX71S1F‐3‐5 inverted microscope equipped with UAPON40X Universal apochromatic water immersion objective (Olympus, Tokyo, Japan), DV2 2‐channel imager (Photometrics), Rolera EM‐C2 (Qimaging), and MetaMorph software (Molecular Devices). To immobilize worms, we glued them to agarose pads (2% agarose in M9) using Dermabond topical skin adhesive (Ethicon). These worms were trapped inside a 500 μm deep rectangular PDMS chamber (12 mm wide, 0.5 mm deep, and 17 mm long). In all of our imaging experiments, humidified gases were delivered to the microfluidic chambers using a PHD 2000 syringe pump (Harvard Apparatus, Holliston, MA, United States) at a flow rate of 0.5 ml per minute. Teflon valves, regulated by a ValveBank Controller (Automate Scientific, Berkeley, California, USA), were used to rapidly switch between gas mixtures. Image analysis was performed using custom‐written MATLAB software.
Aldicarb assays
We adapted the aldicarb assay protocol described in.
In brief, a day before the experiment, we prepared experimental NGM plates containing 1.5 mM of aldicarb (33386‐100MG Sigma Aldrich; stock solution: 50 mM in 70% ethanol). After 5 hours at RT, we seeded the plates with 100 µL of OP50 bacteria (OD600 ~ 0.6) and let them dry for 24 hours at RT. On these plates, we placed young adult hermaphrodites that were either exposed to hypoxia (1% O2) for 24 hours as described above or remained at 21% O2. We measured paralysis every 20 minutes for 2 hours. Paralysis is defined by the absence of movement when poked three times with an eyelash on the head and tail. The 1% and 21% O2‐aldicarb assays included five biological replicates, in which at least 120 worms were examined.
Statistical analysis
For comparison between two groups, we used an unpaired t test with Welch's correction. For comparison between more than two groups, when two parameters are explored (ie, the effect of O2 level and genetic background), we used two‐way ANOVA with Bonferroni posttest. For the aldicarb assays, we used the log‐rank (Mantel‐Cox) test. Data are presented as mean ± SEM (standard error of the mean) or as a Kaplan‐Meier curve (ie, in the case of the aldicarb experiments). All statistical analysis was performed using GraphPad Prism version 8.0.0 for Windows, GraphPad Software, San Diego, California USA, www.graphpad.com.
RESULTS
PITP‐1 is required for fast recovery of the bordering eating behavior after hypoxia
GLB‐5 accelerates the rate of bordering feeding behavior recovery after hypoxia.
,
To explore the underlying mechanism, we performed an EMS mutagenesis screen and isolated glb‐5(+); npr‐1(−) mutants that fail to display wild‐type bordering behavior (after 24 hours exposure at 1% O2, followed by 30 min recovery at 21% O2). To avoid isolating mutants that are generally defective in their O2‐sensing and/or bordering behavior, we verified that these mutants resume wild‐type bordering behavior after 4 hours recovery at 21% O2. We used single nucleotide polymorphism (SNP) mapping
to locate one of these mutations, heb2, to a 1.6 Mb genomic interval on chromosome III. By whole‐genome sequencing (Illumina), we identified 10 point mutations in this genomic region. Among these mutations, a G > A transition mutation disrupts a splice‐acceptor site in the pitp‐1 gene (Figure 1B). pitp‐1 encodes the sole class IIA phosphatidylinositol transfer protein (PITP) in C elegans,
sharing high sequence similarity with the humanNir2 protein.
To confirm the pitp‐1(heb2) phenotype, we tested the bordering recovery of a pitp‐1 deletion mutant, that is, pitp‐1(tm1500),
which lacks approximately 950 bp from the pitp‐1 genomic region. The bordering behavior of pitp‐1(heb2) and pitp‐1(tm1500) mutants was similar (Figure 1C), suggesting that pitp‐1(heb2) is nonfunctional. Since tracking the tm1500 allele is easier than heb2, we have continued working with pitp‐1(tm1500) allele (hereafter referred to as pitp‐1(−)). Finally, we confirmed the necessity of pitp‐1 to the rapid recovery of bordering behavior by generating transgenic pitp‐1(−); glb‐5(+); npr‐1(−) worms that express either the genomic or coding region of pitp‐1 under its promoter region (Figure 1D).
PITP‐1 is not generally required for GLB‐5‐dependent O2 responses
So far, all the characterized GLB‐5 suppressors have been found to affect the O2‐sensing pathway in the AQR, PQR, and URX neurons. This occurs by directly changing the function of the O2‐sensors GCY‐35 and GCY‐36
or by interfering with their transport to the dendritic endings of URX.
To explore the mechanism by which PITP‐1 regulates GLB‐5 activity, we first set out to determine whether it is co‐expressed with glb‐5 in the AQR, PQR, and URX neurons. For this, we generated transgenic worms co‐expressing a GFP and a mCherry fluorescent tags under the promoter regions of pitp‐1 and glb‐5, respectively. As previously reported, pitp‐1 is expressed in many neurons.
We observed a co‐localization of the GFP and mCherry fluorescence in the AQR, PQR, and URX neurons (Figure 2A), as well as in the BAG O2‐sensing neurons (the BAG neurons act reciprocally to the URX neurons and are activated at low O2 levels
), indicating that pitp‐1 is co‐expressed with glb‐5 in O2‐sensing neurons.
FIGURE 2
Expression of pitp‐1 in the O2‐sensing neurons AQR, PQR, URX, and BAG does not rescue the fast recovery of bordering behavior after hypoxia. A, Bright‐field and fluorescence images of transgenic worms co‐expressing Pglb‐5::glb‐5::mCherry and Ppitp‐1::pitp‐1::GFP. Arrows indicate the AQR, BAG, PQR, and URX neurons in the merge images. Scale bars: 25 μm. B, A bar graph presenting the recovery of bordering after 24 hours in hypoxia. Asterisks indicate significance for comparisons with glb‐5(+); npr‐1(−) animals after 24 hours at 1% O2. C, Speed measurements. The speed of worms was measured at 21% and 17% O2 in the presence of bacteria. Asterisks indicate significance for comparisons with glb‐5(+); npr‐1(−) worm speed at 17% O2. Data represent the average of at least six independent experiments. Two‐way ANOVA with Bonferroni posttest. ****P < .0001, NS, nonsignificant. Error bars represent SEM
Expression of pitp‐1 in the O2‐sensing neurons AQR, PQR, URX, and BAG does not rescue the fast recovery of bordering behavior after hypoxia. A, Bright‐field and fluorescence images of transgenic worms co‐expressing Pglb‐5::glb‐5::mCherry and Ppitp‐1::pitp‐1::GFP. Arrows indicate the AQR, BAG, PQR, and URX neurons in the merge images. Scale bars: 25 μm. B, A bar graph presenting the recovery of bordering after 24 hours in hypoxia. Asterisks indicate significance for comparisons with glb‐5(+); npr‐1(−) animals after 24 hours at 1% O2. C, Speed measurements. The speed of worms was measured at 21% and 17% O2 in the presence of bacteria. Asterisks indicate significance for comparisons with glb‐5(+); npr‐1(−) worm speed at 17% O2. Data represent the average of at least six independent experiments. Two‐way ANOVA with Bonferroni posttest. ****P < .0001, NS, nonsignificant. Error bars represent SEMNext, we asked whether pitp‐1 activity in these neurons is sufficient to rescue the fast bordering recovery after hypoxia. To answer this, we generated two transgenic pitp‐1(−); glb‐5(+); npr‐1(−) worm strains expressing the coding region of pitp‐1 just in AQR, PQR, and URXL/R (under the promoter region of gcy‐37) or in all six O2‐sensing neurons (i.e., URX L/R, AQR, PQR, and BAG L/R) under the promoter region of glb‐5.
,
The expression of pitp‐1, either in the four or the six neurons, did not restore fast bordering after hypoxia (Figure 2B), suggesting the pitp‐1 acts in other neurons to accelerate the recovery of feeding behavior after hypoxia.To further explore pitp‐1 function, we asked whether it is generally important for glb‐5 activity. To explore this, we performed speed assay experiments in which we exposed npr‐1(−), glb‐5(+); npr‐1(−), and pitp‐1(−); glb‐5(+); npr‐1(−) worms to a 21%‐17% O2 shift. We use this O2 shift as an indicator for glb‐5 activity because it induces a significant decrease in the foraging speed of glb‐5(+); npr‐1(−) worms, however, only a mild decrease in npr‐1(−) worms that bear a nonactive glb‐5.
,
As expected, the shift from 21% to 17% O2 induced a robust decrease in the speed of glb‐5(+); npr‐1(−) worms, and a mild decrease in npr‐1(−) worms (Figure 2C). The speed‐shift observed in pitp‐1(−); glb‐5(+); npr‐1(−) worms was similar to that seen in glb‐5(+); npr‐1(−) controls, suggesting that pitp‐1 is not required for glb‐5 activity in this behavioral paradigm.
O2‐evoked Ca2+ responses in URX in pitp‐1 mutants
To directly explore whether pitp‐1 is important for glb‐5‐mediated O2 responses, we performed Ca2+ imaging experiments. For this, we expressed the ratiometric YC2.60 Ca2+ sensor in the URX neurons (as described in
) of npr‐1(−), glb‐5(+);npr‐1(−), and pitp‐1(−); glb‐5(+); npr‐1(−) worms. These worm strains were exposed to 1% O2 for 24 hours or remained at 21% O2 as a control (i.e., naïve worms). We chose the 21% to 14% shift because it mimics the O2‐concentration change that the worms experience when entering the bacterial lawn border, where the O2 level is lower than the lawn center and its surroundings.
The 21% to 14% O2 shift elicited a significant drop in Ca2+ concentration in naïve worms of the three strains (Figure 3A‐C). After 24 hours exposure to hypoxia, the 21%‐14% O2 shift did not induce a Ca2+ drop in npr‐1 worms (Figure 3A). This result is in line with previous studies that show that a functional glb‐5 is essential for decreasing the sensitivity of the URX neurons after exposure to hypoxia.
By contrast, the shift from 21% to 14% O2 induced a significant decrease in the Ca2+ level of both glb‐5(+); npr‐1(−) and pitp‐1(−); glb‐5(+); npr‐1(−) worms (Figure 3B,C). The magnitude of the Ca2+ drop was similar between the two strains. Together, these results suggest that pitp‐1 is not required for glb‐5‐dependent O2 responses in the O2‐sensing neurons.
FIGURE 3
PITP‐1 is not required for glb‐5‐dependent O2 responses in the URX neurons. A‐C, Ca2+ imaging in animals expressing the Cameleon YC2.60 in the URX neurons. The responses were measured in naïve worms (grown in 21% O2, blue Ca2+ traces) or in worms exposed to 1% O2 (red Ca2+ traces). Asterisks indicate significant differences at times specified by bars between the Ca2+ levels, unpaired t test with Welch's correction. Gray shading indicates 14% O2. n = the number of worms imaged in each strain. Error bars (lighter shading) represent SEM
PITP‐1 is not required for glb‐5‐dependent O2 responses in the URX neurons. A‐C, Ca2+ imaging in animals expressing the Cameleon YC2.60 in the URX neurons. The responses were measured in naïve worms (grown in 21% O2, blue Ca2+ traces) or in worms exposed to 1% O2 (red Ca2+ traces). Asterisks indicate significant differences at times specified by bars between the Ca2+ levels, unpaired t test with Welch's correction. Gray shading indicates 14% O2. n = the number of worms imaged in each strain. Error bars (lighter shading) represent SEM
PITP‐1 roles in salt/benzaldehyde chemotaxis and recovery from hypoxia are distinct
Where does PITP‐1 act to facilitate fast recovery from hypoxia? A previous study by Iino and colleagues showed that PITP‐1 acts in the ASEL/R neurons to regulate salt (NaCl) chemotaxis/chemorepulsion and in the AWC and the ASH neurons to promote, respectively, olfaction and osmotic avoidance.
Therefore, we explored whether pitp‐1 acts in the ASE and the AWC neurons to facilitate fast recovery from hypoxia. We generated transgenic pitp‐1(‐); glb‐5(+); npr‐1(‐) worms expressing pitp‐1 under the flp‐6 promoter region (ASE neurons
) or under the ceh‐36 promoter region (ASE and AWC neurons
). Expression of pitp‐1 in just the ASE or the ASE/AWC neurons did not rescue the fast recovery of bordering behavior after 24 hours exposure to hypoxia (Figure 4A). By contrast, the expression of pitp‐1 in these neurons (just in ASE or in both ASE and AWC) restored the chemotaxis attraction to salt (Figure 4B), albeit to a lower level than wild type. Together, these results indicate that the place of activity of PITP‐1 in the control of salt chemotaxis is different from the place of its activity in the regulation of feeding behavior after hypoxia.
FIGURE 4
PITP‐1 functions in salt chemotaxis and recovery from hypoxia are distinct. A, A bar graph presenting the recovery of bordering after 24 hours in hypoxia. Asterisks indicate significance for comparisons with pitp‐1(−); glb‐5(+); npr‐1(−) worms after 24 hours at 1% O2. B, Bar graph showing the attraction of worms to sodium chloride (NaCl). Asterisks indicate significance for comparisons with pitp‐1(−); glb‐5(+); npr‐1(−) worms. Data represent the average of at least six independent experiments. Two‐way ANOVA with Bonferroni posttest. *P < .05, ****P < .0001. Error bars represent SEM
PITP‐1 functions in salt chemotaxis and recovery from hypoxia are distinct. A, A bar graph presenting the recovery of bordering after 24 hours in hypoxia. Asterisks indicate significance for comparisons with pitp‐1(−); glb‐5(+); npr‐1(−) worms after 24 hours at 1% O2. B, Bar graph showing the attraction of worms to sodium chloride (NaCl). Asterisks indicate significance for comparisons with pitp‐1(−); glb‐5(+); npr‐1(−) worms. Data represent the average of at least six independent experiments. Two‐way ANOVA with Bonferroni posttest. *P < .05, ****P < .0001. Error bars represent SEM
PITP‐1 acts in MOD‐1‐expressing neurons to promote fast recovery from hypoxia
To determine where pitp‐1 activity is required for the fast bordering recovery after hypoxia, we restricted the expression of pitp‐1 cDNA to a subset of neurons using the following promoters: cng‐3 [AFD, ASE, ASI, AWB, and AWC
], gpa‐14 [ADE, ALA, ASH, ASI, ASJ, ASK, AVA, CAN, DVA, PHA, PHB, PVQ, and RIA
], mod‐1 [AIA, AIB, AIY, AIZ, DD, RIC, RID, RIM, RME, and VD
,
,
], ocr‐2 [expressed in PHA, PHB, ASH, ADL, ADF and AWA
,
], tax‐2 [expressed in the AWC, AFD, ASE, ASG, ASJ, AQR, BAG, ASK, ASI, AWB, and PQR neurons
,
], and ttx‐7 [ADF, ADL, AFD, ASE, ASH, ASI, ASJ, ASK, AWB, AWC, RIA, and VNC
]. In addition, we used the rgef‐1 promoter to express pitp‐1 in all neurons
as a positive control. As expected, transgenic pitp‐1(−); glb‐5(+); npr‐1(−) worms expressing pitp‐1 under the rgef‐1 promoter region showed full rescue of bordering behavior after hypoxia (Figure 5). Intriguingly, the expression of pitp‐1 under the mod‐1 promoter region resulted in a significant rescue of the bordering behavior after hypoxia, albeit to a lesser degree than the rgef‐1 promoter (Figure 5). These results suggest that PITP‐1 acts in mod‐1‐expressing neurons to regulate the recovery of feeding behavior after hypoxia.
FIGURE 5
Expression of pitp‐1 under the promoter region of mod‐1 rescues the fast recovery of bordering behavior after hypoxia. A bar graph presenting the recovery of bordering after 24 hours in hypoxia in transgenic pitp‐1(−); glb‐5(+); npr‐1(−) worms expressing wild‐type pitp‐1(cDNA) under different neuron promoter regions. Asterisks indicate significance for comparisons with npr‐1(−) animals after 24 hours at 1% O2. Data represent the average of at least six independent experiments. Two‐way ANOVA with Bonferroni posttest. ****P < .0001, NS, nonsignificant. Error bars represent SEM
Expression of pitp‐1 under the promoter region of mod‐1 rescues the fast recovery of bordering behavior after hypoxia. A bar graph presenting the recovery of bordering after 24 hours in hypoxia in transgenic pitp‐1(−); glb‐5(+); npr‐1(−) worms expressing wild‐type pitp‐1(cDNA) under different neuron promoter regions. Asterisks indicate significance for comparisons with npr‐1(−) animals after 24 hours at 1% O2. Data represent the average of at least six independent experiments. Two‐way ANOVA with Bonferroni posttest. ****P < .0001, NS, nonsignificant. Error bars represent SEMGABAergic neurons appear to be involved in neuronal response to hypoxia
and food‐sensing in C elegans.
Since the mod‐1‐expressing neurons DD, RME, and VD are GABAergic,
we asked whether PITP‐1 acts in these neurons to facilitate fast bordering recovery after hypoxia. To explore this, we expressed pitp‐1 under the GABAergic‐specific promoter unc‐25.
in pitp‐1(−); glb‐5(+); npr‐1(−) worms and examined the recovery of bordering after hypoxia. The bordering recovery of transgenic GABAergic::pitp‐1 worms was similar to pitp‐1(−); glb‐5(+); npr‐1(−) mutants (Figure S1A), suggesting that PITP‐1 function in these GABAergic neurons is not sufficient to facilitate fast bordering recovery after hypoxia.
Impairment of PKC‐1 activity restores rapid bordering recovery after hypoxia in pitp‐1 mutants
A previous study showed that the function of PITP‐1 in salt chemotaxis plasticity is mediated, in part, by diacylglycerol (DAG) signaling.
A key target of DAG is the protein kinase C 1 (PKC‐1), which is predicted to require DAG for activation.
,
,
To explore the function of PKC‐1 in the PITP‐1‐dependent recovery from hypoxia, we first asked whether PKC‐1 itself is required for the fast recovery. To answer this, we generated glb‐5(+); npr‐1(−) worms bearing the pkc‐1(ok563) deletion allele (hereafter referred to as pkc‐1(−)), which removes ~1.3 Kb from the 5′ UTR of the pkc‐1B splice isoform (including the ATG translation initiation site).
The bordering recovery of glb‐5(+) pkc‐1(−); npr‐1(−) worms from hypoxia was similar to glb‐5(+); npr‐1(−) controls (P > .9999; Figure 6A), indicating that PKC‐1 is not required for the fast recovery of bordering behavior. We next introduced pkc‐1(ok563) into pitp‐1(−); glb‐5(+); npr‐1(−) worms and examined their bordering recovery after hypoxia. Intriguingly, pkc‐1 loss‐of‐function rescued the fast bordering recovery after hypoxia (Figure 6A), albeit to a lesser degree in comparison to glb‐5(+); npr‐1(−) controls. This result suggests that PKC‐1 activity inhibits the fast recovery from hypoxia in pitp‐1(−); glb‐5(+); npr‐1(−) worms.
FIGURE 6
PKC‐1 loss of function enhances the bordering recovery of pitp‐1(−); glb‐5(+); npr‐1(−) worms after hypoxia. A‐D, Bar graphs presenting the recovery of bordering after 24 hours in hypoxia. In (A), asterisks indicate significance for comparisons with pitp‐1(−); glb‐5(+) pkc‐1(−); npr‐1(−) worms after 24 hours at 1% O2. In (C), significance was calculated in PMA‐treated animals, each strain compared to the 21% O2 controls. In (D), asterisks indicate significance for comparisons with glb‐5(+); npr‐1(−) worms after 24 hours at 1% O2. All bordering experiments represent at least six biological repeats. For (A), (B), and (D), we used two‐way ANOVA with Bonferroni posttest. For (C), we used multiple t tests with Holm‐Sidak multiple comparison correction. **P < 0. 01, ***P < .001, ****P < .0001, NS, nonsignificant. Error bars represent SEM. E and F, Aldicarb paralysis assays. E, Aldicarb response of npr‐1(−), glb‐5(+), npr‐1(−), and pitp‐1(−); glb‐5(+); npr‐1(−) worms after exposure to 1% O2 (24 hours) or 21% O2 (control). Inset table, p values indicate significance for comparisons with glb‐5(+); npr‐1(−) worms at 1% and 21% O2; the npr‐1(−)/glb‐5(+); npr‐1(−) comparison in blue, the glb‐5(+); npr‐1(−)/pitp‐1(−); glb‐5(+); npr‐1(−) comparison in black. F, In addition to the strains explored in (E), we investigated the response of transgenic pitp‐1(−); glb‐5(+); npr‐1(−); heb39Ex[Pmod‐1::pitp‐1(cDNA)] worms to aldicarb. Inset table, P values indicate significance for comparisons with pitp‐1(−); glb‐5(+); npr‐1(−) worms at 1% and 21% O2; the npr‐1(−)/pitp‐1(−); glb‐5(+); npr‐1(−) comparison in blue, the glb‐5(+); npr‐1(−)/pitp‐1(−); glb‐5(+); npr‐1(−) comparison in red, and the pitp‐1(−); glb‐5(+); npr‐1(−); heb39Ex/pitp‐1(−); glb‐5(+); npr‐1(−) comparison in orange. Log‐rank (Mantel‐Cox) test. The aldicarb experiments represents five biological repeats
PKC‐1 loss of function enhances the bordering recovery of pitp‐1(−); glb‐5(+); npr‐1(−) worms after hypoxia. A‐D, Bar graphs presenting the recovery of bordering after 24 hours in hypoxia. In (A), asterisks indicate significance for comparisons with pitp‐1(−); glb‐5(+) pkc‐1(−); npr‐1(−) worms after 24 hours at 1% O2. In (C), significance was calculated in PMA‐treated animals, each strain compared to the 21% O2 controls. In (D), asterisks indicate significance for comparisons with glb‐5(+); npr‐1(−) worms after 24 hours at 1% O2. All bordering experiments represent at least six biological repeats. For (A), (B), and (D), we used two‐way ANOVA with Bonferroni posttest. For (C), we used multiple t tests with Holm‐Sidak multiple comparison correction. **P < 0. 01, ***P < .001, ****P < .0001, NS, nonsignificant. Error bars represent SEM. E and F, Aldicarb paralysis assays. E, Aldicarb response of npr‐1(−), glb‐5(+), npr‐1(−), and pitp‐1(−); glb‐5(+); npr‐1(−) worms after exposure to 1% O2 (24 hours) or 21% O2 (control). Inset table, p values indicate significance for comparisons with glb‐5(+); npr‐1(−) worms at 1% and 21% O2; the npr‐1(−)/glb‐5(+); npr‐1(−) comparison in blue, the glb‐5(+); npr‐1(−)/pitp‐1(−); glb‐5(+); npr‐1(−) comparison in black. F, In addition to the strains explored in (E), we investigated the response of transgenic pitp‐1(−); glb‐5(+); npr‐1(−); heb39Ex[Pmod‐1::pitp‐1(cDNA)] worms to aldicarb. Inset table, P values indicate significance for comparisons with pitp‐1(−); glb‐5(+); npr‐1(−) worms at 1% and 21% O2; the npr‐1(−)/pitp‐1(−); glb‐5(+); npr‐1(−) comparison in blue, the glb‐5(+); npr‐1(−)/pitp‐1(−); glb‐5(+); npr‐1(−) comparison in red, and the pitp‐1(−); glb‐5(+); npr‐1(−); heb39Ex/pitp‐1(−); glb‐5(+); npr‐1(−) comparison in orange. Log‐rank (Mantel‐Cox) test. The aldicarb experiments represents five biological repeatsPITP‐1 controls the recovery from hypoxia through its activity in mod‐1‐expressing neurons (Figure 5). Therefore, we asked whether pkc‐1 acts in these neurons to inhibit the fast bordering recovery of pitp‐1(−);glb‐5(+) pkc‐1(−); npr‐1(−) worms after hypoxia. To explore this, we generated transgenic pitp‐1(−); glb‐5(+) pkc‐1(−); npr‐1(−) worms expressing pkc‐1(wt) under the promoter region of mod‐1. Moreover, we generated two additional rescuing strains. One expresses pkc‐1(wt) in all neurons, under the promoter region of rgef‐1, and thus, serves as a positive control for neuronal pkc‐1 activity. The second expresses pkc‐1(wt) just in GABAergic neurons, under the promoter region of unc‐25. Expression of pkc‐1(wt) in all neurons or only in the mod‐1‐expressing neurons resulted in a significant and a similar degree of bordering‐inhibition (P = .6755), suggesting that no other neurons, besides the mod‐1‐neurons, are important for this inhibitory activity of PKC‐1. Unexpectedly, restoring PKC‐1 activity under the unc‐25 promoter resulted in a small, yet, a significant decrease in the recovery of bordering after hypoxia (Figure S1B), suggesting that the activity of PKC‐1 in one or more of the GABAergic neurons DD, RME, and VD contributes to the inhibition of bordering after hypoxia. Together, these results indicate that PKC‐1 activity in the mod‐1‐expressing neurons decreases bordering recovery after hypoxia. Furthermore, they suggest that the function of PITP‐1 is to attenuate PKC‐1 activity in these neurons. Finally, although pitp‐1 expression in DD, RME, and VD is not sufficient for rescuing the fast recovery of bordering after hypoxia, the above pkc‐1‐rescuing experiments suggest that one or more of these neurons play a role in the recovery of bordering after hypoxia.To further explore PKC‐1 function, we asked whether constitutive PKC‐1 activity in the mod‐1‐expressing neurons would interfere with the recovery of bordering after hypoxia. For this, we overexpressed a gain of function mutation of pkc‐1 in the mod‐1‐expressing neurons of glb‐5(+); npr‐1(−) worms. This pkc‐1 mutation, hereafter referred to as pkc‐1(gf), substitutes alanine 160 for glutamate in the pseudosubstrate domain of PKC‐1, resulting in constitutive enzymatic activity.
The bordering index of transgenic pkc‐1(gf) worms after hypoxia was significantly lower compared to the parental glb‐5(+);npr‐1(−) worms (~54% compared to ~96%, after hypoxia, Figure 6B). In fact, the bordering index of transgenic pkc‐1(gf) worms was low even in control worms that did not experience hypoxia, indicating that constitutive PKC‐1 activity in the mod‐1‐expressing neurons inhibits the bordering behavior of glb‐5(+);npr‐1(−) worms irrespective of the exposure to hypoxia. These data support the hypothesis that excessive PKC‐1 activity in the mod‐1‐expressing neurons suppresses bordering after hypoxia.
PMA inhibits fast recovery of bordering behavior after hypoxia
How does PITP‐1 attenuate PKC‐1 activity? Since PKC‐1 is thought to be activated by DAG.
and PITP‐1 can potentially affect the level of DAG,
, we hypothesized that excess of DAG would attenuate the recovery of bordering behavior of glb‐5(+); npr‐1(−) worms. To test this, we put glb‐5(+); npr‐1(−) and npr‐1(−) worms on NGM‐plates containing PMA, a DAG analog that binds and activates PKC,
and measured the recovery of bordering behavior after hypoxia. In addition, we measured the effect of PMA on the bordering behavior of control worms that were not exposed to hypoxia and in worms exposed to 0.002% DMSO (the PMA vehicle). PMA significantly decreased the bordering recovery of both glb‐5(+); npr‐1(−) and npr‐1(−) worms after hypoxia (Figure 6C); however, it did not affect the bordering behavior of control worms.To further explore whether increased DAG signaling can inhibit the recovery of bordering behavior after hypoxia, we introduced the dgk‐1(ok1462) null mutation (hereafter referred to as dgk‐1(−)) into the genetic background of both glb‐5(+); npr‐1(−) and pitp‐1(−); glb‐5(+); npr‐1(−) worms. The dgk‐1 gene encodes a diacylglycerol (DAG) kinase,
which decreases DAG level by converting it to phosphatidic acid (PA).
,
,
Therefore, DAG signaling is increased in dgk‐1 mutants. DGK‐1 was not important for the bordering behavior of control worms that did not experience hypoxia (Figure 6D). However, the dgk‐1 mutation significantly decreased the bordering recovery of glb‐5(+); npr‐1(−) worms after hypoxia. Together, these results suggest that excess of DAG attenuates the recovery of bordering behavior after hypoxia. Moreover, this inhibitory effect is specific to the recovery period and does not apply to worms that were not exposed to hypoxia.
PITP‐1 effect on aldicarb resistance
So far, our results are consistent with a working model in which PITP‐1 restricts PKC‐1 activity during the recovery of worms from hypoxia by limiting DAG availability. Previous studies show that increased PKC‐1 activity and exogenous PMA sensitize worms to the paralytic effect of the acetylcholinesterase inhibitor aldicarb.
,
,
Therefore, we hypothesized that pitp‐1 loss‐of‐function would result in increased sensitivity of worms to aldicarb after hypoxia. To test this, we measured aldicarb‐induced paralysis in npr‐1(−), glb‐5(+); npr‐1(−), and pitp‐1(−); glb‐5(+); npr‐1(−) worms after exposure to hypoxia or in worms that were not exposed to hypoxia (as a control). pitp‐1(−); glb‐5(+); npr‐1(−) worms were significantly more sensitive to aldicarb than glb‐5(+); npr‐1(−) worms in both experimental conditions (Figure 6E, P = .0020, and P = .0008, after hypoxia and control conditions, respectively), suggesting that pitp‐1 is generally important for restricting neurotransmission in C elegans. Moreover, npr‐1(−) worms were significantly more sensitive to aldicarb than glb‐5(+); npr‐1(−) animals (P < .0001 for both experimental conditions). Notably, the exposure to hypoxia increased the aldicarb resistant of the three strain; Hazard ratios (after 24 hours exposure to hypoxia vs controls) were npr‐1(−) [0.4564]; glb‐5(+); npr‐1(−) [0.4095]; and pitp‐1(−); glb‐5(+); npr‐1(−) [0.3958], suggesting that hypoxia attenuates neurotransmitter release in a glb‐5 and pitp‐1 independent manner.Our data suggest that PITP‐1 acts in mod‐1‐expressing neurons to facilitate the fast recovery of bordering after hypoxia (Figure 5). Therefore, we hypothesized that restoring pitp‐1 function in these neurons would rescue pitp‐1(−); glb‐5(+); npr‐1(−) worms from the hypersensitivity to aldicarb. To test this, we repeated the above aldicarb experiments with transgenic Pmod‐1::pitp‐1(cDNA) worms (these worms display rescued bordering phenotype after hypoxia, Figure 5). As predicted, Pmod‐1::pitp‐1(cDNA) animals were significantly more resistant to aldicarb compared with their pitp‐1(−); glb‐5(+); npr‐1(−) parental strain, and this was true for both experimental conditions (i.e., after 24 hours at 1% O2 and in control worms that did not experience hypoxia) (Figure 6F). Intriguingly, the resistance of Pmod‐1::pitp‐1(cDNA) worms to aldicarb was significantly higher than of glb‐5(+); npr‐1(−) animals (P < .0001 for 21% O2‐aldicarb assays and P = .0002 for the 1% O2 assays). Since resistance to aldicarb can arise from impaired pre‐ or postsynaptic activity,
the hyper‐resistance of transgenic Pmod‐1::pitp‐1 worms to aldicarb may reflect the consequence of over‐PITP‐1 activity in these neurons, which decreases synaptic transmission to a suboptimal level. In conclusion, this set of experiments further supports the hypothesis that PITP‐1 facilitates the fast recovery of bordering after hypoxia by restricting neurotransmission in mod‐1‐expressing neurons.
DISCUSSION
PITP‐1 acts downstream to O2‐sensing neurons to facilitate fast bordering recovery from hypoxia
We previously isolated the pdl‐1(db508), gcy‐35(heb1), and gcy‐36(heb3) mutations that suppress the recovery of bordering eating behavior after hypoxia in glb‐5(+); npr‐1(−) worms.
,
These mutations directly interfere with the O2‐sensing machinery in the AQR, PQR, and URX neurons. The pdl‐1 mutation prevents the targeting of the soluble guanylate cyclases GCY‐35 O2‐sensor to URX endings, while the gcy‐35/gcy‐36 mutations make GCY‐35 and GCY‐36 insensitive to GLB‐5. As a result, pdl‐1(db508), gcy‐35(heb1), and gcy‐36(heb3) mutants are generally defective for glb‐5‐dependent‐O2 responses. That is, they do not display a robust decrease in speed when shifted from 21% to 17% O2 and display high URXCa2+ level at 14% and 17% O2 after hypoxia.
,
Notably, the expression of wild‐type pdl‐1, gcy‐35, and gcy‐36 in the AQR, PQR, and URX neurons of pdl‐1(db508), gcy‐35(heb1), and gcy‐36(heb3) mutants, respectively, rescues both the bordering behavior after hypoxia and the slowing response phenotypes, indicating that the activities of these genes in AQR, PQR, and URX are both necessary and sufficient for GLB‐5‐dependent O2 responses.By contrast, pitp‐1 mutants display a robust slowing response when shifted from 21% to 17% O2 and a sharp drop in URXCa2+ level upon a 21% to 14% O2 transition, after hypoxia (Figures 2C and 3C, respectively). Moreover, the recovery of bordering behavior after hypoxia cannot be rescued by restoring wild‐type PITP‐1 activity in AQR, PQR, and URX or even by expressing pitp‐1 under the promoter region of glb‐5 (Figure 2B), which is expressed in additional neurons, including ADF, ASG, and BAG.
Together, these results indicate that PITP‐1 is not required for the function of GLB‐5 in O2‐sensing per se, and suggest that it acts in other neurons to facilitate the fast recovery from hypoxia. In this respect, it is worth emphasizing that our previous studies suggest that the recovery of bordering behavior after hypoxia depends on the desensitization of the URX neurons by GLB‐5.
The results of this study imply that URX‐desensitization is not sufficient for the fast recovery phenotype and that downstream neurosignaling pathways are also required.Indeed, our results suggest that the recovery process is mediated by PITP‐1‐dependent neurosignaling in mod‐1‐expressing neurons. Interestingly, some of the mod‐1‐expression neurons are involved in dietary choice behavior in C. elegans and have chemical‐synapse connections with the GLB‐5‐expressing neurons. For example, the AIY interneurons, which have chemical synapses with BAG,
appear to be important for finding high‐quality food in diverse environments
and for regulating food‐leaving behavior.
Moreover, a recent paper suggests that the transcription factor DAF‐16 regulates salt avoidance by controlling neurotransmission between ASER, AIA, and AIY
; the same neurons in which PITP‐1 acts to regulate salt chemotaxis. Intriguingly, DAF‐16 regulates the aversive response to pheromone signals,
which could be affected by the O2 concentration that the worm experience
and potentially change the attraction/avoidance of worms to food. In the future, it will be exciting to explore whether DAF‐16 controls the recovery of bordering behavior after hypoxia and whether it affects the functions of PITP‐1 in both salt and recovery from hypoxia responses.
PITP‐1 functions in salt chemotaxis and recovery of bordering after hypoxia
A previous study by Iino and colleagues showed that PITP‐1 acts in the ASEL/R neurons to facilitate salt (NaCl) chemotaxis.
Intriguingly, pitp‐1‐loss‐of‐function did not affect the Ca2+ response to salt in ASER, suggesting that PITP‐1 is not important for salt sensing per se but rather for neurotransmission from ASER. Indeed, salt‐induced Ca2+ responses in the AIB interneurons, which are directly connected through chemical synapses to the ASER neurons, were significantly decreased in pitp‐1 mutants, further supporting the function of PITP‐1 in neurotransmission. Our results show that PITP‐1 function in the ASEL/R neurons is not important for the fast recovery of bordering behavior after hypoxia (Figure 4A), suggesting that PITP‐1 roles in salt chemotaxis and recovery of eating behavior after hypoxia are distinct. In fact, we show that the expression of wild‐type pitp‐1 under the promoter region of mod‐1 results in near‐to‐wild‐type recovery of the bordering behavior after hypoxia (Figure 5). Intriguingly, one of the interneurons in which mod‐1 is expressed is AIB (36). In future studies, we will investigate the effect of PITP‐1 on O2‐evoked‐Ca2+ responses (after hypoxia) in AIB and other interneurons that expresses mod‐1 (e.g., AIY).
Is PITP‐1 the “GLB‐5” of mod‐1‐expressing neuron(s)?
Our previous studies suggest a model in which GLB‐5 accelerates the recovery of bordering behavior after hypoxia by decreasing the sensitivity of URX neurons to O2.
In this way, the URX neurons are fine‐tuned to respond to subtle changes in the ambient O2 level. For example, the 21% to ~14% shift that the worm experience when it re‐enters the bacterial lawn after leaving it in hypoxia. Here, we would like to extend this model and propose that PITP‐1’s function in the mod‐1‐expression neurons is analogous to GLB‐5 in the O2‐sensing neurons. That is, PITP‐1 acts to decrease PKC‐1 activity in mod‐1‐expression neurons, and thus, prevent excessive neurotransmission that attenuates the recovery from hypoxia. Below, we present the experimental evidence and logic supporting this working model (Figure 7).
FIGURE 7
An illustrated working model explaining the function of PITP‐1 in the mod‐1‐expressing neuron(s) following the exposure of worms to 21% O2 after 24 hours at 1% O2. The figure describes the recovery of wild‐type (WT) animals (i.e., glb‐5(+);npr‐1(−) worms) vs pitp‐1(−); glb‐5(+); npr‐1(−) mutants; left of the dash line are the mod‐1 neurons of WT animals and right of the dash line are the mutant neurons. In WT, PITP‐1 inhibits PKC‐1 by limiting DAG availability. It does so by limiting PIP2 synthesis and PtdOH dephosphorylation. As a result, PKC‐1 is overactivated after hypoxia, a situation that triggers the release of a yet‐to‐be‐identified neurotransmitter(s)/neuropeptide(s), which inhibits the recovery of bordering eating behavior. The larger fonts and arrows, the pitp‐1‐mutants panel, indicate excessive substrate concentration and PKC‐1 activity
An illustrated working model explaining the function of PITP‐1 in the mod‐1‐expressing neuron(s) following the exposure of worms to 21% O2 after 24 hours at 1% O2. The figure describes the recovery of wild‐type (WT) animals (i.e., glb‐5(+);npr‐1(−) worms) vs pitp‐1(−); glb‐5(+); npr‐1(−) mutants; left of the dash line are the mod‐1 neurons of WT animals and right of the dash line are the mutant neurons. In WT, PITP‐1 inhibits PKC‐1 by limiting DAG availability. It does so by limiting PIP2 synthesis and PtdOH dephosphorylation. As a result, PKC‐1 is overactivated after hypoxia, a situation that triggers the release of a yet‐to‐be‐identified neurotransmitter(s)/neuropeptide(s), which inhibits the recovery of bordering eating behavior. The larger fonts and arrows, the pitp‐1‐mutants panel, indicate excessive substrate concentration and PKC‐1 activity(1)
pkc‐1 loss‐of‐function rescues the recovery phenotype of pitp‐1 mutants (Figure 6A). (2) Restoring the activity of wild‐type PKC‐1 in the mod‐1‐expressing neurons of pitp‐1(−); glb‐5(+) pkc‐1(−); npr‐1(−) worms attenuates the recovery of bordering behavior after hypoxia (Figure 6A). Moreover, expression of constitutively active PKC‐1 in these neurons (pkc‐1(gf)) significantly inhibits the bordering behavior of the worms (Figure 6B). (3) PKC‐1 activation increases worms' sensitivity to the acetylcholinesterase inhibitor aldicarb. Indeed, pitp‐1 mutants are more sensitive to aldicarb compared with their parental glb‐5(+); npr‐1(−) strain (Figure 6E). Moreover, rescuing PITP‐1 function in the mod‐1‐expression neurons of pitp‐1 mutants increases the resistance to aldicarb (Figure 6F). Together, these observations support the hypothesis that PITP‐1 facilitates the fast recovery of bordering after hypoxia by decreasing PKC‐1 activity in the mod‐1‐expressing neurons.PKC‐1 is thought to be activated by diacylglycerol (DAG), which can be generated by the following processes: (a) Through the cleavage of phosphatidylinositol 4,5‐bisphosphate (PIP2) to DAG and inositol 1,4,5‐trisphosphate (IP3) by phospholipase Cβ (PLCβ)
; (b) Through the dephosphorylation of phosphatidic acid (PtdOH) to DAG by PtdOH phosphatase.
The C. elegans PITP‐1 is a member of the class II phosphatidylinositol transfer proteins, which can control DAG availability by binding & shuttling phosphatidylinositol‐4‐phosphate (PIP2 precursor
) and PtdOH.
,
Therefore, we suggest that PITP‐1 decreases the activity of PKC‐1 by restricting DAG availability.Two observations support this claim. First, exogenous PMA (DAG analog) decreases the bordering recovery of wild‐type glb‐5(+); npr‐1(−) worms after hypoxia (Figure 6C). Second, dgk‐1 loss of function, which increases DAG level, also decreases the recovery of glb‐5(+); npr‐1(−) worms from hypoxia. PKC‐1 regulates the secretion of neuropeptides
and may function in neurotransmitter release.
We hypothesize that excessive secretion of a yet to be identified neuropeptide(s) and/or neurotransmitter(s) inhibit(s) the recovery of bordering behavior after hypoxia.In this respect it is interesting to mention that amphetamine‐induced anorexia in the rat is associated with the expression of several PKC isotypes,
including the eta isotype that is structurally similar to the C. elegansPKC‐1.
The anorectic effect is decreased by PKC alpha‐antisense knockdown, suggesting that PKC suppresses food consumption.
Finally, the amphetamine treatment decreases the level of neuropeptide Y (NPY) mRNA in the rathypothalamus.
This is interesting because NPY is known to regulate food‐consumption and is the ligand of the NPY receptor that resembles the C. elegans NPR‐1, which is also involved in eating‐behavior control.
In conclusion, despite the significant difference in the complexity of controlling eating behavior between the C. elegans worm and the rat and other mammals, we would like to suggest (with due caution) that the link between PKC, neuropeptides, and suppression of eating behavior may be ancient and preserved in evolution. In this context, it would be fascinating to investigate whether orthologs of PITP‐1 in the rat are involved in the suppression of eating behavior due to amphetamines and to explore whether mammalian PITP and PKC are involved in high‐altitude anorexia.
CONFLICT OF INTEREST
The authors declare that they have no conflict of interest.
AUTHOR CONTRIBUTIONS
Z. Abergel, M. Shaked, V. Shukla, and E. Gross designed and performed the experiments, contributed new reagents, analyzed the data, and wrote the paper; Z.X. Wu Contributed to paper writing; all authors read and approved the final manuscript.Fig S1Click here for additional data file.
Authors: Lesley Tilleman; Francesca Germani; Sasha De Henau; Eva Geuens; David Hoogewijs; Bart P Braeckman; Jacques R Vanfleteren; Luc Moens; Sylvia Dewilde Journal: IUBMB Life Date: 2011-03 Impact factor: 3.885
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5
FIGURE 6
FIGURE 7