Perilipin-related protein regulates lipid metabolism in C. elegans

Sequence analysis

Perilipin orthologues and W01A8.1 sequences were extracted from UniProt, NCBI and OMA (omabrowser.org) databases. Chordate and nematode sequences were aligned separately using the T-Coffee algorithm (Notredame, Higgins & Heringa, 2000) (server tcoffee.crg.cat) and submitted to PSI-BLAST (Altschul et al., 1997) (E-value inclusion threshold <10⁻³, 5 iterations) and HHpred (Remmert et al., 2011; Biegert & Soding, 2008) searches as implemented in MPItoolkit (toolkit.tuebingen.mpg.de). Repeat detection used HHrepID module in MPItoolkit. Alignments were displayed and analyzed in Jalview app (www.jalview.org).

Strains, transgenic lines and genome editing

Wild type animals, N2 (var. Bristol), were used unless otherwise noted and all strains were maintained as described (Brenner, 1974). Transgenic lines were prepared using microinjections into gonads of young adult N2 hermaphrodites as described (Tabara et al., 1999; Timmons, Court & Fire, 2001; Vohanka et al., 2010). All injections also included mCherry co-injection markers: pCFJ90, pCJ104 and pGH8 (Dickinson et al., 2013).

To create mutants, we employed CRISPR/Cas9 system as described (Dickinson et al., 2013). The following plasmids were constructed: pCK001 targeting the sgRNA (+323) to the second exon of the W01A8.1 gene (forward primer #7992), and pCK023 targeting the sgRNA (+1,372) to the sixth exon (forward primer #8078). The reverse primer was #7993. A scheme of known expressed isoforms listed in WormBase WS246 and the strategy for the disruption of W01A8.1 gene is shown in Figs. S1 and S2. Primers used in this study are listed in Table S1.

The following transgenic lines regulated by W01A8.1 natural promoter were prepared: W01A8.1a/c::gfp and W01A8.1b::gfp (containing the whole coding sequence of isoforms a and b). W01A8.1 isoforms a and c have identical 3′ ends which both could be expressed from W01A8.1a/c::gfp. This construct also includes complete untagged isoform b. The GFP-tagged isoform a (plasmid pCK28 {P_W01A8.1::W01A8.1(a)synth::gfp::unc-54 3′ UTR}) was constructed by synthesizing the W01A8.1a sequence with modified codons to allow protection from CRISPR/Cas9 targeted sgRNA and prepared as a GeneArt^® Strings™ DNA Fragment from Invitrogen (Invitrogen, Carlsbad, California, USA) and cloned using GeneArt^® Seamless Cloning System (Invitrogen) into pPD95.75(NeoR). Schemes for isoforms expressed from W01A8.1 gene and preparation of GFP tagged transgenes are given in Figs. S1 and S2.

Human PLIN2 and PLIN3 were cloned from a collection of anonymous unmarked samples (PLIN2), and from human peripheral lymphocytes (PLIN3) donated by a volunteer with a written consent in compliance with the legislation of the Czech Republic and European Union (Act No 372/2011 of 11. 11. 2011 on Health Care Services, Coll., Paragraph 81, section 1a and section 4a, which is in accordance with the declaration of Helsinki) and was approved by the Ethics Committee of the First Faculty of Medicine, Charles University in Prague (Ref. No. MZ13-UK1LF-KostrouchZdenek). Human PLIN1 optimized for C. elegans was prepared as a synthetic sequence requested as a GeneArt^® Strings™ DNA Fragment from Invitrogen™.

Transgenic lines expressing human PLIN1, PLIN2, PLIN3 tagged by GFP under W01A8.1 natural promoter were prepared using N2 animals and animals with disrupted W01A8.1. Primers used for cloning PLIN2 and PLIN3 are listed in Table S1.

Downregulation of gene expression by RNA interference

Downregulation of W01A8.1 expression used the RNAi protocol of injection of dsRNA into gonads of young adult hermaphrodites as well as RNAi through feeding animals bacteria producing dsRNA as previously described (Tabara et al., 1999; Timmons, Court & Fire, 2001; Vohanka et al., 2010).

Injection RNAi protocol

Double stranded RNA (dsRNA) was prepared for injection by in vitro transcription reactions (SP6/T7 Riboprobe^® in vitro Transcription Systems; Promega, Madison, Wisconsin, USA) from opposing promoters and subsequent annealing of each single stranded RNA (ssRNA) product prior to injection. For RNAi directed against W01A8.1, BamHI or ApaI linearized pCK014 plasmid preparations were used in separate reactions to generate complementary ssRNA. After linearization, the DNA was phenol-chloroform extracted and ethanol precipitated. BamHI linearized DNA was transcribed using T7 RNA Polymerase while ApaI linearized DNA with SP6 RNA Polymerase. After in vitro transcription (∼2 h), equal volumes of sense and antisense RNA were incubated at 75 °C for 10 min and then cooled at room temperature for 30 min. Control RNAi was prepared from the promoter region of nhr-60 as previously described (Simeckova et al., 2007). The dsRNA concentration was measured using a UV spectrophotometer and ∼1 µg/µl was used for injections.

Feeding RNAi protocol

Nematode Growth Medium (NGM) agar plates were prepared according to standard protocols and were supplemented with Ampicillin (100 µ g/ml final concentration) and isopropyl β-D-1-thiogalactopyranoside (IPTG) (1.5 mM final concentration). E. coli strain HT115 was transformed with pCK015 and control L4440 vector. After transformation, a single colony from each was used to inoculate LB medium with Ampicillin (100 µg/ml final concentration). The culture was grown to OD₆₀₀ ≈ 1.0; 900 µl of culture was poured onto NGM agar plates to completely cover the surface and 750 µl of the suspension was removed to leave 150 µl of the suspension on the plates. The bacteria were allowed to grow and were induced overnight at room temperate (∼22 °C).

Fecundity and brood size assay

Fecundity measurement following RNAi (injection method) was conducted using a total of 50 young adult worms (25 control and 25 inhibited by RNAi specific for W01A8.1). Progeny was counted 24 h and again 48 h after injections. Brood size assay was performed for W01A8.1 disrupted animals and controls (n = 15 for each group). The progeny was determined during 6 days. The experiments were conducted at room temperature ∼22 °C.

Fecundity measurement after RNAi using feeding protocol was performed over two generations to maximize the effect of knockdown. For this, a semi-synchronized population was isolated using standard WormBook (http://www.wormbook.org/) bleaching protocol. Hatched L1 stage worms were placed on NGM agar RNAi (W01A8.1 specific and control) plates at ∼22 °C. Small, synchronized populations, of parents (P0) were transferred to fresh RNAi plates and allowed to lay progeny (F1). F1 generation animals were transferred to new RNAi plates and F2 generation was scored for a total of 21 F1 parents in each group. The experiment was repeated twice to confirm the results.

RNA isolation and cDNA synthesis

Total RNA was extracted as previously described (Vohanka et al., 2010). Briefly, C. elegans washed, pelleted and re-suspended in re-suspension buffer with proteinase K. After lysis TRIzol^® Reagent (Invitrogen) was added to the mixture and the standard manufacturer’s protocol was followed to obtain total RNA. Samples were then treated with DNaseI (New England Biolabs, Ipswich, Massachusetts, USA) and again TRIzol-chloroform extracted (Invitrogen) to obtain DNA free total RNA.

Human total RNA was also extracted using TRIzol^® Reagent (Invitrogen) from peripheral blood lymphocytes and fat tissue. cDNA was prepared from total RNA by reverse transcription using standard protocols for the SuperScript^® III First-Strand Synthesis System (Invitrogen) and oligo(dT) priming.

Transcript quantification

Quantitative polymerase chain reaction (qPCR) was performed with cDNA prepared from total RNA isolated as described above, using the Roche Universal Probe Library technique (Hoffmann-La Roche, Basel, Switzerland). Primers and probes for determination of number of transcripts of W01A8.1 are given in Table S1. Levels of W01A8.1 expression were normalized against ama-1.

Single worm PCR

Single animals were placed into 5 µl of worm lysis buffer (10 mM Tris–HCl pH 8.3, 50 mM KCl, 2.5 mM MgCl2, 0.45% NP-40, 0.45% Tween 20, 0.01% Gelatin and 500 µg/ml fresh proteinase K) in a PCR tube. Animals were frozen at −80 °C for 5 min before placing the tube into a thermal cycler and run under the following conditions: heat to 60 °C for 60 min followed by inactivation of proteinase K by heating to 95 °C for 20 min. Post-lysis, a PCR reaction mix (45 µl) targeting the template of choice was added and cycled for ∼35 times with Q5^® Hot Start DNA polymerase (New England Biolabs).

LipidTox staining

The lipid staining protocol was done as described (O’Rourke et al., 2009) with modifications. Approximately 200–500 animals were harvested from NGM plates with 1X PBS and washed several times to remove E. coli and pelleted at 1,500 × g. To the pellet, 500 µl 2X MRWB (160 mM KCl, 40 mM NaCl, 14 mM Na₂EGTA, 1 mM Spermidine 3HCl, 0.4 mM Spermine, 30 mM NaPIPES pH 7.4, 0.2% beta-ME) and 100 µl 20% paraformaldehyde were added and the volume was adjusted up to 1 ml with 1X PBS. Inverting the tube several times mixed the worms in solution after which it was allowed to fix for ∼60 min at room temperature with gentle shaking.

After fixation, animals were pelleted at 1,500 × g and washed 3 times with 1 ml Tris–HCl buffer (100 mM, pH 7.4). After the third wash, the supernatant was discarded down to 100 µl and 650 µl of Tris-HCL buffer was added followed by 250 µl of fresh/frozen reduction buffer (100 mM Tris-Cl pH 7.4, 40 mM DTT). Worms were then left shaking for ∼30 min at room temperature. After reduction, worms were washed 3 times in 1X PBS. After the final PBS wash, the volume was brought up to 500 µl and then 500 µl of LipidTox (Red) (1:500 dilution) (Invitrogen) was added to make a final volume of 1 ml. The final concentration of 1:1,000 dilution of LipidTox was used. The worms were left in the dark for at least 60 min with shaking before viewing.

Microinjections

Microinjections of plasmids, DNA amplicons or dsRNA into gonads of young adult hermaphrodites were done using an Olympus IX70 microscope equipped with a Narishige microinjection system (Olympus, Tokyo, Japan).

Microscopy

Fluorescence microscopy and Nomarski optics microscopy were done using an Olympus BX60 microscope equipped with DP30BW CD camera (Olympus, Tokyo, Japan).

Coherent Anti-Stokes Raman Scattering microscopy (CARS)

For label-free CARS imaging the worms were anesthetized with levamisole and placed on a cover glass with a thin layer of 2% agarose. The CARS images were acquired with a Leica TCS SP8 CARS system (Leica Microsystems, Mannheim, Germany) consisting of a TCS SP8 confocal microscope combined with a picoEmerald laser (APE, Berlin, Germany) offering a fixed Stokes laser line of 1064.5 nm and a tuneable Pump line from an optical parametric oscillator (780 nm–940 nm). A HC PL IRAPO 40x water immersion objective was used for the imaging and CARS signal was detected with a non-descanned photon multiplier tube detector at the transmitted light side. For imaging of CH₂ vibration with Raman shift of 2,868 cm⁻¹ pump wavelength of 815.5 nm was used. The CARS signal was selected with a CARS2000 filter cube placed in front of the detector. Adult animals and embryos were completely scanned and recorded as stacks of focal planes. Recordings for quantitative analysis were done at fixed settings for mutant and wild type.

Image analysis

Single focal planes (containing the highest number of CARS positive structures) from stacks of representative embryos and adult hermaphrodites were selected and analyzed using ImageJ computer program (http://imagej.nih.gov/ij/). Analysis for embryos was done on images of seven different embryos (seven mutant and seven wild type embryos) inside gravid hermaphrodites (only one and two cell early embryonic stages were chosen for comparison). Analysis of adult somatic tissue was performed on distal body region and we compared five different mutant adults with five different wild type adults. The images selected for analysis have been provided as supplement.

Automatic particle counting feature of ImageJ program was used for determination of the number and area of CARS positive structures with manual thresholding as described on http://imagej.net/Particle_Analysis. Image area required for analysis was first selected (the area outside the selected zone was cleared) then the image was converted to an 8-bit scale. Manual threshold was applied with settings yielding the biggest number of individually recognizable structures (adult tissue threshold setting range was 36–200 and for embryos 11–13 to 200). Overlapping structures were separated using the “Watershed” command and also by manual line draw feature. “Analyze particles” command generated data sets containing the number and area of particles. Microsoft Excel 2003 was used to perform statistical analysis and two-tailed Student’s t-test for determining the p-value. Raw data sets are provided as supplement and labels in the Excel tables correspond to the marked images also provided as supplement.

Identification of a perilipin orthologue in C. elegans

We performed BLASTp searches with individual protein sequences of human perilipins that generated no significant hits to Nematoda sequences in the UniProt database, consistent with previous efforts that failed to identify a perilipin-related protein in this phylum. However, when a sequence alignment of chordate perilipins 2 and 3 (OMA database) was submitted as query in PSI-BLAST, the C. elegans protein W01A8.1a (Q23095_CAEEL) was identified as a highly significant hit (E = 3 × 10⁻¹³). A reciprocal PSI-BLAST search with the aligned closest nematode homologues of W01A8.1a identified chordate perilipins as strong hits with human Perilipin 2 (significance score E = 10⁻⁵³) appearing in the second iteration of the search. Similarly, HHpred profile-to-profile searches with human perilipin sequences as a query of the C. elegans proteome identified proteins coded by W01A8.1 (a, b or c) and reciprocally W01A8.1a showed profile homology to all human perilipins and the corresponding Pfam (Punta et al., 2012) perilipin profile (PF03036). Each available nematode proteome contained only a single such perilipin-related sequence, in stark contrast to the insect and chordate proteomes that had 2–5 perilipin paralogues. A sequence alignment of Plin2 and 3 from two selected vertebrates is compared with their nematode homologues (Fig. 1). Although the sequence-to sequence comparisons are not sufficient to unravel the sequence homology between vertebrate and nematode Plins, the similarity appear clearly in the profile-to sequence (PSI-BLAST) and profile-to-profile (HHpred) searches. We conclude that vertebrate Plins and nematode W01A8.1 are statistically highly significant homologues.

The alignment encompasses a substantial part of C. elegans and human sequences (e.g., 90% of W01A8.1 and 87% of Perilipin 2) and covers all three domains characteristic for perilipins (N-terminal PAT, imperfect amphiphilic 11-mer repeat (Brasaemle, 2007) and C-terminal four-helix bundle (Hickenbottom et al., 2004)) covering approximately amino acids 10–100, 125–190 and 220–380 respectively in W01A8.1a. As W01A8.1 and human perilipins appear to be the best mutual reciprocal PSI-BLAST and HHpred hits, W01A8.1 is a very good candidate for a C. elegans orthologue of perilipin.

Protein databases annotate W01A8.1 as Mediator Complex subunit 28, hence the official protein name assignment of MDT-28 in WormBase (WS246). Pfam database (Punta et al., 2012) based the Mediator 28 Hidden Markov model profile on a seed alignment of bovine and mosquito Mediator 28 sequences with W01A8.1. This very profile was probably used subsequently in all automatic annotations of the nematode sequences. However, no substantial homology between W01A8.1 and Mediator 28 exists as shown in the above searches. Since using the WormBase name of W01A8.1 (MDT-28) would be misleading, the gene is referred here by the cosmid name W01A8.1, which gives rise to at least three protein isoforms designated W01A8.1a, W01A8.1b, and W01A8.1c from at least seven different transcripts (W01A8.1a.1, W01A8.1a.2, W01A8.1b.1, W01A8.1b.2, W01A8.1b.3, W01A8.1c.1, W01A8.1c.2). The three protein isoforms are 415, 385, and 418 amino acid residues in length for isoform a, b, and c, respectively (Fig. S1). According to the C. elegans nomenclature, we suggest to rename W01A8.1 as Cel-plin-1 (isoform a, b, and c) and proteins Cel-PLIN-1 (isoform a, b, and c).

W01A8.1 protein products are cytoplasmic and reside primarily on lipid droplets

If the proteins encoded by W01A8.1 act as perilipins, they would be expected to be associated with lipid droplets (Kozusko et al., 2015). To test this, we created translational reporter transgenes regulated by the putative endogenous promoter expressing isoform b and lines in which the genomic locus was tagged by an in-frame C-terminal GFP cassette. The second transgene, W01A8.1a/c::gfp, is likely to express not only high levels of a and c tagged isoforms, but also the native isoform b (Fig. S1). The translational fusion constructs resulted in high levels of cytoplasmic proteins present in intestinal and epidermal cells on vesicular structures with the characteristic appearance of lipid droplets. This pattern of expression and cellular distribution was observed beginning at the three-fold embryonic stage and continued throughout development to adulthood (Fig. 2). To confirm that the observed GFP-associated vesicular structures were indeed lipid droplets, transgenic animals were stained with the lipophilic reagent LipidTox as previously described (O’Rourke et al., 2009). The translational GFP fusion protein reporters were localized at the periphery of fat droplets that were LipidTox positive (Fig. 2).

Human PLINs 1 and 2 label identical compartments as W01A8.1 proteins in C. elegans

We prepared transgenic C. elegans lines expressing human PLIN1, PLIN2 and PLIN3 fused to GFP and regulated by the W01A8.1 promoter. PLIN1::GFP and PLIN2::GFP were localized on spherical cytoplasmic structures primarily in gut and epidermal cells (Figs. 3A, 3C, 3D and 3F) with identical appearance as W01A8.1 translational reporter GFP tagged proteins and Drosophila PLIN1::GFP expressed in C. elegans as reported by Liu et al. (2014). PLIN3 expression was diffusely cytoplasmic and only faintly defined spherical structures (Figs. 3G and 3I). The structures clearly labeled with PLIN1::GFP and PLIN2::GFP were also positive in LipidTox staining (shown for PLIN2::GFP in Figs. 3J–3L). We conclude that W01A8.1 proteins are localized on the same structures as human PLIN1 and PLIN2.

W01A8.1 reduction-of-function alters the appearance of lipid droplets in early embryos and causes a reduction of brood size

To test the function of W01A8.1, we used RNAi done by germline injection and by feeding. W01A8.1 RNAi made by microinjections and feeding resulted in a significantly smaller brood size, with approximately 30% less progeny. RNAi made by microinjections resulted in ∼52% reduced progeny laid in the first 24 h after microinjections and after 48 h ∼28% reduction in progeny laid compared to controls (n = 260, n = 550, for day one and n = 1,000, n = 1,400 for day two).

Repetition of knockdown by RNAi feeding over two generations confirmed this observation (Fig. S4). dsRNA feeding caused the W01A8.1 specific group to produce ∼30% less larvae compared to controls, experiment was repeated twice independently with consistent findings. We confirmed, using RT-qPCR, that feeding based knockdown (represented in Fig. S4) resulted in approximately 45% decrease in W01A8.1 transcripts (data not shown).

Staining of adult hermaphrodites with LipidTox (after formaldehyde fixation) revealed larger lipid droplets in early embryos derived from adults inhibited for W01A8.1 (Figs. 4A and 4B) compared to controls (Figs. 4C and 4D).

Targeted disruption of W01A8.1 results in early embryonic defects but not lethality

In order to eliminate the W01A8.1 function completely, we designed a CRISPR/Cas9-mediated gene editing approach to eliminate almost the entire coding region (Fig. S2). We also included a rescuing plasmid consisting of isoform a that was prepared as cDNA synthesized in vitro using synonymous codons (W01A8.1(a)synth::gfp) that is protected against CRISPR/Cas9 targeted editing but allows the production of the wild type isoform a at the protein level. Lines that expressed the GFP fusion transgene were morphologically normal and W01A8.1(a)synth::GFP was found on lipid droplet-like structures as expected (Figs. 4E and 4G) that also stained positive by LipidTox (Figs. 4F and 4G). This transgenic strain yielded lines either carrying or losing the rescuing transgene in the background of a disrupted endogenous W01A8.1. The elimination of W01A8.1 was easily monitored by PCR (Fig. S3). Surprisingly, animals with the deleted W01A8.1 locus that lost the extra-chromosomal rescuing array were able to reproduce normally. From several lines that had a confirmed disruption of W01A8.1 and a confirmed loss of the extrachromosomal array, the line CK123 (KV001) was selected and used for subsequent analyses. As was observed in W01A8.1 RNAi embryos, loss of W01A8.1 activity resulted in the formation of large LipidTox-positive structures (Figs. 4H and 4I) that were clearly bigger than droplets observed in control embryos using the same protocol (Figs. 4C and 4D). These large lipid-containing structures were observable in live mutant embryos (Fig. 4J) but not in wild type embryos (Fig. 4K) using Nomarski optics. Viewing through multiple focal planes in live, developing embryos lacking W01A8.1 showed that these large lipid droplets are present in embryos during the early mitotic divisions and were localized around the nucleus. Staining with LipidTox (after fixation) confirmed the lipid content in the vesicular structures arranged around dividing nucleus (Figs. 4L and 4M). These large vesicles persist through the two-cell stage, disappearing in most embryos with more than 6 cells. On fixed embryos stained with LipidTox, larger than wild type lipid droplets are visible until late embryonic stages, including three fold embryos.

In order to visualize lipid-containing structures in W01A8.1 null mutants and in controls in vivo, we used CARS microscopy (done with kind help from Dr. Zhongxiang Jiang, Leica Microsystems, Mannheim, Germany). The CARS systems allow visualization of lipids of specific categories by tuning into symmetric CH₂ vibrations of specific fat composing molecules (Zumbusch, Langbein & Borri, 2013). CARS microscopy clearly confirmed the formation of large lipid containing vesicles in early embryos and allowed detailed analysis of the W01A8.1 null phenotype. CARS microscopy also confirmed the gradual increase of the size of lipid containing structures during oogenesis (Figs. 5A and 5B), the sudden re-localization of these structures to the periphery of the dividing nucleus in the first embryonic division (Fig. 5E), and the propagation of this phenotype, although with gradually diminishing appearance, throughout embryonic development (Figs. 5F and 5H). CARS microscopy detected this phenomenon also in wild type animals, although the size of lipid containing structures was smaller making the phenomenon of the sudden re-localization of lipid containing structures less obvious (Figs. 5C, 5D and 5G) than in W01A8.1 null embryos. In contrast to embryos, lipid-containing structures in intestinal and epidermal cells of adult W01A8.1 null mutants (Figs. 5J and 5L) were smaller than lipid-containing structures in control animals (Figs. 5I and 5K).

Morphometric analysis confirmed that W01A8.1 null mutant embryos contained larger lipid positive structures recognized by CARS microscopy (Figs. 6A and 6B) that represent a larger total area (Fig. 6C), as determined by quantitating individual focal planes images. Morphometry revealed many small structures with area 1–4 AU (arbitrary units) with the provided threshold settings (Figs. 6A and 6E). There was a clear inverse relation in the number of large and small structures (with area <5 AU) for embryos as well as for adult tissues. The analysis in Figs. 6B and 6F shows that inclusion of small structures into analysis does not significantly affect the results but affects only standard deviation of the particle size distribution indicating that the results are independent on the setting of the limit for the size of lipid containing structures. The probability of the results were assayed using Student’s t-test and found to be statistically significant as the probability of this result, assuming the null hypothesis (no difference between control and experimental sets) was less than 0.0001.

Despite the fact that there were a larger number of individually recognizable lipid containing structures in wild type embryos (Fig. 6D), the mean area of these structures in mutants was considerably larger (Fig. 6C). In contrast, adult mutant animals contained smaller, more numerous lipid-containing structures (Figs. 6E, 6F and 6H) that covered a smaller total area (Fig. 6G) (and therefore volume) compared to wild type controls.

The morphometric analysis confirms that there is more CARS positive signal and therefore most likely more fat in W01A8.1 null embryos (despite lower threshold used for analysis of wild type embryos) and less CARS positive signal (and less fat) in adult tissues of W01A8.1 null animals compared to controls.

The analysis of the number of progeny laid by animals lacking W01A8.1 in comparison to wild type animals showed a decrease of progeny in mutant animals statistically significant on the day 3 (Fig. S5).