1. Microbiology and Infectious Disease
  2. Neuroscience
Download icon

Neuronal hyperexcitability is a DLK-dependent trigger of herpes simplex virus reactivation that can be induced by IL-1

  1. Sean R Cuddy
  2. Austin R Schinlever
  3. Sara Dochnal
  4. Philip V Seegren
  5. Jon Suzich
  6. Parijat Kundu
  7. Taylor K Downs
  8. Mina Farah
  9. Bimal N Desai
  10. Chris Boutell
  11. Anna R Cliffe  Is a corresponding author
  1. Department of Microbiology, Immunology and Cancer Biology, University of Virginia, United States
  2. Neuroscience Graduate Program, University of Virginia, United States
  3. Department of Pharmacology, University of Virginia, United States
  4. MRC-University of Glasgow Centre for Virus Research (CVR), Garscube Campus, United Kingdom
Research Article
Cite this article as: eLife 2020;9:e58037 doi: 10.7554/eLife.58037
7 figures, 1 table and 2 additional files

Figures

HSV-1 Reactivation from sympathetic neurons is induced by adenylate cyclase activation.

(A) Schematic of the primary sympathetic superior cervical ganglia (SCG)-derived model of HSV latency. Reactivation was quantified based on Us11-GFP-positive neurons in presence of WAY-150168, which prevents cell-to-cell spread. (B) Schematic of the cellular pathways activated by forskolin treatment. Forskolin can act both intracellularly to activate adenylate cyclase (AC) and increasing the levels of cAMP or extracellularly on ion channels. (C) Numbers of Us11-GFP-positive neurons following addition of either forskolin (60 μM) or cell-impermeable dideoxy-forskolin (60 μM) treatment of latently-infected sympathetic neurons. (D) Numbers of Us11-GFP-positive neurons following treatment with a cAMP mimetic 8-Bromo-cAMP (125 μM). (E) Reactivation, quantified by Us11-GFP-positive neurons, was induced by forskolin in the presence or absence of the adenylate cyclase inhibitor SQ22,536 (50 μM). In C-E each point represents a single biological replicate, and the mean and standard errors of the mean (SEM) are also shown. In D statistical comparisons were made using an unpaired t-test. In C and E statistical comparisons were made using a one-way ANOVA with a Tukey's multiple comparisons test. *p<0.05, **p<0.01.

Figure 2 with 2 supplements
Reactivation triggered by forskolin involves a DLK/JNK-dependent phase I of viral gene expression.

(A) Reactivation was induced by forskolin in the presence of JNK inhibitor SP600125 (20 μM). (B) Reactivation was induced by forskolin in the presence of the DLK inhibitor GNE-3511 (4 μM). In A and B each experimental replicate is shown. (C) Reactivation was induced by forskolin or superinfection with a wild-type (F strain) HSV-1 (MOI of 10 PFU/cell) and qualified based on Us11-GFP-positive neurons (n = 3). (D–F) RT-qPCR for viral mRNA transcripts following forskolin treatment of latently infected SCGs. (G–I) RT-qPCR for viral lytic transcripts at 20 hr post-forskolin treatment and in presence of the JNK inhibitor SP600125 (20 μM) and the DLK inhibitor GNE-3511 (4 μM). (J) Neurons were transduced with a non-targeting shRNA control lentivirus or two independent lentiviruses expressing shRNAs that target DLK (shDLK-1, shDLK-2). Western-blotting for DLK or β-III tubulin was carried out 3 days post transduction. The percentage knock-down of DLK normalized to β-III tubulin is shown. (K and L) RT-qPCR for viral mRNA transcripts following forskolin treatment of latently infected SCGs that were either transduced with the shRNA control or shRNA DLK lentiviruses. In D-I, K, and L, each experimental replicate is represented. Statistical comparisons were made using a one-way ANOVA with a Tukey’s multiple comparison. *p<0.05, **p<0.01, ***p<0.001. The mean and SEM are shown.

Figure 2—source data 1

Quantification of GFP-positive neurons, RT-qPCR and western blot band densities for Figure 2.

https://cdn.elifesciences.org/articles/58037/elife-58037-fig2-data1-v3.xlsx
Figure 2—figure supplement 1
Reactivation triggered by forskolin triggers a wave of lytic gene expression that precedes DNA Replication and infectious virus production.

(A) Titers of infectious virus detected from reactivating neurons induced with forskolin (n = 4). (B) Quantification of the relative viral genome copy number following forskolin-mediated reactivation based on gC copy number normalized to cellular GAPDH and expressed relative to the 0 hr time-point (n = 7). (C–E) RT-qPCR for viral mRNA transcripts following forskolin treatment of latently infected SCGs. (F) Quantification of Us11-GFP neurons and (G) RT-qPCR for UL30 mRNA transcript following forskolin treatment of latently infected SCGs that were either transduced with the shRNA control or shRNA DLK lentiviruses. Statistical comparisons were made using a one-way ANOVA with a Tukey’s multiple comparison. *p<0.05, **p<0.01, ***p<0.001. (C–E) In C-G each biological replicate is represented. The means and SEMs are shown.

Figure 2—figure supplement 1—source data 1

Quantification of HSV titer, GFP-positive neurons and RT-qPCR for Figure 2—figure supplement 1.

https://cdn.elifesciences.org/articles/58037/elife-58037-fig2-figsupp1-data1-v3.xlsx
Figure 2—figure supplement 2
Effect of PKA, CREB, Rapgef2 and EPAC Inhibition on HSV-1 Reactivation.

(A) Latently infected cultures were reactivated with forskolin (60 μM) in the presence of the PKA inhibitor KT 5720 (3 µM) and the number of Us11-GFP-positive neurons quantified at 3 days post-reactivation. (B) RT-qPCR for the viral lytic transcript ICP27 at 20 hr post-forskolin treatment and in the presence of KT 5720. (C) Latently infected cultures were reactivated with forskolin in the presence of the CREB inhibitor 666–15 (2 µM). (D) RT-qPCR for ICP27 at 20 hr post-forskolin treatment and in the presence of 666–15. (E) Latently infected cultures were reactivated with forskolin (60 μM) in the presence of the EPAC inhibitor ESI09 (10 µM). (F) Latently infected cultures were reactivated with 8-Bromo-cAMP (125 μM) in the presence of the Rapgef2 inhibitor SQ22,536 (50 µM). Individual experimental replicates are represented along with the means and SEMs. Statistical comparisons were made using a one-way ANOVA with a Tukey’s multiple comparison. *p<0.05, **p<0.01, ***p<0.001.

Figure 3 with 1 supplement
The Initial wave of viral lytic gene expression during forskolin-mediated reactivation is independent on histone demethylase activity.

(A) Quantification of the percentage of genome foci stained using click-chemistry that co-localize with H3K9me3/S10p. At least 15 fields of view with 1–8 genomes per field of view were blindly scored from two independent experiments. Data are plotted around the median, with the boxes representing the 25th–75th percentiles and the whiskers the 1st-99th percentiles. (B) Representative images of click-chemistry based staining of HSV-EdC genomes and H3K9me3/S10p staining at 5 hr post-forskolin treatment. (C and D). Effect of the LSD1 inhibitors OG-L002 and S 2101 on forskolin-mediated Phase I of reactivation determined by RT-qPCR for ICP27 (C) and gC (D) viral lytic transcripts at 20 hr post-forskolin treatment and in the presence of 15 μM OG-L002 and 20 μM S 2102. (E) Effect of the JMJD3 and UTX inhibitor GSK-J4 (2 μM) on forskolin-mediated Phase I measured by RT-qPCR for viral lytic transcripts ICP27 (E) and gC (F) at 20 hr post-forskolin treatment and in the presence of GSK-J4. For C-F each experimental replicate along with the mean and SEM is represented. (C–F). Statistical comparisons were made using a one-way ANOVA with a Tukey’s multiple comparison. *p<0.05, **p<0.01, ***p<0.001.

Figure 3—source data 1

Quantification of genome co-localization and RT-qPCR for Figure 3.

https://cdn.elifesciences.org/articles/58037/elife-58037-fig3-data1-v3.xlsx
Figure 3—figure supplement 1
Forskolin induces hyperexcitability-associated chromatin changes, and hsv reactivation that requires histone demethylase.

(A) SCG neurons were treated with forskolin and immunofluorescence staining was carried out for H3K9me3/S10p, the DNA damage marker γH2AX and the neuronal marker beta III-tubulin. (B) Quantification of neuronal nuclear staining intensity for H3K9me3 (>150 cells/condition). (C) Quantification of neuronal nuclear staining for γH2AX. In B and C data are plotted around the median and whiskers represent the 2.5–97.5 percentile range. (D). Western blotting for pS475-AKT, total AKT, pS73-c-Jun and tubulin at 15 hr post-treatment with the PI3-kinase inhibitor LY294002 (20 µM) or forskolin (60 µM) (E). Effect of the LSD1 inhibitors OG-L002 (15 μM) and S 2101 (20μM) on forskolin-mediated reactivation measured by Us11-GFP-positive neurons (F). Effect of the JMJD3 and UTX inhibitor GSK-J4 (2 μM) on forskolin-mediated reactivation measured by Us11-GFP-positive neurons (G). Statistical comparisons were made using a one-way ANOVA with a Tukey’s multiple comparison. *p<0.05, **p<0.01, ***p<0.001 (E, F). In E and F individual experimental replicates are shown along with the mean and SEM.

Figure 3—figure supplement 1—source data 1

Quantification of nuclear staining intensity and GFP-positive neurons Figure 3—figure supplement 1.

https://cdn.elifesciences.org/articles/58037/elife-58037-fig3-figsupp1-data1-v3.xlsx
Figure 4 with 1 supplement
HSV Reactivation Mediated by Forskolin Requires Neuronal Excitability.

(A) Latently infected cultures were reactivated with forskolin in the presence of the voltage-gated sodium channel blocker tetrodotoxin (TTX; 1 µM) and the number of Us11-GFP-positive neurons quantified at 3 days post-reactivation. (B) Latently infected cultures were reactivated with forskolin in the presence of the voltage-gated potassium channel blocker tetraethylammonium (TEA; 10 mM) and the number of Us11-GFP-positive neurons quantified at 3 days post-reactivation. (C) Forskolin-mediated reactivation in the presence of the HCN channel blockers ZD 7288 (10μM) quantified as the numbers of Us11-GFP-positive neurons at 3 days post-reactivation. (D) The effect of ZD 7288 on the HSV lytic gene transcript ICP27 during Phase I reactivation measured at 20 hr post-forskolin treatment by RT-qPCR. In A-D individual experimental replicates are represented along with the mean and SEM. (E and F) Quantification of the relative nuclear staining for H3K9me3/S10p and γH2AX in SCG neurons at 5 hr post-forskolin treatment and in the presence of ZD 7288 from >800 cells/condition from two independent experiments. Data are plotted around the mean, with the boxes representing the 25th-75th percentiles and the whiskers the 5st-95th percentiles. Statistical comparisons were made using a one-way ANOVA with a Tukey’s multiple comparison (A–D) or two-tailed unpaired t-test (E–F). *p<0.05, **p<0.01, ***p<0.001. In A-D individual experimental replicates are represented.

Figure 4—source data 1

Quantification of GFP-positive neurons, RT-qPCR and nuclear staining intensity for Figure 4.

https://cdn.elifesciences.org/articles/58037/elife-58037-fig4-data1-v3.xlsx
Figure 4—figure supplement 1
HSV Reactivation Mediated by Forskolin Requires Neuronal Excitability.

(A and B) Latently infected cultures were reactivated with forskolin in the presence of the HCN channel inhibitors ivabradine (20 µM; A) and CsCl (3 mM; B). Latently infected cultures were reactivated with forskolin in the presence of the HCN inhibitor ZD 7288 (10 µM) and viral lytic transcripts measured at 20 hr post-reactivation (C and D). Individual experimental replicates are represented in addition to the mean and SEM. Statistical comparisons were made using a one-way ANOVA with a Tukey’s multiple comparison. *p<0.05, **p<0.01.

HSV Reactivation triggered by prolonged neuronal hyperexcitability is DLK/JNK-dependent.

(A) Latently infected SCG cultures were treated with forskolin or KCl (55 mM) for the indicated times followed by wash-out. Reactivation was quantified by number of Us11-GFP-positive neurons at 3 days after the initial stimulus was added. (B) Latently infected neurons were placed in tetrodotoxin (TTX; 1 μM) for 2 days and the TTX was then washed out. At the time of wash-out the JNK inhibitor SP600125 (20 μM) or DLK inhibitor GNE-3511 (4 μM) was added. (C) Latently infected neurons were transduced with either control non-targeting shRNA or shRNA targeting DLK for 3 days, then placed in tetrodotoxin (TTX; 1 μM) for 2 days and the TTX was then washed out. Reactivation was quantified at 3 days post-wash-out. Individual experimental replicates, the mean and SEMs are represented. Statistical comparisons were made using a one-way ANOVA with a Tukey’s multiple comparison. **p<0.01, ***p<0.001.

Figure 6 with 1 supplement
IL-1β Treatment of sympathetic neurons results in changes consistent with heightened neuronal excitability.

(A) Adult P36 SCG neurons were treated with IL-1β (30 ng/mL) for 15 hr and stained for H3K9me3/S10p, γH2AX and beta II-tubulin to mark neurons. (B and C) Quantification of the intensity of H3K9me3/S10p (B) and γH2AX (C) staining following 15 of IL-1β treatment and in the presence of tetrodotoxin (TTX; 1 μM) or anti-IL1 receptor (IL-1R) blocking antibody (2 μg/mL). Data are plotted around the median and whiskers represent the 5th-95th percentiles. (D) Representative images of cytosolic Ca2+ elevations measured using Fura-2-AM in neurons stimulated with 100 µM acetylcholine either pre-treated with IL-1β for 20 hr or mock treated. As a control the neurons were also treated with Ionomycin at the end of the protocol. Bar = 100 μm. (E) Representative experiment for cytosolic Ca2+ elevations in neurons stimulated with 100 µM acetylcholine. Cells were pretreated with IL-1β or vehicle for 20 hr prior to imaging. The plotted values were calculated as a change in fluorescence/initial fluorescence (ΔF/F0). Error bars represent SEM (IL-1β treatment, n = 58 cells and vehicle control, n = 25 cells). (F) Peak cytosolic Ca2+ elevations normalized to untreated controls in neurons stimulated with 100 µM acetylcholine. Cells were pretreated with IL-1β (n = 70, wells) or vehicle (n = 58, wells) for 20 hr prior to imaging. IL-1R blocking antibody (n = 54, wells) was also added. Data points represent individual wells, horizontal line represents mean. Statistical comparisons were made using a one-way ANOVA with a Tukey’s multiple comparison (B–D). ***p<0.001 ****p<0.0001.

Figure 6—source data 1

Quantification of nuclear staining intensity and ratiometric calcium imaging for Figure 6.

https://cdn.elifesciences.org/articles/58037/elife-58037-fig6-data1-v3.xlsx
Figure 6—figure supplement 1
IL-1β Treatment of mature scg neurons induces excitability-associated histone post-translational modifications (supplement to Figure 6).

Quantification of the nuclear staining intensity in P36 sympathetic neurons for H3K9me3/S10 (A) and γH2AX (B) following treatment with IL-1β (30 ng/mL) from 150 nuclei from two independent experiments. Data are plotted around the median and whiskers represent the 5th-95th percentile.

IL-1β-Induced HSV Reactivation is linked to heightened neuronal excitability and dlk activation.

(A) Quantification of Us11-GFP expressing neurons following addition of IL-1β to latently infected cultures of mature SCG neurons. (B) Numbers of Us11-GFP-positive neurons following addition of forskolin or IL-1β to mature SCG neurons, and in the presence of an IL-1R-blocking antibody (2 μg/mL). (C) Quantification of IL-1β induced reactivation in the presence of the voltage-gated sodium channel blocker TTX (1 μM), the HCN-channel blocker ZD 7288 (10 μM) and the DLK inhibitor GNE-3511 (4 μM). (D) Latently infected SCG neurons were transduced with an shRNA control lentivirus or lentiviruses expressing shRNA against DLK. Three days later IL-1β was added to the cultures and the numbers of GFP-positive neurons quantified at 3 days later. Individual experimental replicates, means and SEMs are represented. Statistical comparisons were made using two-tailed unpaired t-test (A) or a one-way ANOVA with a Tukey’s multiple comparison (B–D). *p<0.05, **p<0.01.

Tables

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Strain, strain background (Mus musculus, M/F)CD1Charles RiverCrl:CD1(ICR)
Strain, strain background (Human herpesvirus 1)HSV Us11-GFPI gift from Ian Mohr, NYU. PMID:12915535
Strain, strain background (Human herpesvirus 1)HSV-1 17syn+A gift from Roger Everett, MRC Virology Unit Glasgow
Cell line (Homo sapiens)293LTVCell BiolabsCat # LTV-100
RRID:CVCL_JZ09
Cell line (Cercopithecus aethiops)VeroATCCCat # CCCL-81
RRID:CVCL_0059
Recombinant DNA reagentpCMV-VSV-GA gift from Bob Weinberg/Addgene PMID:12649500Cat # 8454
RRID:Addgene_8454
Recombinant DNA reagentpsPax2A gift from Didier Trono/AddgeneCat # 12260
RRID:Addgene_12260
AntibodyAnti-phospho-Akt (S473) (Rabbit monoclonal)Cell Signalling TechnologiesCat # 4060
RRID:AB_2315049
WB (1:500)
AntibodyAnti-Akt (pan) (Rabbit monoclonal)Cell Signalling TechnologiesCat # C67E7
RRID:AB_915783
WB (1:1000)
AntibodyAnti-phopsho-c-Jun (Rabbit monoclonal)Cell Signalling TechnologiesCat # 3270
RRID:AB_2129575
WB (1:500)
AntibodyAnti-DLK/MAP3K12 (Rabbit polyclonal)Thermo FisherPA5-32173
RRID:AB_2549646
WB (1:500)
AntibodyAnti-a-tubulin (Mouse monoclonal)Millipore sigmaCat # T9026
RRID:AB_477593
WB (1:2500)
AntibodyAnti-Rabbit IgG Antibody (H+L), Peroxidase (Goat polyclonal)Vector LabsCat # PI-1000
RRID:AB_2336198
WB (1:10000)
AntibodyAnti-mouse IgG Antibody (H+L), Peroxidase (Horse polyclonal)Vector LabsCat # PI-2000
RRID:AB_2336177
WB (1:10000)
AntibodyAnti- H3K9me3S10P (Rabbit polyclonal)AbcamCat # Ab5819
RRID:AB_305135
IF (1:250)
AntibodyAnti-Beta-III Tubulin (Chicken polyclonal)Millipore SigmaCat # AB9354
RRID:AB_570918
IF (1:1000)
AntibodyAnti-γH2A.X (Mouse monoclonal)Cell Signalling TechnologiesCat # 80312
RRID:AB_2799949
IF (1:100)
AntibodyAnti-c-Fos (Rabbit polyclonal)NovusCat # NB110-75039
RRID:AB_1048550
IF (1:125)
AntibodyF(ab’)2 Anti-Mouse IgG (H+L) Alexa Fluor 647, (Goat polyclonal)Thermo FisherCat # A21237
RRID:AB_2535806
IF (1:1000)
AntibodyF(ab’)2 Anti-Rabbit IgG (H+L) Alexa Fluor 555 (Goat polyclonal)Thermo FisherCat # A21425
RRID:AB_2535846
IF (1:1000)
AntibodyAnti-Chicken IgY (H+L) Alexa Fluor 647 (Goat pAb)AbcamCat # Ab150175
RRID:AB_2732800
IF (1:1000)
AntibodyAnti-Chicken IgY (H+L) Alexa Fluor 488 (Goat polyclonal)AbcamCat # Ab150173
RRID:AB_2827653
IF (1:1000)
AntibodyF(ab’)2 Anti-Rabbit IgG (H+L) Alexa Fluor 488 (Goat polyclonal)Thermo FisherCat # A-11070
RRID:AB_2534114
IF (1:1000)
AntibodyAnti-Mouse IL-1R (Goat polyclonal)Leinco TechnologiesCat # I-736
RRID:AB_2830857
Blocking (2 ug/mL)
Sequence-based reagentmGAP FPMID:19515781PCR primersCATGGCCTTCCGTGTGTTCCTA
Sequence-based reagentmGAP RPMID:19515781PCR primersGCGGCACGTCAGATCCA
Sequence-based reagentICP27 FPMID:21285374PCR primersGCATCCTTCGTGTTTGTCATTCTG
Sequence-based reagentICP27 RPMID:21285374PCR primersGCATCTTCTCTCCGACCCCG
Sequence-based reagentICP8 FPMID:23322639PCR primersGGAGGTGCACCGCATACC
Sequence-based reagentICP8 RPMID:23322639PCR primersGGCTTAAATCCGGCATGAC
Sequence-based reagentICP4 FThis paperPCR primersTGCTGCTGCTGTCCACGC
Sequence-based reagentICP4 RThis paperPCR primersCGGTGTTGACCACGATGAGCC
Sequence-based reagentUL30 FPMID:22383875PCR primersCGCGCTTGGCGGGTATTAACAT
Sequence-based reagentUL30 RPMID:22383875PCR primersTGGGTGTCCGGCAGAATAAAGC
Sequence-based reagentUL48 FThis paperPCR primersTGCTCGCGAATGTGGTTTAG
Sequence-based reagentUL48 RThis paperPCR primersCTGTTCCAGCCCTTGATGTT
Sequence-based reagentgC FThis paperPCR primersCAGTTTGTCTGGTTCGAGGAC
Sequence-based reagentgC RThis paperPCR primersACGGTAGAGACTGTGGTGAA
Sequence-based reagentshRNA: DLK-1Broad Institute: Genetic Perturbation Platform/Millipore SigmaTRCN0000022573
Sequence-based reagentshRNA: DLK-2Broad Institute: Genetic Perturbation Platform/Millipore SigmaTRCN0000022572
Sequence-based reagentshRNA: non-targeting controlPMID:16873256
Commercial assay or kitQuick-RNA MiniprepZymo ResearchR1054
Commercial assay or kitSuperScript IV First-Strand Synthesis SystemThermoFisher18091050
Commercial assay or kitSYBR Green PCR Master MixThermoFisher4309155

Chemical compound, drugAcycloguanosineMillipore SigmaA466910 µM, 50 µM
Chemical compound, drugFUDRMillipore SigmaF-050320 µM
Chemical compound, drugUridineMillipore SigmaU-300320 µM
Chemical compound, drugSP600125Millipore SigmaS556720 µM
Chemical compound, drugGNE-3511Millipore Sigma5331684 µM
Chemical compound, drugGSK-J4Millipore SigmaSML07012 µM
Chemical compound, drugL-Glutamic AcidMillipore SigmaG56383.7 µg/mL
Chemical compound, drugForskolinTocris109960 µM
Chemical compound, drugLY 294002Tocris113020 µM
Chemical compound, drug666–15Tocris56612 µM
Chemical compound, drugSQ 22,536Tocris143550 µM
Chemical compound, drugKT 5720Tocris12883 µM
Chemical compound, drugTEATocris306810 mM
Chemical compound, drugCsClTocris47393 mM
Chemical compound, drugOG-L002Tocris624430 µM
Chemical compound, drugS2101Tocris571420 µM
Chemical compound, drugTetrodotoxinTocris10691 µM
Chemical compound, drugESI-09Tocris477310 µM
Chemical compound, drugZD 7288Cayman1522820 µM
Chemical compound, drug8-bromo-cyclic AMPCayman14431125 µM
Chemical compound, drugNGF 2.5SAlomone LabsN-10050 ng/mL
Chemical compound, drugPrimocinInvivogenant-pm-1100 µg/mL
Chemical compound, drugAphidicolinAG ScientificA-10263.3 µg/mL
Chemical compound, drugIL-1βShenendoah Bio.100–16730 ng/mL
Chemical compound, drugWAY-150138Pfizer, gift from Lynn Enquist and Jay Brown.NA10 µg/mL
Chemical compound, drugFura-2 AMThermo FisherF12215 µM
OtherHoescht StainThermo622492 µM

Data availability

All data generated or analysed during this study are included in the manuscript and supporting files. Source data files have been provided for all figures.

Additional files

Supplementary file 1

Table 1.

Cell Body Score for Neuronal Health and Degeneration IndexScoring system used to determine neuronal health based on morphology of the soma following treatment with compounds used in this study. Table 2. Axon Score for Neuronal Health and Degeneration IndexScoring system used to determine neuronal health based on morphology of the axons following treatment with compounds used in this study.

https://cdn.elifesciences.org/articles/58037/elife-58037-supp1-v3.docx
Transparent reporting form
https://cdn.elifesciences.org/articles/58037/elife-58037-transrepform-v3.pdf

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)