Loxl2 is a mediator of cardiac aging in Drosophila melanogaster, genetically examining the role of aging clock genes

Abstract Transcriptomic, proteomic, and methylation aging clocks demonstrate that aging has a predictable preset program, while transcriptome trajectory turning points indicate that the 20–40 age range in humans is the likely stage at which the progressive loss of homeostatic control, and in turn aging, begins to have detrimental effects. Turning points in this age range overlapping with human aging clock genes revealed five candidates that we hypothesized could play a role in aging or age-related physiological decline. To examine these gene’s effects on lifespan and health-span, we utilized whole body and heart-specific gene knockdown of human orthologs in Drosophila melanogaster. Whole body lysyl oxidase like 2 (Loxl2), fz3, and Glo1 RNAi positively affected lifespan as did heart-specific Loxl2 knockdown. Loxl2 inhibition concurrently reduced age-related cardiac arrythmia and collagen (Pericardin) fiber width. Loxl2 binds several transcription factors in humans and RT-qPCR confirmed that a conserved transcriptional target CDH1 (Drosophila CadN2) has expression levels which correlate with Loxl2 reduction in Drosophila. These results point to conserved pathways and multiple mechanisms by which inhibition of Loxl2 can be beneficial to heart health and organismal aging.


Introduction
Molecular aging clocks use transcriptomics (Mamoshina et al. 2018;Gonzá lez-Velasco et al. 2020), proteomics (Lehallier et al. 2019;Johnson et al. 2020), and epigenetic markers (Horvath 2013;Bergsma and Rogaeva 2020;Galkin et al. 2021) to predict a subject's age with high accuracy. In addition, studies examining the transcriptomes and proteomes during aging have shown that the majority of genes change their expression trends (transcriptome trajectory turning points) between ages 20 and 40 in multiple tissues (Somel et al. 2010;Skene et al. 2017;Lehallier et al. 2019). The age range of these expression turning points corresponds to the time at which most adults will begin to visibly detect aging phenotypes ranging from skin wrinkles to graying hair (Albert et al. 2007;Panhard et al. 2012). We, therefore, hypothesized that a number of the molecules having turning points in this age range should present in aging clock data, and play a role in aging or age-related phenotypes.
Through the examination of aging clocks (Horvath 2013;Lehallier et al. 2019) and transcriptome trajectory turning points (Skene et al. 2017), we identified Lysyl oxidase like 2 (Loxl2). Lysyl oxidases oxidatively deaminate lysine and hydroxylysine residues forming aldehydes on collagens which can then covalently bond, cross-linking the collagens and forming fibers (de Jong et al. 2011;Erasmus et al. 2020). Collagen is not only the most abundant protein in the extracellular matrix (Frantz et al. 2010), but when excessive fibrillar collagen accumulates in the heart it is recognized as cardiac fibrosis (Manabe et al. 2002) and the incidence of cardiac arrhythmia is frequently dependent on collagen texture (de Jong et al. 2011).
Loxl2 also binds transcription factors in the nucleus to inhibit and promote the transcription of genes that play a role in age-related diseases ranging from cancers to cardiovascular diseases (Wen et al. 2020). In human cells, Loxl2 binds the snail transcription factor to regulate E-Cadherin 1 (CDH1) expression (Peinado et al. 2005), and it has been demonstrated in flies that E-cadherin is dependent on the snail1 homolog escargot (esg; Tanaka-Matakatsu et al. 1996). Cadherins, Snail1, and LOXL2 are major components in the endothelial-to-mesenchymal transition (EMT; Cuevas et al. 2014) and EMT is a key modulator of age-related diseases including cancer and cardiovascular disease (Brabletz et al. 2018;Santos et al. 2019;Chen et al. 2020). Further linking Loxl2, cadherins, and collagen is the observation that the EMT program is highly correlated with epigenetic regulation that remodels the ECM. (Peixoto 2019). We therefore aimed to examine the multifaceted role of Loxl2 in aging and aging phenotypes using Drosophila melanogaster.

Lifespan assay
All assays were performed in females unless noted. Fly lines are listed in Supplementary Table S1. Flies were maintained at 25 , 60% relative humidity, and 12-h light/dark cycle in vials containing 25 adults. Larvae were reared on a standard cornmeal and yeast-based diet. The standard cornmeal diet consists of the following master mix of materials pumped into vials: water (1700 ml), agar (15.8 g), yeast (50 g), Cornmeal (104 g), sugar (220 g), TEGOSEPT (4.76 g), and 95% EtOH (18.4 ml). Five-day old adults received standard food with either RU486 mifepristone in 95% ethanol to induce RNAi (200 mM) or 95% ethanol in control food. Food vials were exchanged every 2-3 days. Calculations and chart were generated using Log Rank Mantel Cox test and JMP statistical discovery software. Graphs were generated using GraphPad Prism version 7.03.

Fluorescent immunohistochemistry
Fly abdomens were dissected in 1Â PBS, fixed in 4% paraformaldehyde diluted in 0.3% 1Â Phosphate-buffered saline plus Triton X-100 (PBST) for 20-30 min. Samples were washed in PBST, blocked with PBST plus Donkey or Goat Serum Albumin for 1 h, and then washed in PBST three times for 10 min each. The samples were incubated with anti-Pericardin (Developmental Studies Hybridoma Bank # EC11) at 2 lg/ml overnight at 4 C, washed in PBST three times for 10 min each, then incubated with Alexa Fluor 594 anti-mouse IgG (Jackson ImmunoResearch # 115-585-166) for 2 h at room temperature. Samples were washed in PBST three times for 10 min each. Hoechst (Immunochemistry Technologies # 639) and Alexa Fluor 488 Phalloidin (Thermo Fisher scientific # 12379) were applied for 30 min followed by washing in 1Â PBS and mounted in ProLong Diamond Antifade Mountant (Invitrogen # P36961) according to manufacturer's protocols.

Confocal microscopy and image analysis
Images were taken on an Olympus FV3000 confocal microscope. Filament widths were quantified with CellSens Software version 2.2. Images finalized using ImageJ Fiji v1.53c.

RT-qPCR
RNA from 10 flies per replicate was extracted using TissueLyser II (Qiagen), Trizol, chloroform, and isopropanol, then washed in RNAse free ethanol. DNase treatment (Turbo DNA-free, Ambion # AM1907) was performed following manufacturer's protocol. RNA was quantified using Thermo Scientific Nanodrop Lite Spectrophotometer. cDNA was synthesized using qScript cDNA Synthesis Kit (Quanta Bio # 95047-100) on the Applied Biosystems ProFlex PCR System. qPCR was performed using primers found in Supplementary Table S2 and PowerUp SYBR Green Master Mix (Thermo Fisher # A25742) on Applied Biosystems QuantStudio 3. Expression was calculated relative to Ribosomal protein L32 (RpL32).

Overlapping genes in aging clocks
We first compared the list of genes that make up the Horvath methylation epigenetic aging clock (Horvath 2013) with a blood plasma proteomic clock (Lehallier et al. 2019). We next selected genes that have transcriptome trajectory turning points occurring between ages 19 and 41 in the brain (Skene et al. 2017; Figure 1A). These ages were selected because the majority of transcriptome trajectory turning points, and an end of key developmental timepoints, occur within this window (Somel et al. 2010;Skene et al. 2017). Only five genes were found to overlap between the brain turning points, plasma proteomic clock, and the methylation clock; Target Of Myb1 Like 1 Membrane Trafficking Protein (TOM1L1), Secreted Frizzled Related Protein 1 (SFRP1), Loxl2, Glyoxalase I (GLO1), and CXADR Ig-Like Cell Adhesion Molecule (CXADR; Figure 1B).
Although at first glance these seem relatively unrelated, however, a gene network of the five genes in a STRING-db analysis suggests that the five genes may function around a TNF/WNT3A axis ( Figure 1C). The genes are also target genes of the CAMP Responsive Element Binding Protein 1 transcription factor (Rouillard et al. 2016), possibly indicating a common regulatory axis.
We employed the model organism D. melanogaster due to its short lifespan and extensively available genetic toolkits to examine the age-related effects of these genes. We used Flybase (Larkin et al. 2021) to identify Drosophila genes with the highest ortholog scores for these five human genes ( Figure 1B). Human CXADR had low homology with Drosophila proteins and was excluded from further study.

Loxl2, fz3, and Glo1 reduction affects lifespan
To determine if the four genes may be relevant to Drosophila aging, UAS RNAi fly lines for the candidate genes were crossed with whole body daughterless-GeneSwitch GAL4 (da(GS)-GAL4) lines. RNAi was initiated by RU486 (mifepristone) chemical feeding in food starting on day five post-eclosion (to avoid impacting development) and continued until death. RNAi of Loxl2, fz3, and Glo1 significantly affected lifespan of female flies while CG3529 (TOM1L1 homolog) did not (Figure 2, A-D). Mean lifespan increased in the three fly lines by 3.7%, 6.4%, and 9.1%, respectively ( Figure 2E).

Loxl2 RNAi reduces Pericardin filament width with age
Lysl oxidases function to deaminate collagen fibers, allowing their crosslinking (Erasmus et al. 2020). We, therefore, examined the effect of Loxl2 on Drosophila collagen IV; Pericardin. Pericardin (prc) levels and filament width have been shown to increase with age, and reducing collagen-interacting protein SPARC increases lifespan while its overexpression increases arrhythmicity (Vaughan et al. 2018). We immuno-stained for Pericardin (Figure 3, A-D) and measured Pericardin filament width in alary muscles near the heart tube ( Figure 3, E-H). Quantification revealed that filament widths increased with age and RNAi for Loxl2 significantly reduced this effect ( Figure 3I). Western blots against Pericardin were then performed to determine if the filament width may be a result of changing Pericardin levels ( Figure 3J, Supplementary Figure S1), but no significant difference of Pericardin levels between age matched þ/ÀRU fed lines was seen ( Figure 3K).

Loxl2 RNAi improves cardiac arrhythmia
Cadherins play a key role in cell matching during Drosophila heart development (Zhang et al. 2018). CadN2 is an underexplored protein in Drosophila and is the best matched ortholog to 17 Cadherins in humans (Larkin et al. 2021 RNAi flies þ/ÀRU induction (starting on day 5) from four time points in the Drosophila lifespan. CadN2 was significantly upregulated when Loxl2 is repressed at week three and seven; one outlier was removed from each sample before analysis. Unpaired t-test *P 0.05, **P 0.01, *** P 0.001, **** P 0.0001, ns, not significant.
CDH5 (Bouwens et al. 2020), and CDH13 (Teng et al. 2015;Verweij et al. 2017)-play a role in, or correlate with, cardiovascular disease. Furthermore, Loxl2 has been implicated in collagen-related cardiovascular fibrosis, in particular, its reduction can reduce not only cardiac fibrosis (Yang et al. 2016) but also vascular stiffening (Steppan et al. 2019) with age in mammals.
To test if Loxl2 reduction could affect the Drosophila heart we made use of the SOHA software (Ocorr et al. 2009), to measure cardiac arrhythmia indices; which can be used as measures of heart health and cardiac aging (Ocorr et al. 2007). SOHA on crosses of UAS-Loxl2 RNAi flies with whole body daughterless-GeneSwitch GAL4 (da(GS)-GAL4) and RU486 feeding starting on day five revealed that the arrythmia indices of knockdown flies significantly improved vs. controls at aged timepoints ( Figure 5A).
Due of the benefit of Loxl2 RNAi on the heart, we employed a Hand-GeneSwitch GAL4 (Hand(GS)-GAL4) heart-specific driver cross with UAS-Loxl2 RNAi to test if the lifespan benefits may be partially derived from heart Loxl2. We saw a significant mean lifespan benefit in both male and female flies ( Figure 5, B and C).

Discussion
This work is the first to demonstrate that the reduction of Loxl2 has a protective mechanism in the Drosophila heart and extends mean lifespan, consistent with Loxl2's role in mammalian cardiovascular disease (Yang et al. 2016;Steppan et al. 2019). We also show inhibiting Loxl2 can reduce Pericardin filament width and increase CadN2 expression. Our future work aims to decouple their respective contribution to the protective effects of Loxl2 reduction during aging.
Our data suggest that Loxl's action as a transcriptional regulator may be conserved, which we can now explore in more detail utilizing the advantages of the Drosophila genetic system (McGuire et al. 2004;Allocca et al. 2018;Heigwer et al. 2018). Loxl2's known interactions indicate that its role in aging could be through EMT which involves a transitional state that may be partially activated during aging (Brabletz et al. 2018) and interestingly EMT is even being pursued in age-related telomere shortening (Imran et al. 2021). Further investigation into the extent Loxl2 regulation in the ECM and EMT pathways in Drosophila plays with age, particularly the molecular partners and signaling of Loxl2 in the ECM and nucleus, can now be genetically explored in Drosophila. It is tempting to speculate that aging clocks and transcriptomic trajectory turning points may be picking up signals of the EMT process at a critical timepoint (20-40 years), the time at which aging phenotypes begin to be seen in humans.
generally occurs with age, although it is not fully understood what causes this increase in variation between individuals with age (even when the population starts from the same or similar genotypes). It may be that the effects of epigenetic changes during development generate phenotypic and molecular variations that are exaggerated with age (Zhang et al. 2020). The reduction of Loxl2 may help delay certain transcriptional aspects of aging and in turn delay the effects of those variations. Although we cannot totally rule out off target effects of the RNAi, according to the Harvard Transgenic RNAi Project led by Norbert Perrimon group, there were no off targets predicted for the RNAi line used (Perkins et al. 2015). We confirmed Loxl2 knockdown with RT-qPCR (Supplementary Figure S2), and the Drosophila RNAi phenotypes are consistent with the known crosslinking functions and cardiac effects in mouse Loxl2 knockdown models (Yang et al. 2016;Martínez Rodríguez and Gonzá lez 2019). It is also true that some drivers do have an effect on lifespan when combined with RU feeding in mated female flies (Landis et al. 2015), the GAL4 drivers utilized within this manuscript have been previously tested for cardiac arrythmia, lifespan, and stress impacts under RU feeding (Shaposhnikov et al. 2015;Cannon et al. 2017). Several lines of evidence including reduced lifespan or increased arrythmia under RU for several crosses from the aforementioned studies, as well as our male Loxl2 and female CG3529 lifespan data, provide support that these drivers are less likely to be affected by a positive sexually dimorphic effect of RU feeding on lifespan.
Because we saw cardiac aging related phenotypes in our Loxl2 data, but no turning point studies have been done on the heart, we wanted to see if Loxl2 had turning points in the human heart. We utilized the GTEx Portal [from The Genotype-Tissue Expression (GTEx) Project] to determine if Loxl2 had a transcriptome trajectory turning point in the human atria or ventricles within the timeframe that was seen in the brain or blood transcriptome trajectory turning point studies (Skene et al. 2017;Lehallier et al. 2019). Importantly, Loxl2 changes its average expression trajectory in the heart chambers in the 30-39 age decile consistent with the majority of transcriptome trajectory turning points seen in the blood study (Lehallier et al. 2019). Further turning point studies using human heart data would reveal other molecules important in cardiovascular aging.

Data availability
Fly lines are available upon request. The authors affirm that all data necessary for confirming the conclusions of this article are represented fully within the article, its tables, and figures.
Supplementary material is available at G3 online.