Regulation of fatty acid desaturase- and immunity gene-expression by mbk-1/DYRK1A in Caenorhabditis elegans

In the nematode Caenorhabditis elegans, longevity in response to germline ablation, but not in response to reduced insulin/IGF1-like signaling, is strongly dependent on the conserved protein kinase minibrain-related kinase 1 (MBK-1). In humans, the MBK-1 ortholog DYRK1A is associated with a variety of disorders, most prominently with neurological defects observed in Down syndrome. To better understand mbk-1’s physiological roles and their dependence on genetic background, we analyzed the influence of mbk-1 loss on the transcriptomes of wildtype and long-lived, germline-deficient or insulin-receptor defective, C. elegans strains by RNA-sequencing. mbk-1 loss elicited global changes in transcription that were less pronounced in insulin-receptor mutant than in germline-deficient or wildtype C. elegans. Irrespective of genetic background, mbk-1 regulated genes were enriched for functions in biological processes related to organic acid metabolism and pathogen defense. qPCR-studies confirmed mbk-1 dependent induction of all three C. elegans Δ9-fatty acid desaturases, fat-5, fat-6 and fat-7, in wildtype, germline-deficient and insulin-receptor mutant strains. Conversely, mbk-1 dependent expression patterns of selected pathogen resistance genes, including asp-12, dod-24 and drd-50, differed across the genetic backgrounds examined. Finally, cth-1 and cysl-2, two genes which connect pathogen resistance to the metabolism of the gaseous messenger and lifespan regulator hydrogen sulfide (H2S), were commonly suppressed by mbk-1 loss only in wildtype and germline-deficient, but not in insulin-receptor mutant C. elegans. Our work reveals previously unknown roles of C. elegans mbk-1 in the regulation of fatty acid desaturase- and H2S metabolic-genes. These roles are only partially dependent on genetic background. Considering the particular importance of fatty acid desaturation and H2S for longevity of germline-deficient C. elegans, we propose that these processes at least in part account for the previous observation that mbk-1 preferentially regulates lifespan in these worms.


Introduction
Caenorhabditis elegans mbk-1 (minibrain-related kinase 1) encodes an evolutionarily conserved dual-specificity tyrosine-regulated kinase (DYRK) orthologous to Drosophila melanogaster minibrain and human DYRK1A/B [1][2][3][4]. Human DYRK1A maps to chromosome 21q22. 13 in the Down syndrome critical region, and its overexpression has been implicated in the neurological defects Open Access *Correspondence: Hildegard.I.Mack@gmail.com; pidder.jansen-duerr@uibk. ac.at 1 Institute for Biomedical Aging Research, University of Innsbruck, Rennweg 10, 6020 Innsbruck, Austria Full list of author information is available at the end of the article associated with this disorder [5,6]. Inhibitory single nucleotide variants, small insertions/deletions or complete deletion of one DYRK1A-allele in patients or in mice also cause intellectual disability and other abnormalities, highlighting the sensitivity of DYRK1A-activity to gene dosage [7,8]. DYRK1A is expressed in neuronal and nonneuronal tissues during development and adulthood [2,9] and has been described to contribute to tumor development, both as an oncogene and as a tumor suppressor [2]. Indeed, individuals with Down syndrome display an elevated risk for certain types of childhood leukemia, but a lower risk for solid tumors across all ages [10,11].
C. elegans mbk-1 differs from its related kinase mbk-2 (DYRK2) and from a more distant family member, hpk-1 (HIPK2) in its expression and subcellular localization patterns, as well as in its physiological role [3]. MBK-1 is expressed in all somatic cells during development and adulthood and localizes predominantly to the nucleus [3]. Overexpression of mbk-1 impairs olfactory behavior, paralleling the DYRK1A-overexpression induced neurological defects observed in other species [3]. On the other hand, mbk-1 inactivation does not cause obvious phenotypic alterations under standard culture conditions [3]. Yet, upon exposure to Pseudomonas aeruginosa strain PAO1, mbk-1 is required for protecting C. elegans from fast killing by this pathogen [12]. Moreover, mbk-1 loss strongly reduces longevity in response to germline ablation, while exerting smaller effects on longevity in response to reduced insulin-like signaling, or on wildtype lifespan [13]. The mechanisms by which mbk-1 promotes pathogen defense and longevity have not been fully elucidated.
In this study, we aimed to better understand the mechanisms through which mbk-1 preferentially modulates C. elegans lifespan in response to GSC deficiency. Towards this goal, we examined the effect of mbk-1 loss on stress resistance and gene transcription of normal and long-lived, wildtype, germline-deficient or daf-2(−) mutant C. elegans strains, in functional assays and by RNA-sequencing.

Loss of mbk-1 decreases heat and oxidative stress resistance in germline-deficient worms
We previously reported that the longevity of GSC deficient C. elegans is strongly dependent on mbk-1 [13]. Conversely, long-lived daf-2(−) or wildtype worms do not, or only to a minor extent, require mbk-1 for lifespan regulation [13]. As changes in longevity are frequently accompanied by changes in resistance to environmental stress such as heat or oxidative stress [23], we examined how mbk-1 influences these properties in GSC(−), daf-2(−) and corresponding GSC(+) or daf-2(+) control worms (cf. Methods). Upon exposure to heat stress, the loss of function allele mbk-1(pk1398) (hereafter referred to as mbk-1(−); [3]), consistently reduced survival of GSC(−) worms, while a more moderate, or no reduction of survival was observed in daf-2(−) and in GSC(+)/ daf-2(+) worms (Fig. 1a, c, Additional file 1: Table S1). Oxidative stress resistance on the other hand was consistently decreased in mbk-1(−) worms in GSC(−), GSC(+) and daf-2(+), but not in daf-2(−) background ( Fig. 1b, d, Additional file 1: Table S1). Loss of daf-16 produced a statistically significant decrease in survival in response to each of the two stressors in all genetic backgrounds in all repetitions of the experiment, although effects were generally smaller in GSC(+)/daf-2(+) than in long-lived GSC(−) or daf-2(−) worms [24] (Additional file 1: Table S1). Yet, when daf-16 and mbk-1 loss were combined, further reduction of survival was consistently observed only in heat-stressed GSC(−) worms ( Fig. 1; Additional file 1: Table S1). Collectively, these results are consistent with and expand the concept [13] that mbk-1 activity is particularly important for GSC(−) physiology and longevity. Moreover, our data rise the possibility that mbk-1 exerts its lifespan-and stress modulatory function in these worms through both, daf-16 dependent andindependent mechanisms.

Mbk-1 modulates the expression of fatty acid desaturase genes
The GOterm "organic acid metabolic process" showed significant enrichment among mbk-1 induced genes irrespective of genetic background, and contained two of the three C. elegans Δ9-fatty acid desaturase genes [34], fat-5 and fat-6, in GSC(−) and in daf-2(+) worms (Fig. 2e, Additional file 1: Table S3). Differential expression of the third desaturase, fat-7, was not statistically significant in any of the four backgrounds examined (Additional file 2: Table S9). Previous reports indicate that fat-5; fat-6; fat-7 triple mutants are not viable, and that both, fat-6 and fat-7 together, are required for GSC(−) longevity, but not for normal wildtype lifespan [35,36].
As none of the mbk-1 dependent DEGs validated above has been implicated in the same form of pathogen resistance than mbk-1 itself has been implicated in, we searched or RNA-seq data for genes that may mediate this known role of mbk-1. Specifically, mbk-1 protects C. elegans from fast-killing by a particular strain of the gram-negative bacterium P. aeruginosa, PAO1 [12]. Given that growing C. elegans in the presence of hydrogen sulfide (H 2 S) also confers PAO1-resistance [38], and that PAO1 kills C. elegans by hydrogen cyanide (HCN) poisoning [39], we focused on genes implicated in H 2 S-and/or HCN-metabolism. Indeed, qPCR-analysis indicated mbk-1 loss-induced downregulation of the cystathionine gamma lyase ortholog cth-1 [40] in all four backgrounds examined, confirming results or trends also observed in our RNA-seq data ( Fig. 4; Additional file 1: Table S3). In addition, a known PAO1-resistance gene, the cyanoalanine synthase cysl-2 [38] was downregulated upon mbk-1 loss in GSC(−) and in GSC(+)/daf-2(+) C. elegans (Fig. 4). Collectively, these data indicate that mbk-1 modulates the expression of pathogen defense genes in a genetic background-dependent manner, and rise the possibility that mbk-1's function in pathogen resistance/ immunity is not limited to fast-killing by gram-negative P. aeruginosa PAO1.

Discussion
In this study, we show that the DYRK1-ortholog mbk-1, which is required for full longevity of germline-deficient C. elegans, also promotes stress resistance in these animals. Moreover, using a transcriptomics approach, we discovered that mbk-1 modulates the expression of genes implicated in pathogen defense and in fatty acid desaturation, a process that, as mbk-1, strongly contributes to GSC(−), but not to daf-2(−) longevity. Thus, our study helps explain mbk-1's previously described activities in C. elegans lifespan regulation and pathogen resistance. To our knowledge, this report also is the first in any organism to profile the influence of a DYRK1A-family member on gene expression in the postmitotic adult stage.

Genes and pathways deregulated upon loss of mbk-1
Our RNA-seq profiling of normal and long-lived GSC(−) and daf-2(−) C. elegans revealed global changes in gene expression that were more limited in daf-2(−) than in GSC(−) or GSC(+)/daf-2(+) animals, but over-all similar across all backgrounds examined. Thus, it is not surprising that mbk-1 regulated genes were functionally enriched for some of the same biological processes. Remarkably, mbk-1 most consistently, i.e. across multiple or all backgrounds analyzed here, modulated the expression of genes implicated in organic acid metabolism and pathogen defense. Indeed, similar GO terms, such as "metabolic process" and "immune system process" were also detected in published gene expression studies on the mbk-1 ortholog in other systems, including zebrafish embryos [41], and HeLa cells [42]. Collectively, these data provide evidence for mbk-1/DYRK1A playing an evolutionarily conserved role in these processes.

Role of mbk-1 in pathogen defense
Mbk-1 has been implicated in the resistance against a particular strain of the gram-negative bacterium P. aeruginosa, PAO1 [12], which produces HCN to rapidly kill C. elegans [39,45]. More specifically, mbk-1 contributes to transcriptional activation of the key PAO1-resistance factor HIF-1, but the precise mechanisms remained unclear [12]. The results we present here suggest a potential solution for the so far "missing link" between mbk-1 and hif-1. Specifically, mbk-1 promotes the expression of cth-1, a cystathionine gamma lyase involved in the synthesis of H 2 S [40]. Moreover, mbk-1 induces the cyanoalanine synthase cysl-2, which directly detoxifies the PAO1 virulence factor HCN [38] in a reaction that also generates H 2 S [38]. H 2 S in turn was proposed as an endogenous activator of HIF-1 [46]. Beyond PAO1-resistance, it is interesting to note that GSC(−) worms display elevated levels of H 2 S relative to wildtype C. elegans and require transsulfuration pathway enzymes, and by extension: H 2 S, for their longevity [47]. Therefore, it is tempting to speculate that ensuring beneficial levels of H 2 S constitutes another mechanism via which mbk-1 ensures GSC(−) longevity.
Remarkably, we observed that mbk-1 in wildype C. elegans also promotes the expression of genes such as dod-24 and drd-50, which are upregulated upon exposure to another pathogenic P. aeruginosa strain, PA14 [37]. PA14, in contrast to PAO1, kills C. elegans slowly by gut colonization [12,37]. Even though some PA14-induced genes such as asp-12 [37] are suppressed by mbk-1, our results provide evidence for a broader function of mbk-1 in pathogen defense that is not restricted to the previously reported protection from P. aeruginosa PAO1 [12]. Clearly, it will be interesting to examine how mbk-1 loss modulates C. elegans' response to other pathogens in the future.

Conclusions
In summary, the data reported here suggest specific mechanisms underlying previously known activities of the C. elegans DYRK1-family kinase mbk-1 in the regulation of lifespan and pathogen defense. Moreover, they provide further evidence for genetic background-restricted, as well as for genetic background-independent functions of mbk-1. Specific novel regulatory influences of mbk-1 emerging from our studies include fatty acid desaturation, H 2 S-metabolism and P. aeruginosa PA14-resistance. As many components of these pathways, and mbk-1 itself, are evolutionarily conserved, our results may also be applicable to mbk-1 orthologs in postmitotic tissues of other species.

C. elegans strains and culture
Strains used in this study are listed in Additional file 1: Table S7. Worms were cultured following standard protocols on NG agar plates seeded with E. coli OP50 [48]. C. elegans carrying the glp-1(e2144ts)-mutation served as a genetic model for germline-deficiency [15,47]. To eliminate germ cells, glp-1(ts) strains (referred to as GSC(−) in Results/Discussion), and corresponding glp-1(+) (i.e. GSC(+)) control strains, were incubated at 25 °C for the first 24 h of postembryonic development and subsequently, shifted to 20 °C for the remainder of the experiment. Daf-2(e1370) worms and corresponding daf-2(+) control worms were continuously cultured at 20 °C.

Stress resistance assays
Worms were synchronized by hypochlorite treatment and transferred to assay plates on day 2 of adulthood at a density of 20-30 worms per 3 cm plate for heat stress experiments and 50-60 worms per 3 cm plate for oxidative stress experiments. Survival was scored every 1-2 h. Heat stress was imposed by incubation at 35 °C. For oxidative stress experiments, assay plates contained 15.4 mM tert-butyl hydroperoxide (TBHP) and were prepared 12 h before starting the experiment.

Growing worms for RNA-extraction
To obtain synchronized populations, gravid adults were treated with hypochlorite and eggs were allowed to hatch in M9 overnight. ~ 700 L1 larvae per strain were plated on 10 cm NG agar plates seeded with concentrated E. coli OP50 and cultured at the required temperatures (cf. above). At the L4-stage, 500 worms per strain were transferred to two E. coli OP50-seeded 6 cm NG agar plates supplemented with 20 μM FUDR to inhibit germ cell proliferation and progeny production. At day 2 of adulthood, worms were harvested by washing them off their plates with M9. After additional washing with M9 and RNAse-free water, worms were suspended in 1 ml Trizol, snap-frozen in liquid nitrogen and stored at − 80 °C until RNA-extraction.

RNA-extraction
RNA was extracted using Trizol and cleaned up with the Monarch ® Total RNA Miniprep Kit or RNA Cleanup Kit (New England Biolabs) according to the manufacturer's instructions.

RNA-sequencing
RNA quality control, library preparation, and 50 bp single-end RNA-sequencing on the Illumina HiSeq4000 platform was performed at Eurofins Genomics, Ebersberg, Germany. All samples were analyzed in 3 biological replicates and a minimum of 33.2 * 10 6 reads (maximum 49.9 * 10 6 , median 39.7 * 10 6 reads) per replicate were obtained, resulting in at least 16.5x genome coverage (maximum 24.9x, median 20x). A minimum of 91.4% of reads (maximum 96.9%, median 95.7%) could be mapped to the C. elegans reference genome (cf. below).

Overlap of DEG-lists
Overlaps between lists of DEGs were visualized using the Venn and Euler Diagrams-App in Cytoscape (v3.8.2) [58]. Statistical significance of overlaps was calculated as the hypergeometric probability of detecting at least as many common genes as observed in the two lists, using the phyper-function in R. Representation factors were calculated as the number of overlaping genes divided by the expected number of overlapping genes in the two lists (http:// nemat es. org/ MA/ progs/ overl ap_ stats. html; last accessed 12. April 2019). For all calculations, the number of genes in the genome was set to 18,980, i.e. the number of genes in Wormbase WS271 that passed low-count filtering during DESeq2-analysis (cf. above). DEG-lists from published studies were, if necessary, converted to current WBGene-IDs using WormMine (http:// inter mine. wormb ase. org/ tools/ wormm ine/ begin. do; last accessed 04. September 2019) and adjusted to genes in Wormbase WS271 using custom R-scripts and manual curation.

qPCR
One microgram total RNA was reverse-transcribed using LunaScript ® RT SuperMix (New England Biolabs). qPCR-reactions were performed in 2-3 technical replicates in 20 μl reaction volume on an CFX Connect ™ Real-Time PCR Detection System (Bio-Rad Laboratories) with iTaq ™ Universal SYBR ® Green Supermix (Bio-Rad Laboratories). The thermal cycling protocol comprised one activation step at 95 °C for 3 min, followed by 40 cycles of denaturation at 95 °C for 10 s and combined annealing/extension at 60 °C for 30 s. Melting curve analysis was performed from 65 °C to 95 °C with 0.5 °C increments at 5 s per step. Data were analyzed by the ΔΔCt method and target gene expression levels were normalized to the geometric mean of cdc-42, tba-1 and Y45F10D.4 [61,62]. Primer sequences are listed in Additional file 1: Table S8.

Statistical analysis
Statistical analysis was performed using Prism 5 (Graph-Pad Software, San Diego, CA, USA). Details on the particular tests used are specified in the figure legends.
Additional file 1: Table S1. Statistical analysis of stress resistance data. Accompanies Fig. 1. Worms of the strains indicated were subjected to heat or oxidative stress as specified in the individual table headlines. For each biological replicate of the respective experiment, the table lists mean and median survival times in h, standard deviations and standard errors of the mean, the number of worms scored, and p-values from Kaplan-Meier survival analysis with Mantel-Cox tests against the control-, daf-16(−) or mbk-1(−) strain of the same genetic background. In case of the GSC(−) and daf-2(−) control strain, survival analysis was performed relative to the GSC(+) or daf-2(+) control strain. Repeated measures two-way ANOVA with Bonferroni post tests was performed on the mean survival times determined in the individual biological replicates of the experiments. Statistically significant p-values (* p < 0.05, ** p < 0.01, *** p < 0.001) are highlighted in blue. Table S2. Genes regulated by daf-2(−) and GSC(−) in this work and in published studies. For each overlap, the number of genes shared between the two particular lists, the representation factor, and the p-value (in italics; hypergeometric probability) are indicated. Statistically significant overlaps (p < 0.05) are highlighted in blue. For these overlap analyses, lists of published studies were adjusted to WS271 and limited to genes that passed low-count filtering in our DESeq2-analyses. (cf. Methods). See main text for full references of the published studies included in the table. Table S3. Genes differentially expressed between mbk-1(−) and mbk-1(+) worms of the genetic backgrounds analyzed in this study. Differentially expressed genes (DEGs) were defined by p adj < 0.05 in DESeq2-analysis and by the following expression fold-change thresholds: FC < 0.6667 (Log 2 FC < − 0.58) for genes downregulated upon mbk-1 loss. Expression of these genes is normally promoted by mbk-1 (mbk-1 induced DEGs). FC > 1.5 (Log 2 FC > 0.58) for genes upregulated upon mbk-1 loss. Expression of these genes is normally repressed by mbk-1 (mbk-1 repressed DEGs). In addition, the table lists DEGs detected in control comparisons between the following strains (all mbk-1(+)): downregulated < 0.6667-fold or upregulated > 1.5-fold in daf-2(−) relative to daf-2(+); downregulated < 0.25-fold (Log 2 FC < − 2) or upregulated > 4-fold (Log 2 FC > 2) in GSC(−) relative to GSC(+). See Additional file 2 for full DESeq2-data. Table S4. Overlaps of mbk-1 dependent DEGs in the genetic backgrounds analyzed in this study. Accompanies Fig. 2c, d. (a) Pairwise overlaps between DEG-lists. For each overlap, the number of genes shared between the two particular lists, the representation factor, and the p-value (in italics; hypergeometric probability) are indicated. Statistically significant overlaps are highlighted in blue. (b) Absolute numbers of mbk-1 dependent DEGs that are induced/repressed exclusively in the background indicated, or that are also induced/repressed in one or more of the three other backgrounds. (c) Percentages of mbk-1 dependent DEGs that are induced/repressed exclusively in the background indicated, or that are also induced/repressed in one or more of the three other backgrounds. Table S5. Overlap of genes regulated by mbk-1 in this work and by daf-16 in published studies. For each overlap, the number of genes shared between the two particular lists, the representation factor, and the p-value (in italics; hypergeometric probability) are indicated. Statistically significant overlaps (p < 0.05) are highlighted in blue. For these overlap analyses, lists of published studies were adjusted to WS271 and limited to genes that passed low-count filtering in our DESeq2-analyses. (cf. Methods). See main text for full references of the published studies included in the table.  Fig. 2e, f. Note that Fig 2e, f are further limited to GO terms of the bp-category with N observed ≥ 2 and enrichment fold-change ≥ 2 and that some GO terms were not plotted in these figures for the reasons specified in the Comment-column. Table S7. List of C. elegans strains used in this study. Table S8. List of qPCR-primers used in this study. Table S9. DESeq2-analysis of the RNA-seq data set generated in this study. Differential gene expression was determined between mbk-1(−) and mbk-1(+) worms (abbreviated as "-VS+" in column headers) in GSC(+), GSC(−), daf-2(+) and daf-2(−) background. In addition, control comparisons were made between the mbk-1(+) strains of the daf-2(−) and daf-2(+), and the GSC(−) and GSC(+) backgrounds. The table shows DESeq2-results for all genes that passed our filter criteria in at least one of the comparisons (cf. Methods).