Worms were cultivated on the surface of NGM agar seeded with the Escherichia coli strain OP50 as the primary food source and grown in 20°C incubators using standard protocols unless indicated otherwise. All experiments were performed on hermaphrodites. The wild-type strain N2, variety Bristol (Brenner, 1974) and other strains used in this study are listed in Supplementary Table 1. Strains were constructed using standard genetic methods (Fay, 2006) and genotypes were confirmed either by phenotype (for example, the transgenic strain was marked by fluorescence) or by PCR (for example, by identifying small deletions in mutant strains).

To generate transgenic worms expressing the circ-crh-1 in muscle cells, exon 4 of crh-1 and intronic sequences flanking exon 4 (Figure 4A) were cloned into the pMC10 plasmid (a kind gift from the Sengupta Lab). Next, promoter sequences of myo-3 (∼2.5 kb) were cloned at the 5′-end of the circ-crh-1 sequence using the multi cloning site (MCS) of pMC10. The generated myo-3p::circ-crh-1 construct along with the unc-122p::RFP co-injection marker (AddGene) was injected into VDL1300 crh-1(syb385); dvIs100[unc-54p::Aβ_1–42::unc-54 3′-UTR, mtl-2p::GFP] animals to create the VD12 strain. Transgenic worms carrying extrachromosomal arrays overexpressing pie-1p::circ-crh-1 (VDL975), rab-3p::circ-crh-1 (VDL1104) were crossed with the VDL1300 strain to create VDL1306 and VDL1307 strains (Supplementary Table 1). A col-49 mutant allele (syb8747) harboring a 1180bp deletion was generated using a Co-CRISPR method to create the VD10 strain (SunyBiotech), which was confirmed by PCR and Sanger sequencing. sgRNAs used to generate the col-49(syb8747) mutant were Sg1: 5′-cctcatcatcatgtggaaattcg and Sg2: 5′-cccacctagaactgcttgattcg. We crossed the VD10 with the GMC101 strain to create VDL1308 col-49(syb8747); dvIs100[unc-54p::Aβ_1–42::unc-54 3′-UTR, mtl-2p::GFP]. We then crossed the VDL1308 with the VDL1300 strain to create VDL1310 crh-1(syb385); col-49(syb8747); dvIs100[unc-54p::Aβ_1–42::unc-54 3′-UTR, mtl-2p::GFP] using standard genetic methods. We generated a transgenic line that overexpresses col-49 from a transgene carrying a multiple copy array of the col-49 genomic sequence (∼1.2 kb) under control of its endogenous promoter upstream sequence (∼2 kb) and col-49 3′-UTR sequence (∼1 kb). The resulting plasmid (col-49p::col-49 genomic::col-49 3′-UTR) along with the unc-122p::RFP co-injection marker was injected into wild-type worms to generate VD16 (SunyBiotech). Next, we crossed the VD16 with the GMC101 strain to create the VDL1309 strain using standard genetic methods. Supplementary Table 1 shows all strains created and used in this study.

All strains were maintained at 20°C for at least two generations before the lifespan assay. Adult worms age-synchronized by hypochlorite treatment and collected eggs were hatched overnight at 20°C. L1 larvae were then plated onto NGM plates seeded with E. coli OP50 bacteria. At the L4 larval stage, 90–150 worms per genotype were transferred to new 6 cm NGM plates seeded with 10x concentrated E. coli OP50 bacteria containing 0.5 μM 5-fluorodeoxyuridine (FUdR) to inhibit the development of self-progeny, and then shifted at the young adult stage to 25°C. Each strain was assayed in parallel and each plate contained 10–15 worms. Worms were blindly scored every day and were considered dead when they did not respond to touch of the platinum wire pick and were subsequently removed from the plate. Worms that experienced ventral rupture, bagging, or walling were censored from the lifespan analysis.

The paralysis assay was performed using GMC101 transgenic animals expressing unc-54p::Aβ_1–42 as described previously (McColl et al., 2012). Briefly, worms were age-synchronized by hypochlorite bleaching and cultivated at 20°C. After they reached the L4 larval stage, worms were transferred to assay plates freshly seeded with E. coli OP50 bacteria, containing 0.5 μM of FUdR to inhibit the development of self-progeny, and then shifted at the young adult stage to the higher permissive 25°C temperature to induce paralysis unless indicated otherwise. About 15 worms were placed on each 6 cm NGM plate, and animals were blindly scored every 24 h as “paralyzed” if they failed to perform a full body wave propagation following a repeated touch-provoked response.

Worms were age-synchronized worms by hypochlorite treatment and collected eggs were hatched overnight at 20°C in 1x M9 buffer. L1 larvae were then plated onto NGM plates seeded 10x concentrated E. coli OP50 bacteria and allowed to develop to the L4 larval stage at 20°C. L4 larvae were then collected, washed and re-plated onto E. coli OP50 seeded NGM plates containing 0.5 μM FUdR. Worms were either upshifted to 25°C or kept at 20°C. Adult worms were collected at different aging time-points and washed with 1x M9 buffer through 35 μM nylon mesh to remove bacteria. Worm pellets of 100–300 μl were then transferred into green RINO tubes (Next Advance) and TRizol LS reagent (ThermoFisher Scientific, Cat #10296028) was added in a 1:3 ratio. Worms were immediately lysed by bead beating them for 5 min using a Bullet Blender Pro Storm (Next Advance). Total RNA was extracted using the Purelink RNA mini-kit, followed by a DNAse I treatment following the manufacturer’s protocol (Ambion, Cat #12183020). RNA was quantified by a Nanodrop. Bioanalyzer or tapestation (Agilent) were used for qualification as needed. and samples were stored at −80°C.

To quantify and confirm individual circular or linear transcripts, 0.5 μg total RNA was reverse transcribed using Superscript III to prepare cDNA using random hexamers (Invitrogen, Cat #18080051). Next, cDNA samples were diluted and used with PowerUp SYBR Green Master Mix (Applied Biosystems, Cat #A25471) for RT-qPCR analysis analyzed on a CFX96 Real-Time System (Bio-Rad). For RT-qPCRs of circRNAs, we used outward-facing primers. For host gene linear RNA counterparts, one primer was located in the circularizing exon and the other was located in the upstream or downstream non-circularizing exon. For linear mRNAs such as collagen-encoding genes and Aβ mRNA, we used forward-facing primers. Fold-change values were calculated using wild-type (N2) ΔCt as control values for the 2^–ΔΔCt method. Data is normalized to housekeeping genes (cdc-42 or act-1) mRNA. Primer sequences are listed in Supplementary Table 2.

Four-day old adults were randomly collected during paralysis assays and stained with 1 mM X-34 (Sigma, Cat # SML1954) 10 mM Tris-HCl pH 8.0 for 2 h as previously described (Link et al., 2001). Stained worms were then washed twice with 1x M9 and transferred to E. coli seeded NGM plates containing 0.5 μM FUdR for 24 h to de-stain worms. De-stained 5-day old adults were then placed onto a 2% agarose pad with 10 μM levamisole to anesthetize worms. Confocal microscopy was used to acquire and capture images of the worm head using a 40x oil objective (405 nm excitation, 470–520 nm emission range). ImageJ software (NIH) was used to quantify Aβ aggregates.

As previously described (Teuscher et al., 2019), worms were age-synchronized by hypochlorite treatment and collected eggs were hatched overnight at 15°C, 20°C, or 25°C in 1x M9 buffer. L1 larvae were then plated onto NGM plates seeded E. coli OP50 bacteria and allowed to develop 24 h post-L4 larval stage (1-day adult) at 15°C, 20°C, or 25°C. 1-day adult worms were then collected with 1x M9 buffer and washed with dH₂O through a 35 μM nylon mesh to remove bacteria. Worm pellets of 300 μl were then transferred into green RINO tubes (Next Advance) and lysed by bead beating them for 10 min using a Bullet Blender Pro Storm (Next Advance). Total collagen level was determined using the QuickZyme Total Collagen Kit (QuickZyme Biosciences), following the manufacturer’s protocol. Briefly, lysate samples were mixed with 12M HCl solution and incubated for 20 h at 95°C. Then assay buffer was added, and the 96-well plate was incubated at room temperature for 20 min, followed by the addition of the detecting reagent and incubation at 60°C for 60 min. Total collagen level was measured and quantified as a fraction of total protein abundance using a Synergy HT BioTek microplate reader. Total protein levels were quantified using the Pierce™ BCA Protein Assay Kit (ThermoFisher Scientific) following the manufacturer’s protocol.

Statistical comparisons and graphical representations were performed with the Online Application for Survival Analysis, OASIS 2 (Han et al., 2016). For lifespan survival and paralysis curves, we used the Mantel-Cox log-rank test. Other data were analyzed using Graphpad Prism 9 software and statistical comparisons made include the Mann-Whitney t-test or the one-way ANOVA followed by a post hoc multiple-comparisons test. p-values are reported in the figure legends.

We previously demonstrated that expression of circ-crh-1 accumulates with C. elegans aging (Cortés-López et al., 2018) and that loss of circ-crh-1 expression results in a significant extension of mean lifespan (Knupp et al., 2022). To directly test whether circ-crh-1 expression plays a role in Aβ-induced toxicity, we used the GMC101 strain (further referred as unc-54p::Aβ_1–42) in which expression of human Aβ_1–42 peptides constitutively in muscle cells provokes an inducible age-progressive full-body paralysis when shifted from the non-restrictive 20°C temperature to a higher permissive 25°C temperature (McColl et al., 2012). We observed that mutations in circ-crh-1 (syb385 and syb2657), which exhibit a complete loss of circ-crh-1 expression but normal linear crh-1 expression, showed less severe paralysis in unc-54p::Aβ_1–42 animals at the higher permissive 25°C temperature (Figure 1A; Knupp et al., 2022). Thus, loss of circ-crh-1 expression significantly delayed the onset of Aβ-induced paralysis.

To further test the protective effect of circ-crh-1 expression Aβ-induced toxicity, we next tested its loss on the lifespan of unc-54p::Aβ_1–42 expressing animals. Previous work has shown that the expression of Aβ_1–42 in body wall muscle cells severely decreased lifespan (Gallrein et al., 2021). Similarly, we found that unc-54p::Aβ_1–42 expressing animals led to a significantly shorter mean lifespan compared to wild-type controls (20.5% reduction, 8.10 days for GMC101 versus 10.19 days for wild-type, p < 0.0001) when animals were shifted from 20°C to the higher permissive 25°C temperature (Figure 1B). Interestingly, crh-1(syb385) was able to restore the reduced lifespan of unc-54::Aβ_1–42 animals back to wild-type levels (9.7 days for crh-1(syb385); unc-54::Aβ_1–42 versus 8.1 days for GMC101, p < 0.0001) (Figure 1B). No significant differences were observed in mean lifespan between unc-54::Aβ_1–42 and wild-type animals (p = 0.614) at the 20°C temperature (no Aβ-induction) (Supplementary Figure 1). Thus, circ-crh-1 mutations prevented lifespan shortening induced by Aβ. Together, these findings suggest that loss of circ-crh-1 expression protects C. elegans from Aβ-induced toxicity.

We next examined the impact of Aβ-induction on the expression of the age-accumulated circ-crh-1 in aged unc-54::Aβ_1–42 animals compared to wild-type. We therefore conducted RT-qPCR analysis to measure the RNA fold change of circ-crh-1 expression (i.e., cel-circ_0000439) in 4-day old adults compared to 1-day old adults for wild-type and between unc-54::Aβ_1–42 animals shifted from 20°C to the higher permissive 25°C temperature (after Aβ-induction). As expected, the fold-change ratio of circ-crh-1 gene expression was higher in unc-54::Aβ_1–42 animals than wild-type by 1.25-fold (Figure 2A). Importantly, expression of linear crh-1 was not significantly affected after Aβ-induction in 1-day and 4-day old adults (Figure 2A). Similar results were observed when normalizing circ-crh-1 gene expression of 4-day old unc-54::Aβ_1–42 adults at 25°C (after Aβ-induction) to the 20°C temperature (before Aβ-induction) (Figure 2B). These results suggest that Aβ-induction positively regulates circ-crh-1 expression, consistent with a detrimental role for age-accumulated circRNAs in AD.

Next, we asked whether loss of circ-crh-1 expression reduced Aβ-aggregate deposits in unc-54::Aβ_1–42 animals, which results in age-progressive paralysis and a shortened lifespan (Figure 1). In order to rule out the effects of circ-crh-1 loss on the transcription of Aβ_1–42 expression rather than simply reducing Aβ aggregates in muscle, we first used RT-qPCR analysis to measure Aβ mRNA levels at different aging time-points. We found no significant differences in Aβ gene expression between unc-54::Aβ_1–42 and crh-1(syb385);unc-54::Aβ_1–42 animals at 1, 2, and 3-day old adulthood at the higher permissive 25°C temperature (Figure 3A). We then utilized the sensitive amyloid-binding dye X-34 (Link et al., 2001) to specifically stain and visualize in vivo Aβ aggregates in muscle cells of C. elegans as reported (McColl et al., 2012) and assessed whether circ-crh-1(-) mutations reduce the number of Aβ aggregates puncta in unc-54::Aβ_1–42 expressing worms. Consistent with the delayed onset of the age-progressive Aβ-induced paralysis and restoration of the shortened lifespan, we found that the number of X-34 positive Aβ-aggregates were significantly reduced (p < 0.05) in crh-1(syb385) mutants carrying unc-54::Aβ_1–42 compared to GMC101 in 5-day old adults at the higher permissive 25°C temperature (Figures 3B, C). Thus, loss of circ-crh-1 expression inhibits the accumulation of Aβ aggregates.

We next asked whether circ-crh-1 expression in muscle could restore the delayed onset of Aβ-induced paralysis of crh-1(syb385) mutants expressing Aβ_1–42 in muscle cells. To investigate this question, we used tissue-specific rescue experiments. We previously showed that circ-crh-1 expression in neurons is an important determinant for lifespan regulation (Knupp et al., 2022). We decided to create crh-1(syb385); unc-54::Aβ_1–42 transgenic animals that express circ-crh-1 under select tissue-specific promotors. We cloned the crh-1(exon 4) circularizing sequence in between the left and right reverse complementary match (RCM) sequences, and used muscle (myo-3), pan-neural (rab-3), or germline (pie-1) specific promoters to drive the circ-crh-1 expression transgene (Figure 4A). Expression of circ-crh-1 under control of the muscle-specific myo-3 promoter could partially restore the delayed onset of Aβ-induced paralysis of crh-1(syb385); unc-54::Aβ_1–42 animals at the higher permissive 25°C temperature (Figure 4B), suggesting that circ-crh-1 expression in muscle is an important determinant for Aβ-induced toxicity. Surprisingly, however, circ-crh-1 expression driven by rab-3 and pie-1 promoters also partially restored Aβ-induced paralysis at 25°C (Figure 4B). These results suggest that in addition to muscle, circ-crh-1 expression may have additional requirements in other tissues to alter Aβ-induced paralysis.

We previously showed that circ-crh-1(-) mutants exhibit widespread transcriptomic changes that might impact various age-related pathways (Cortés-López et al., 2018). Notably, among the identified genes, a subset included collagen-encoding genes with many of which showed increased expression levels in circ-crh-1(-) mutants. Interestingly, cuticular collagens are implicated in Aβ aggregate formation and clearance pathways of Aβ in C. elegans (Jongsma et al., 2023). We hypothesized that the increased expression of cuticular collagen genes contributes to the amelioration of Aβ-induced toxicity in circ-crh-1(-) mutants carrying the unc-54::Aβ_1–42 transgene. To test this hypothesis, we selected six cuticular collagen genes of interest from the 21 collagen genes with elevated expression previously identified in crh-1(syb385) mutants (Knupp et al., 2022), including collagens that have a known association with lifespan such as col-49 and col-179 (Palani et al., 2023). Among the collagen genes tested by RT-qPCR analysis, only col-49 showed elevated gene expression levels in crh-1(syb385) mutants expressing unc-54::Aβ_1–42 compared to GMC101 controls at 25°C in 3-day old adults (Figure 5A). No detectable differences were observed in total collagen levels in crh-1(syb385) mutants compared to wild-type under different cultivation temperatures (Supplementary Figure 2). Thus, the elevated expression of col-49 in crh-1(syb385) mutants likely represents a specific response to Aβ-induced toxicity, rather than a general increase in collagen production driven by circ-crh-1.

We next generated col-49 deletion mutants using a CRISPR/Cas9 strategy and crossed the mutant with GMC101 to assess Aβ-induced toxicity through a paralysis assay at 25°C. We found that col-49(syb8747); unc-54::Aβ_1–42 animals exhibited an exacerbated, age-progressive paralysis compared to the control (GMC101) at the higher 25°C permissive temperature (Figure 5B). col-49(syb8747) mutants without the unc-54::Aβ_1–42 transgene did not display paralysis at 25°C (Figure 5B).

To test whether col-49 expression is required for the delayed onset of Aβ-induced toxicity observed in crh-1(syb-385) mutants, we crossed the col-49(syb8747) mutation into crh-1(syb385) mutants, both of which carried the unc-54::Aβ_1–42 transgene. Mutations in col-49(syb8747) significantly suppressed the delayed Aβ-induced paralysis phenotype of crh-1(syb385) animals with the unc-54::Aβ_1–42 transgene when 1-day old adults were shifted from 20°C to the higher 25°C permissive temperature (Figure 6A). We also examined overexpression (OE) of col-49 on Aβ-induced paralysis, and found that col-49 OE transgenic animals weakly but significantly (p < 0.05) delayed the onset of Aβ-induced paralysis (Figure 6B) similar as circ-crh-1(-) mutants. Collectively, these results suggest that col-49 gene expression is elevated in circ-crh-1(-) mutants, and that loss of circ-crh-1 promotes Aβ-induced paralysis in a col-49 dependent manner.