1. Chromosomes and Gene Expression
  2. Neuroscience
Download icon

A KDM5–Prospero transcriptional axis functions during early neurodevelopment to regulate mushroom body formation

  1. Hayden AM Hatch
  2. Helen M Belalcazar
  3. Owen J Marshall
  4. Julie Secombe  Is a corresponding author
  1. Dominick P. Purpura Department of Neuroscience Albert Einstein College of Medicine, United States
  2. Department of Genetics Albert Einstein College of Medicine, United States
  3. Menzies Institute for Medical Research University of Tasmania, Australia
Research Article
Cite this article as: eLife 2021;10:e63886 doi: 10.7554/eLife.63886
7 figures, 1 table and 7 additional files

Figures

kdm5140 pharate adults have neuromorphological defects of the mushroom body (MB).

(A) Representative α/β lobe Z projections of pharate wild-type (NP4707rev2 revertant) and kdm5140 strains. Arrows indicate growth defects, and arrowheads indicate guidance defects. The α/β lobes are revealed with anti-fasciclin 2. (B) Quantification of α/β lobe defects in wild-type and kdm5140 strains. n = 19–22 (mean n = 21). ***p<0.001 (chi-square test with Yates’ correction). (C) Western blot of w1118 and kdm5:HA adult heads confirming wild-type expression levels of lysine demethylase 5 (KDM5):HA within our endogenously tagged kdm5:HA strain. Anti-HA (top) and anti-KDM5 (bottom) loading control. (D) Schematic of an adult MB with its associated Kenyon cells (left). OK107-Gal4 is used to drive expression of UNC-84:GFP, an inner nuclear membrane GFP reporter, within MB neuroblasts, mushroom body-ganglion mother cells, and Kenyon cells (right). Brains are counterstained with anti-HA to demonstrate the presence of endogenously tagged KDM5:HA within Kenyon cell nuclei. Scale bars represent 20 μm. 

lysine demethylase 5 (KDM5) is broadly expressed in nuclei of the Drosophila larval central nervous system (CNS) and adult brain.

(A) Maximal Z projection of third-instar larval CNS revealing broad expression of endogenously tagged KDM5:HA and stained with anti-HA and anti-Bruchpilot (anti-Brp). (B) Cortical region of a larval (WL3) brain lobe stained with anti-HA and DAPI, showing nuclear localization of endogenously tagged KDM5:HA. (C) Neuroblast (NB)-specific wor-Gal4 driving expression of UNC-84:GFP to demonstrate endogenously tagged KDM5:HA expression within nuclei of WL3 central brain NBs (marked by *) and presumptive ganglion mother cells. UNC-84:GFP perdures for approximately 2–3 cell divisions. (D) Maximal Z projection of an adult brain revealing broad expression of endogenously tagged KDM5:HA with anti-HA and counterstained for neuropil with anti-Brp. (E) Dorsoanterior cortical region of an adult brain with endogenously tagged KDM5:HA. The nuclear localization of KDM5:HA is revealed by anti-HA, DAPI, and anti-Brp staining. Scale bars represent 50 μm in (A) and (D), 5 μM in (B) and (E), and 20 μM in (C).

Figure 3 with 1 supplement
Depletion of lysine demethylase 5 (KDM5) within neural precursors results in neuroanatomical defects of the mushroom body (MB).

(A) Schematic showing the sequential generation of distinct subclasses of Kenyon cells throughout development. MB neuroblasts (MBNBs) are eliminated via apoptosis (dotted line) immediately prior to eclosion. (B) OK107-Gal4 driving expression of UNC-84:GFP to reveal shRNA-mediated KDM5:HA depletion within adult Kenyon cell nuclei. (C) Representative Z projections of adult kdm5 knockdown animals exhibiting significant α/β lobe defects and their respective kdm5 shRNA and GAL4 controls. The antibody anti-fasciclin 2 is used to visualize α/β lobes. Arrows indicate growth defects, and arrowheads indicate guidance defects. (D) Quantification of α/β MB lobe defects in flies expressing kdm5 shRNA driven by neural progenitor cell- and Kenyon cell-specific drivers. ‘KD’ indicates shRNA-mediated knockdown of kdm5. n = 16–49 (mean n = 29). ****p<0.0001 (chi-square test with Bonferroni correction). (E) Z projection of larval cortex revealing wor-Gal4-driven expression of kdm5 shRNA and unc-84:gfp transgenes. KDM5:HA depletion is observed in presumptive ganglion mother cells and post-mitotic cells surrounding NBs (marked by asterisks). Scale bars represent 20 μm in (B) and (C) and 10 μm in (E).

Figure 3—figure supplement 1
Validation of kdm5 knockdown within subpopulations of mature mushroom body (MB) neurons.

MB Gal4 drivers promoting the expression of CD8:GFP to reveal shRNA-mediated lysine demethylase 5 (KDM5):HA depletion within subpopulations of mature Kenyon cells within the adult brain. Scale bar represents 5 μM.

Expression pattern of R71C09-Gal4 within ganglion mother cells and immature neurons of the Drosophila central nervous system.

(A) Whole-mount Z projections of a larval ventral nerve cord (left), brain lobe (middle), and brain cortical region (right). Z projections show R71C09-Gal4-driven expression of UNC-84:GFP, counterstained with neuroblast (NB)-specific anti-Dpn. (B) Z projection of larval cortex revealing R71C09-Gal4-driven expression of UNC-84:GFP with and without kdm5 shRNA. NBs are marked by an asterisk. (C) Optical sections of an adult mushroom body (MB) with its associated pedunculus, Kenyon cell bodies, and calyces expressing an R71C09-Gal4-driven mVenus reporter. R71C09-Gal4 strongly drives mVenus expression in newly born neurons located within core fibers of the pedunculus. (D) Representative Z projections (left) and quantification (right) of adult α/β MB lobe defects in flies expressing kdm5 shRNA driven by R71C09-Gal4. The antibody anti-fasciclin 2 (anti-Fas2) is used to visualize α/β lobes. n = 15–39 (mean n = 26). Arrowhead indicates a guidance defect. ****p<0.0001 (chi-square test with Bonferroni correction). (E) Representative α/β lobe Z projections and quantification of pharate wild-type (NP4707rev2 revertant), kdm5140, and kdm5140; R71C09 > kdm5 rescue strains. Arrows indicate growth defects. The α/β lobes are revealed with anti-Fas2. n = 20–28 (mean n = 24). **p<0.01; ***p<0.001 (chi-square test with Bonferroni correction). Scale bars represent 20 μm.

Figure 5 with 5 supplements
Transcriptome profiling of lysine demethylase 5 (KDM5)-depleted ganglion mother cells (GMCs) and immature neurons by targeted DamID (TaDa) reveals KDM5-regulatory networks critical for ganglion mother cell (GMC) proliferation and neurodevelopment.

(A) Timeline of TaDa induction within GMCs and immature neurons of kdm5140 WL3 animals and pupae. (B) Volcano plot of differentially expressed genes (DEGs) within GMCs and immature neurons of kdm5140 animals compared to wild type. Genes with a false discovery rate (FDR) < 0.05 are in red, with those labeled involved in GMC proliferation and neurodevelopment. TaDa analyses were performed in quintuplicate. (C) Representative distribution of ontology terms for DEGs in kdm5140 GMCs and immature neurons using a PANTHER Overrepresentation Test (Fisher’s exact test with FDR < 0.05). (D) Timeline of TaDa and kdm5 shRNA induction within GMCs and immature neurons of WL3 animals and pharate adults. (E) Volcano plot of DEGs within kdm5 shRNA GMCs and immature neurons compared to those expressing a scrambled shRNA. Genes with an FDR < 0.05 are in red, with those labeled involved in GMC proliferation and neurodevelopment. TaDa analyses were performed in triplicate. (F) Correlation of Z scores between DEGs of kdm5140 and kdm5 shRNA TaDa datasets (Deming regression; p<0.0001). (G) Correlation of Z scores between overlapping DEGs of kdm5140 and kdm5 shRNA TaDa datasets (Deming regression; p<0.0001). (H) Representative distribution of ontology terms for DEGs in overlapping kdm5140 and kdm5 shRNA TaDa datasets utilizing a PANTHER Overrepresentation Test (Fisher’s exact test with FDR < 0.05).

Figure 5—figure supplement 1
kdm5140 pharate adults expressing targeted DamID (TaDa) genetic elements present with significant mushroom body (MB) morphological defects.

(A) Representative α/β lobe Z projections of pharate kdm5140 animals expressing TaDa genetic elements and their respective controls. The antibody anti-fasciclin 2 is used to visualize α/β lobes. Arrows indicate growth defects. Scale bar represents 20 μm. (B) Quantification of α/β lobe defects in pharate kdm5140 animals expressing TaDa genetic elements and their respective controls. n = 8–22 (mean n = 13). ****p<0.0001 (chi-square test with Bonferroni correction).

Figure 5—figure supplement 2
Knockdown of myc within Kenyon cells does not result in morphological defects of the mushroom body.

Representative α/β lobe Z projections of adults expressing myc dsRNA driven by OK107-Gal4. No gross morphological defects were detected in the KD or control conditions n = 29–30.

Figure 5—figure supplement 3
Chromatin accessibility profiling using targeted DamID (CATaDa) analyses reveal minimal changes to chromatin accessibility within ganglion mother cells (GMCs) and immature neurons upon loss of kdm5.

(A) Representative CATaDa profiles of R71C09-Gal4 expressing cells for kdm5140 and wild type. Heat maps show Dam binding profiles for the greatest 8181 peaks for each genotype. (B) Volcano plot showing changes to chromatin accessibility within GMCs and immature neurons of kdm5140 animals compared to wild type. Genes with an FDR < 0.01 are in yellow. CATaDa analyses were performed in quintuplicate. (C) Volcano plot showing changes to chromatin accessibility for differentially expressed genes (DEGs) from kdm5140 TaDa. In yellow are kdm5140 TaDa DEGs with an FDR < 0.05. In red are kdm5140 TaDa DEGs with associated changes in chromatin accessibility for FDR < 0.01 (Fisher’s exact test, p=4.79E-09).

Figure 5—figure supplement 4
Adults expressing R71C09-Gal4-driven kdm5 shRNA in tandem with targeted DamID (TaDa) genetic elements present with significant mushroom body morphological defects.

(A) Representative α/β lobe Z projections of adult R71C09 > kdm5 shRNA animals expressing TaDa genetic elements and their respective controls. The antibody anti-fasciclin 2 is used to visualize α/β lobes. Arrowheads indicate guidance defects. Scale bar represents 20 μm. (B) Quantification of α/β lobe Z projections of adult R71C09 > kdm5 shRNA animals expressing TaDa genetic elements and their respective controls. n = 16–34 (mean n = 28). ****p<0.0001 (chi-square test with Bonferroni correction).

Figure 5—figure supplement 5
Transcriptome analyses reveal that lysine demethylase 5 (KDM5)-regulated gene expression is tissue-specific.

(A) Venn diagram illustrating intersection of kdm5140 and kdm5 shRNA overlapping targeted DamID (TaDa) data with that from a previously published kdm5140 mRNA-seq wing disc dataset (Drelon et al., 2018). (B) Correlation of Z scores between kdm5140 TaDa and kdm5140 wing disc mRNA-seq (Drelon et al., 2018) datasets. Pearson’s R = 0.09.

Network of known and predicted interactions between lysine demethylase 5 (KDM5)-regulated aenes and analysis of direct KDM5 targets.

(A) Gene network analysis and community clustering were performed using Cytoscape with a minimum confidence score of 0.4. Networks with greater than two nodes are shown and color-coded based on cluster. Labels indicate general categories of overlapping kdm5140 and kdm5 shRNA targeted DamID (TaDa) differentially expressed genes (DEGs) within each cluster. (B) Venn diagram illustrating intersection of similarly dysregulated kdm5140 and kdm5 shRNA overlapping DEGs with direct KDM5 targets from R71C09 > dam-kdm5 TaDa. (C) Analysis of the 319 similarly dysregulated kdm5140 and kdm5 shRNA overlapping DEGs (with values plotted from the kdm5 shRNA TaDa). Direct KDM5 targets are labeled in orange. Venn diagram created with BioVenn (Hulsen et al., 2008).

Neuromorphological and transcriptomic analyses reveal a genetic interaction between prospero and lysine demethylase 5 (KDM5).

(A) Integrative genomics viewer (IGV) plot showing average KDM5 occupancy at pros transcriptional start sites across six replicates. Scale bars represent the log2 ratio change between Dam-KDM5 and Dam samples. Called peaks are indicated by gray bars. (B) Venn diagram illustrating intersection of similarly dysregulated kdm5140 and kdm5 shRNA overlapping differentially expressed genes (DEGs) with Dam-KDM5 direct binding data and a previously published Dam-Pros targeted DamID (TaDa) binding dataset (Fisher’s exact test, p=2.20E-16 for KDM5-dysregulated and Pros gene overlap) (Liu et al., 2020). (C) Analysis of the 319 similarly dysregulated kdm5140 and kdm5 shRNA overlapping DEGs (with values plotted from the kdm5 shRNA TaDa). Direct Pros targets from the previously published Dam-Pros TaDa dataset (Liu et al., 2020) are labeled in orange. (D) Representative Z projections of OK107 > pros RNAi adults exhibiting significant α/β lobe defects. The α/β lobes are revealed by anti-fasciclin 2 (Fas2). (E) Quantification of α/β mushroom body (MB) lobe defects in flies expressing pros shRNA driven by OK107-Gal4. n = 16–19 (mean n = 17). ****p<0.0001 (chi-square test with Bonferroni correction). (F) Representative Z projections of representative kdm5140/+ and pros17/+ heterozygous adult α/β MB lobes (top) and kdm5140/+; pros17/+ transheterozygous adult α/β MB lobes (bottom). The α/β lobes are revealed by anti-Fas2. (G) Quantification of α/β MB lobe defects in kdm5140/+; pros17/+ transheterozygous adults and heterozygous controls. n = 28–37 (mean n = 34). ****p<0.0001 (chi-square test with Bonferroni correction). (H) Model proposing a genetic interaction between pros and kdm5 within ganglion mother cells (GMCs) and immature α/β Kenyon cells. KDM5 binds to the pros locus and positively regulates its transcription. Loss of kdm5 leads to downregulation of pros and its targets, resulting in defects to MB neurodevelopment and cognitive function. Image was created with BioRender and Venn diagrams with BioVenn (Hulsen et al., 2008). Scale bars represent 20 μm.

Tables

Appendix 1—key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
AntibodyMouse monoclonal anti-brpDSHBCat# nc82; RRID:AB_2314866IF (1:50)
AntibodyRabbit monoclonal anti-HA-TagCell Signaling TechnologyCat# 3724; RRID:AB_1549585IF (1:100)
AntibodyMouse monoclonal anti-HA-TagCell Signaling TechnologyCat# 2367; RRID:AB_10691311IF (1:100)
WB (1:1000)
AntibodyRat monoclonal anti-DeadpanAbcamCat# ab195173; RRID:AB_2687586IF (1:100)
AntibodyGoat polyclonal anti-mouse Alexa-488Thermo Fisher ScientificCat# A32723; RRID:AB_2633275IF (1:500)
AntibodyGoat polyclonal anti-mouse Alexa-568Thermo Fisher ScientificCat# A11004; RRID:AB_2534072IF (1:500)
AntibodyGoat polyclonal anti-rabbit Alexa-488Thermo Fisher ScientificCat# A11034; RRID:AB_2576217IF (1:500)
AntibodyGoat polyclonal anti-rabbit Alexa-568Thermo Fisher ScientificCat# A11004, RRID:AB_2534072IF (1:500)
AntibodyGoat polyclonal anti-rat Alexa-568Thermo Fisher ScientificCat# A11077; RRID:AB_2534121IF (1:500)
AntibodyMouse monoclonal anti-Fas2DSHBCat# 1D4; RRID:AB_528235IF (1:25)
AntibodyRabbit polyclonal anti-KDM5Secombe Lab; Secombe et al., 2007RRID:AB_2569502WB (1:1000)
AntibodyIRDye 680RD donkey monoclonal anti-mouse IgG secondary antibodyLI-COR BiosciencesLI-COR Biosciences Cat# 925-68072, RRID:AB_2814912WB (1:8000)
AntibodyIRDye 800CW donkey monoclonal anti-rabbit IgG secondary antibodyLI-COR BiosciencesLI-COR Biosciences Cat# 926-32213, RRID:AB_621848WB (1:8000)
Cell line (Escherichia coli)NEB 5-alpha competent E. coliNew England BioLabsCat# C2987
Commercial assay or kitClontech CloneAmp HiFi PCR PremixClontechCat# 639298
Commercial assay or kitAdvantage 2 cDNA polymeraseClontechCat# 639201
Commercial assay or kitAgencourt AMPure XP BeadsBeckman CoulterCat# A63880
Commercial assay or kitTakara In-Fusion HD Cloning PlusTakaraCat# 638909
Commercial assay or kitQuick​ ​Ligation ​Kit​New England BioLabsCat# M2200S
Commercial assay or kitZymo​ ​Quick-DNA​ ​miniprep​ ​plusZymo ResearchCat#​ ​D4069
Commercial assay or kitMacherey-Nagel​ ​NucleoSpin​ ​Gel​ ​and​ ​PCR​ ​Clean-up​ KitTakaraCat# 740609.250
Commercial assay or kitQubit dsDNA HS Assay KitInvitrogenCat# Q32851
Genetic reagent (Drosophila melanogaster)Drosophila: kdm5:3xHAThis studyN/AEndogenous kdm5:HA strain (Figures 24). Available from lead contact.
Genetic reagent (D. melanogaster)Drosophila: OK107-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_854
Genetic reagent (D. melanogaster)Drosophila: 5XUAS-unc84::2XGFPJanelia Research Campus; Henry et al., 2012N/A
Genetic reagent (D. melanogaster)Drosophila: wor-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_56553
Genetic reagent (D. melanogaster)Drosophila: UAS-kdm5RNAIBloomington Drosophila Stock CenterRRID:BDSC_35706
Genetic reagent (D. melanogaster)Drosophila: insc-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_8751
Genetic reagent (D. melanogaster)Drosophila: c708a-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_50743
Genetic reagent (D. melanogaster)Drosophila: c305a-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_30829
Genetic reagent (D. melanogaster)Drosophila: H24-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_51632
Genetic reagent (D. melanogaster)Drosophila: 201Y-Gal4Bloomington Drosophila Stock CenterRRID:BDSC_4440
Genetic reagent (D. melanogaster)Drosophila: UAS-Dcr-2Bloomington Drosophila Stock CenterRRID:BDSC_24650
Genetic reagent (D. melanogaster)Drosophila: GMR71C09-GAL4Bloomington Drosophila Stock CenterRRID:BDSC_39575
Genetic reagent (D. melanogaster)Drosophila: 20XUAS-IVS-CsChrimson.mVenusBloomington Drosophila Stock CenterRRID:BDSC_55136
Genetic reagent (D. melanogaster)Drosophila: kdm5NP4707Kyoto Stock Center; Hayashi et al., 2002; stock# 104754
Genetic reagent (D. melanogaster)Drosophila: kdm5140Secombe Lab; Drelon et al., 2018
Genetic reagent (D. melanogaster)Drosophila: UASt-kdm5Secombe Lab; Secombe et al., 2007
Genetic reagent (D. melanogaster)Drosophila: UAS-dMycRNAiVienna Drosophila Resource CenterStock# KK106066
Genetic reagent (D. melanogaster)Drosophila: UAS-prosRNAiBloomington Drosophila Stock CenterRRID:BDSC_42538
Genetic reagent (D. melanogaster)Drosophila: tubP-Gal80tsBloomington Drosophila Stock CenterRRID:BDSC_7019
Genetic reagent (D. melanogaster)Drosophila: UAS-LT3-NDamBrand Lab; Southall et al., 2013
Genetic reagent (D. melanogaster)Drosophila: UAS-LT3-NDam-RpII215Brand Lab; Southall et al., 2013
Genetic reagent (D. melanogaster)Drosophila: w1118Bloomington Drosophila Stock CenterRRID:BDSC_5905
Genetic reagent (D. melanogaster)Drosophila: UAS-LT3-NDam-kdm5This studyN/AUsed for KDM5 TaDa (Figures 6 and 7). Available from lead contact.
Sequence-based reagentAdRtVogel et al., 2007PCR primersCTAATACGACTCACTATAGGGCAGCGTGGTCGCGGCCGAGGA
Sequence-based reagentAdRbVogel et al., 2007PCR primersTCCTCGGCCG
Sequence-based reagentDamID_PCRVogel et al., 2007PCR primersGGTCGCGGCCGAGGATC
Sequence-based reagentscram_shRNAThis studyPCR primersGGATAATAGAATAGTTATATTCAAGCATATTCTATTATCC
Sequence-based reagentFw DsRed_KDM5_AarIThis studyPCR primerstatagtgtcttcggggccgaCAGGAGCTGTGGCGCATTCTAGAAAC
Sequence-based reagentPAM RvThis studyPCR primersAATCTGGAACATCGTATGGGTACTGCGGCCGCGCTCGCGC
Sequence-based reagentPAM FwThis studyPCR primersAGCAGCGGGCGGTGCAATCGGCGCGAGCGCGGCCGCAGTA
Sequence-based reagentRv DsRed_KDM5_AarIThis studyPCR primersgattatctttctagggttaaAGGAAAAAGTCAAATAAAACGTAAGAAAACTTTGC
Sequence-based reagentFw DsRed_KDM5_SapIThis studyPCR primersgactatctttctagggttaaTCAAAGGCGAAGGCGACTCT
Sequence-based reagentRv DsRed_KDM5_SapIThis studyPCR primersatatggtcttcttttcccggAACATGTTCCTCTTTTAAGGTGCTCTTT
Sequence-based reagentDam-kdm5_NotI-FwThis studyPCR primerscgcagatctgcggccgATGTCCGCCAAAACTGAGG
Sequence-based reagentDam-kdm5_XbaI-RvThis studyPCR primersacaaagatcctctagCTACCGCGCCGATTGCAC
Recombinant DNA reagentpU6-BbsI-chiRNAAddgene; Gratz et al., 2013RRID:Addgene_45946
Recombinant DNA reagentpValium20Drosophila Genomics Resource Center; Ni et al., 2009DGRC# 1467
Recombinant DNA reagentpHD-ScarlessDsRedDrosophila Genomics Resource CenterDGRC# 1364
Software, algorithmFijihttps://fiji.sc/RRID:SCR_002285
Software, algorithmPrism 6GraphPadRRID:SCR_002798
Software, algorithmCytoscapehttps://cytoscape.org/RRID:SCR_003032
Software, algorithmGene Ontologyhttp://www.geneontology.orgRRID:SCR_002811
Software, algorithmR 3.5.1The R FoundationRRID:SCR_001905
Software, algorithmggplot2 (R package)CRANRRID:SCR_014601
Software, algorithmdamidseq_pipelineMarshall and Brand, 2015ahttp://owenjm.github.io/damidseq_pipeline/
Software, algorithmfind_peaks
Wolfram et al., 2012
http://github.com/owenjm/find_peaks
Software, algorithmbowtie2
Langmead and Salzberg, 2012
http://bowtie-bio.sourceforge.net/bowtie2/index.shtml
Software, algorithmBioVennHulsen et al., 2008https://www.biovenn.nl/
OtherVectashield Mounting MediumVector LabsCat# H-1000
OtherNormal donkey serumFisher ScientificCat# 50-413-367
OtherDAPI-Fluoromount-GSouthernBiotechCat# OB010020
OtherT4 DNA ligaseNew England BioLabsCat# M0202
OtherT4 polynucleotide kinaseNew England BioLabsCat# M0201
OtherAarIThermo ScientificCat# ER1581
OtherSapINew England BioLabsCat# R0569
OtherBpiIThermo ScientificCat# ER1011
OtherNotINew England BioLabsCat# R0189
OtherPstINew England BioLabsCat# R0140
OtherNheINew England BioLabsCat# R0131
OtherEcoRINew England BioLabsCat# R0101
OtherDpnI​New England BioLabs​Cat# R0176
OtherDpnI​INew England BioLabsCat# R0543
OtherAlwI​New England BioLabsCat# R0513

Additional files

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)