. 2022 Jul 18;13(1):4163.

doi: 10.1038/s41467-022-31591-y.

Orexin neurons inhibit sleep to promote arousal

Roberto De Luca¹, Stefano Nardone^#², Kevin P Grace^#^{1

3}, Anne Venner¹, Michela Cristofolini¹, Sathyajit S Bandaru¹, Lauren T Sohn¹, Dong Kong⁴, Takatoshi Mochizuki⁵, Bianca Viberti¹, Lin Zhu¹, Antonino Zito^{6

7}, Thomas E Scammell¹, Clifford B Saper¹, Bradford B Lowell², Patrick M Fuller^{8

9}, Elda Arrigoni¹⁰

Affiliations

¹ Department of Neurology, Division of Sleep Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, 02215, USA.
² Department of Medicine, Division of Endocrinology, Diabetes and Metabolism. Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, USA.
³ Department of Neurological Surgery, University of California Davis School of Medicine, Davis, CA, USA.
⁴ Department of Pediatrics, Division of Endocrinology, F.M. Kirby Neurobiology Center. Children's Hospital and Harvard Medical School, Boston, MA, USA.
⁵ Department of Biology, Graduate School of Science and Engineering. University of Toyama, Toyama, Japan.
⁶ Department of Molecular Biology, Massachusetts General Hospital, Boston, MA, 02114, USA.
⁷ Department of Genetics, Harvard Medical School, Boston, MA, 02114, USA.
⁸ Department of Neurology, Division of Sleep Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, 02215, USA. pmfuller@ucdavis.edu.
⁹ Department of Neurological Surgery, University of California Davis School of Medicine, Davis, CA, USA. pmfuller@ucdavis.edu.
¹⁰ Department of Neurology, Division of Sleep Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, 02215, USA. earrigon@bidmc.harvard.edu.

^# Contributed equally.

PMID: 35851580
PMCID: PMC9293990
DOI: 10.1038/s41467-022-31591-y

Orexin neurons inhibit sleep to promote arousal

Roberto De Luca et al. Nat Commun. 2022.

. 2022 Jul 18;13(1):4163.

doi: 10.1038/s41467-022-31591-y.

Authors

Affiliations

¹ Department of Neurology, Division of Sleep Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, 02215, USA.
² Department of Medicine, Division of Endocrinology, Diabetes and Metabolism. Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, USA.
³ Department of Neurological Surgery, University of California Davis School of Medicine, Davis, CA, USA.
⁴ Department of Pediatrics, Division of Endocrinology, F.M. Kirby Neurobiology Center. Children's Hospital and Harvard Medical School, Boston, MA, USA.
⁵ Department of Biology, Graduate School of Science and Engineering. University of Toyama, Toyama, Japan.
⁶ Department of Molecular Biology, Massachusetts General Hospital, Boston, MA, 02114, USA.
⁷ Department of Genetics, Harvard Medical School, Boston, MA, 02114, USA.
⁸ Department of Neurology, Division of Sleep Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, 02215, USA. pmfuller@ucdavis.edu.
⁹ Department of Neurological Surgery, University of California Davis School of Medicine, Davis, CA, USA. pmfuller@ucdavis.edu.
¹⁰ Department of Neurology, Division of Sleep Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, 02215, USA. earrigon@bidmc.harvard.edu.

^# Contributed equally.

PMID: 35851580
PMCID: PMC9293990
DOI: 10.1038/s41467-022-31591-y

Abstract

Humans and animals lacking orexin neurons exhibit daytime sleepiness, sleep attacks, and state instability. While the circuit basis by which orexin neurons contribute to consolidated wakefulness remains unclear, existing models posit that orexin neurons provide their wake-stabilizing influence by exerting excitatory tone on other brain arousal nodes. Here we show using in vivo optogenetics, in vitro optogenetic-based circuit mapping, and single-cell transcriptomics that orexin neurons also contribute to arousal maintenance through indirect inhibition of sleep-promoting neurons of the ventrolateral preoptic nucleus. Activation of this subcortical circuit rapidly drives wakefulness from sleep by differentially modulating the activity of ventrolateral preoptic neurons. We further identify and characterize a feedforward circuit through which orexin (and co-released glutamate) acts to indirectly target and inhibit sleep-promoting ventrolateral preoptic neurons to produce arousal. This revealed circuitry provides an alternate framework for understanding how orexin neurons contribute to the maintenance of consolidated wakefulness and stabilize behavioral state.

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interests.

Figures

**Fig. 1. Optogenetic stimulation of orexin terminals in the VLPO rapidly triggers arousals from both NREM and REM sleep.**
a Orexin innervation of the POA and VLPO (scale bars: 1 and 0.5 mm). b *AAV-DIO-ChR2-mCherry* was injected bilaterally into the Ox field of *Ox-IRES-Cre* mice and optical fibers bilaterally placed above the VLPO. c Orexin neurons transduced with ChR2- mCherry (red; left) and optical fiber (*) above the VLPO (right; scale bars: 500 µm). Optical fiber placements (red * ChR2-mCherry and grey * WT; bottom). d ChR2-eYFP(+) neurons double-labeled for Ox-A (in red; scale bars: 200; 100, and 20 µm). Laser-induced arousals from NREM (e) and REM (f) sleep (10 s-long periods before and after 10 s light pulses, top). EMG and EEG recordings and continuous EEG wavelet transforms (bottom). Trial-averaged responses to laser stimulation in NREM (g) and REM (h) sleep. EEG wavelet power for trials containing arousals (top; n = 5). Black contour lines: 5% significance level against sham trials (paired bootstrap confidence interval of the mean wavelet power difference). EMG mean responses (grey region: 95% bootstrap confidence interval; bottom). Mean differences in arousal probability, from NREM (i) and REM sleep (j) in ChR2-mCherry (n = 5) and WT (n = 3; top). On the bottom, the mean differences in arousal probability (stimulation vs. sham) within the ChR2-mCherry group (NREM: effect of genotype F_{(1, 24)} = 22.92, p = 0.003; interaction between genotype and stimulation frequency F_{(4, 24)} = 1.66, p = 0.192, two-way ANOVA one-repeated-measure. REM: interaction between genotype and stimulation frequency F_{(2, 12)} = 7.02, p = 0.010, two-way ANOVA one-repeated-measure). Mean difference in arousal probability from NREM (i, bottom; t = 2.98 and 3.73, p = 0.019 and 0.004 for 10 and 20 Hz stimulations) and REM sleep (j, bottom; t = 3.82 and 5.16, p = 0.002 and 0.0007 for 1 and 10 Hz stimulations; post-hoc Holm–Šidák t-tests, two-way ANOVA one-repeated-measure; n = 5; Source Data file). Error bars: SEM; ‡: significant for factor genotype (two-way ANOVA); #, significant interaction between the factors genotype and stimulation frequency (two-way ANOVA). Mean differences (stimulations vs sham within the ChR2-mcherry group) plotted as bootstrap sampling distributions (bottom). Dots: mean difference in arousal probability; vertical error bars: 95% confidence intervals; *p < 0.05 post-hoc Holm–Šidák t-test. All testing was two-tailed, and n refers to the number of independent animals. 3V 3^rd ventricle, Opt optic tract, f fornix, Ac anterior commissure, CPu corpus striatum, BF basal forebrain and LH lateral hypothalamus. Atlas levels are per Franklin and Paxinos, 2001.

**Fig. 2. Dual response of orexin in VLPO.**
a Whole-cell recordings of VLPO neurons and *post-hoc* labeling of a biocytin-filled neuron (scale bars: 500 and 20 µm). b Map of the recorded neurons in VLPO (n = 29; scale bar: 500 µm). c Most of VLPO neurons exhibit LTS (80 out of 116). d Ox-A (0.3-1 µM) excites 35% of the VLPO neurons: it increases action potential firing frequency (n = 21; one-way ANOVA, F_{(2, 60)} = 30.49; p < 0.0001; CTRL vs Ox-A and Ox-A vs. Wash, *adj-p* < 0.0001) and depolarizes the membrane potential (n = 26; one-way ANOVA, F_{(2, 75)} = 49.89; p < 0.0001; CTRL vs Ox-A and Ox-A vs Wash, *adj-p* < 0.0001). e Ox-A inhibits 60% of the VLPO neurons: it reduces action potential firing frequency (n = 37; one-way ANOVA, F_{(2, 108)} = 28.56; p < 0.0001; CTRL vs Ox-A and Ox-A vs Wash, *adj-p* < 0.0001) and hyperpolarizes membrane potential (n = 44; one-way ANOVA, F_{(2, 129)}=63.18; p < 0.0001; CTRL vs Ox-A and Ox-A vs Wash, *adj-p* < 0.0001). ****p < 0.0001, Bonferroni’s *post-hoc* test. Ac, anterior commissure; 3V, 3^rd ventricle; Opt, optical chiasm. Panel d and e: data are represented as means ± SEM, n refers to the number of recorded neurons and Source Data are provided as a Source Data file.

**Fig. 3. Pharmacological and molecular diversity of the VLPO GABAergic neurons.**
We recorded from TdTomato labeled VLPO Vgat(+) (VLPO^Vgat) neurons in *Vgat-IRES-Cre* mice injected into the VLPO with *AAV-DIO-TdTomato*. VLPO^Vgat neurons (a *top;* TdTomato-labelled; sample #a18; scale bar: 20 μm) that are inhibited by NA (50 μM) express galanin. Single-cell RT-PCR results from 4 VLPO neurons inhibited by NA (b). The VLPO^Vgat neurons (c *top*; TdTomato-labelled; sample #a30; scale bar: 20 μm) that are excited by NA do not express galanin. Single-cell RT-PCR results from 5 VLPO^Vgat neurons excited by NA (d). Gal (182 bp), *Gad1* (177 bp), *Gad2* (248 bp), and *Gapdh* (171 bp; housekeeping gene), M marker ladder, nc negative control. e scRT-sqPCR results confirming the presence of *galanin* mRNA in 7 VLPO^Vgat neurons inhibited by NA and/or Ox-A (NA/Ox-A(−)) and the absence of *galanin* mRNA in 8 VLPO^Vgat neurons excited by NA and/or Ox-A (NA/Ox-A(+)). Mann–Whitney unpaired one-sided t-test, *Gal* NA/Ox-A(−) vs *Gal* NA/Ox-A(+); p = 0.0002, ***p < 0.001; values normalized to the mean *Gapdh*. e data are represented as means ± SEM. Panels b, d, and e Source Data are provided as a Source Data file.

**Fig. 4. Orexin enhances the VLPO^GABA→VLPO^GABA/Gal circuit.**
a To study the VLPO^Vgat→VLPO^GABA/Gal input, we injected a mix of *AAV-fDIO-ChR2-eYFP* and *AAV-DIO-TdTomato* into the VLPO of *Vgat-Flp::Gal-IRES-Cre* mice (n = 10) and recorded from TdTomato-labeled VLPO^GABA/Gal neurons. To study the VLPO^GABA/Gal→VLPO^GABA/Gal input, we injected *AAV-DIO-ChR2-mCherry* into the VLPO of *Gal-IRES-Cre* mice (n = 2) and recorded from mCherry-labeled VLPO^GABA/Gal neurons. VLPO^GABA neurons labelled in green (white arrow) and VLPO^GABA/Gal neurons double-labeled in green and red (magenta arrow; scale bar: 20 µm). b Restricted transduction of *AAV-fDIO-ChR2-eYFP* in the VLPO of *Vgat-Flp::Gal-IRES-Cre* mice (YFP immunolabeling in green, scale bar: 500 μm). Injection sites from 12 mice used for in vitro CRACM recordings. c Raster plots of IPSCs in VLPO^GABA/Gal neurons with photostimulation of VLPO^Vgat→VLPO^GABA/Gal input (*left*) and VLPO^GABA/Gal→VLPO^GABA/Gal input (*right;* bin duration: 50 ms). d IPSC probability in response to photostimulation of the VLPO^Vgat→VLPO^GABA/Gal input (*black*; n = 8) and the VLPO^GABA/Gal→VLPO^GABA/Gal input (*grey*; n = 6; means ± SEM). Photostimulation of VLPO^Vgat neurons evokes GABA_A-mediated oIPSCs in VLPO^GABA/Gal neurons (e; BIC 20 µM). Mean oIPSC amplitude and latency (f; n = 8; in red: means ± SEM). g Opto-evoked IPSCs recorded in TTX (1 µM + 4-AP 25–50 µM, n = 43; in red: mean ± SEM). h Ox-A (1 µM) increases oIPSC amplitude (recordings in TTX + 4-AP). i Mean oIPSC amplitude in CTRL, Ox-A and Wash (n = 9, one-way ANOVA), F_{(2, 24)} = 19.42, p < 0.0001, CTRL vs Ox-A and Ox-A vs Wash adj-p-value = 0.0002; means ± SEM). j Peak-scaled non-stationary fluctuation analysis showing the current/variance relationship for oIPSCs (left; control: N = 84.2, i = 0.84 pA and Ox-A: N = 139.6, i = 0.82 pA). Ox-A increases the numbers of activated GABA_A channels (*center*; n = 8, CTRL vs Ox-A, p = 0.0091, paired one-sided t-test; **p < 0.01) without affecting the GABA_A unitary current (*right*; n = 8, CTRL vs Ox-A, p = 0.2052, paired one-sided t-test; means ± SEM). Recordings at reversal potential of the ChR2-mediated current (Vh = −5 to −15 mV). Blue-light pulses (10 ms; blue bars). ***p < 0.001 Bonferroni’s *post-hoc* test. In grey: 30 individual oIPSCs; in black: average oIPSC. Opt, optical chiasm; 3V, 3^rd ventricle; Atlas levels are per Franklin and Paxinos, 2001. For panels d–j: n refers to the number of recorded neurons. Panels c, d, f, g, and i, j: Source Data are provided as a Source Data file.

**Fig. 5. Differential expression profiles and orexin receptor expression in VLPO^GABA/Gal and VLPO^GABA neurons.**
Heat maps representing the expression levels of the top 30 differentially expressed genes ranked by *adj-p* (panel a; top 15 hypo- and top 15 hyper-expressed) and expression levels of GABAergic genes (*Slc32a1*, *Gad1*, *Gad2*), *Hcrtr1, Hcrtr2*, *Gal* and *Slc17a6* (panel b) in the VLPO^GABA (#0–3; 7–9; 12, 13, 15, 17, 19) and VLPO^GABA/Gal (#5, 10, 14) clusters. *Hcrtr2* positive neurons were largely restricted to VLPO^GABA cluster #1. c Dot plot representing the average expression of *Hcrtr2* gene (*x axis*) in VLPO^GABA and VLPO^GABA/Gal groups (*y axis*). Dot size: percentage of neurons that expresses a specific gene (*bottom right*). Color intensity: expression level (*top right*). d Heat map of *Gal* and *Hcrtr2* gene expression levels in VLPO^GABA cluster #1 (orange; 489 neurons), in all the other VLPO^GABA clusters (#0, 2, 3; 7–9; 12, 13, 15, 17, 19; green; down sampled to 515 neurons) and merged VLPO^GABA/Gal clusters (#5, 10, 14; cyan; 515 neurons). Violin plots for *Hcrtr2* (*bottom left*) and *Gal* (*bottom right*) differential gene expression: VLPO^GABA cluster #1 (orange), all the other VLPO^GABA clusters (#0, 2, 3; 7–9; 12, 13, 15, 17, 19; green) and VLPO^GABA/Gal clusters (#5, 10, 14; cyan). *Hcrtr2* is expressed in VLPO^GABA #1 (VLPO^GABA #1 vs VLPO^GABA/Gal adj-p value = 4.98E−32; VLPO^GABA #1 vs. VLPO^GABA (#0, 2, 3; 7–9; 12, 13, 15, 17, 19) adj-p value = 1.63E−114). *Gal* is expressed in VLPO^GABA/Gal neurons (VLPO^GABA #1 vs VLPO^GABA/Gal adj-p value = 1.26E−144; VLPO^GABA (#0, 2, 3; 7–9; 12, 13, 15, 17, 19) vs VLPO^GABA/Gal adj-p value = 0). Heat map and Dot plot expression values are represented as z-scores. Expression levels is color- coded in the legend. In the heat maps, *x-axis*: individual clusters color-coded bar; *y-axis*: genes. As color dots: Cluster IDs (*right*). Test used: *Wilcoxon Rank Sum two-sided* Bonferroni-corrected Test. e Expression in VLPO of *Gal* (red), *Slc32a1* (green) and *Hcrtr2* (pseudo colored in blue) mRNAs. White arrows: *Slc32a1*(+), *Hcrtr2*(+) and *Gal*(−) neurons; magenta arrows: *Slc32a1*(+), *Hcrtr2*(−) and *Gal*(−) neurons. Scale bars: 500 and 20 µm (*bottom right*). 3V, 3^rd ventricle. Source Data are provided as a Source Data file.

**Fig. 6. Orexin input excites VLPO^GABA neurons by glutamate release (Ox → VLPO^GABA).**
a *AAV-DIO-ChR2-eYFP* injected into the Ox field of *Ox-IRES-Cre* mice and recordings from VLPO neurons. b Injection sites of 15 *Ox-IRES-Cre* mice used for in vitro CRACM (n = 10 mice) and in vivo optogenetics (n = 5 mice). c Photostimulation of Ox → VLPO input evokes AMPA-mediated oEPSCs in VLPO neurons (DNQX 200 µM; in grey: 30 individual oEPSCs; in black: average oEPSC). d Raster plot of EPSCs in a VLPO neuron before, during, and after photostimulation of the Ox → VLPO input (50 ms bins). e Average EPSC probability of VLPO neurons that responded (n = 13; black) and did not respond (n = 48; grey; means ± SEM) to the photostimulation of Ox input. f No changes in holding currents in response to photostimulation trains (60 s at 10 and 20 Hz; 10 ms light pulses) in neurons activated by the Ox → VLPO input with glutamate-mediated oEPSCs. The VLPO^GABA neurons are activated by the Ox → VLPO input. The VLPO neurons that responded to photostimulation of the Ox → VLPO input (n = 7) are excited by NA (g) and/or do not express *galanin* mRNA (h; scRT-sqPCR). Specifically 4 neurons were tested for both the responses to NA and for the presence of *Gal* mRNA, 2 only for NA, and 1 only for *Gal* mRNA. Mean effects of NA on the action potential frequency (g *left*; paired one-sided t-test, CTRL vs NA, p = 0.0033; n = 4) and membrane potential (*g, right;* paired one-sided t-test, CTRL vs NA, p = 0.0017; n = 6). Results from scRT-sqPCR for *Gapdh* and *galanin* (*Gal*) from 5 VLPO neurons Ox → VLPO input connected (conn) are compared to 13 controls (ctrl) VLPO^GABA/Gal neurons (h; ctrl: sampled from recorded VLPO^GABA/Gal TdTomato-labeled neurons from *Gal-IRES-Cre* mice injected with *AAV-DIO-TdTomato*). Mann-Whitney unpaired one-sided t-test, *Gal* (conn) vs *Gal* (ctrl); p = 0.0001. Values are normalized to the mean *Gapdh*. Means ± SEM. i oEPSC amplitude (*left*) and latency (*right*) in 15 VLPO neurons (blue dots: identified VLPO^GABA neurons; in red: means ± SEM). Recordings at V_h= −70 mV; 10 ms blue-light pulses (blue bars). ***p < 0.001 and **p < 0.01. 3V 3^rd ventricle, Opt optical chiasm, f fornix, mt mammillothalamic tract, LH lateral hypothalamus. Atlas levels are per Franklin and Paxinos, 2001. For panels e and g: n refers to the number of recorded neurons. Panel d–e and g–i: Source Data file.

**Fig. 7. Orexin neurons promote wakefulness by inhibiting sleep-promoting VLPO^GABA/Gal neurons via feed-forward inhibition.**
Orexin neurons promote arousal through their projections to the VLPO. Orexin input to the VLPO directly activates VLPO GABAergic neurons that do not express galanin (VLPO^GABA) via co-release of Ox and glutamate. In turn, VLPO^GABA neurons inhibit VLPO^GABA/Gal sleep-promoting neurons to produce wakefulness.

See this image and copyright information in PMC

Cited by

A hypothalamic circuit for circadian regulation of corticosterone secretion.
Ramirez-Plascencia OD, De Luca R, Machado NLS, Eghlidi D, Khanday MA, Bandaru SS, Raffin F, Vujovic N, Arrigoni E, Saper CB. Ramirez-Plascencia OD, et al. Res Sq [Preprint]. 2024 Jul 12:rs.3.rs-4718850. doi: 10.21203/rs.3.rs-4718850/v1. Res Sq. 2024. PMID: 39041039 Free PMC article. Preprint.
The positive impact of smoking on poor sleep quality is moderated by IGF1 levels in cerebrospinal fluid: a case-control study among Chinese adults.
Shan L, Wu Y, Lao J, Ma M, Luo X, Zheng K, Hu W, Kang Y, Wang F, Liu Y, Xu Y, Jin X. Shan L, et al. Front Psychiatry. 2024 May 10;15:1392732. doi: 10.3389/fpsyt.2024.1392732. eCollection 2024. Front Psychiatry. 2024. PMID: 38800060 Free PMC article.
Insomnia in Forensic Detainees: Is Salience Network the Common Pathway for Sleep, Neuropsychiatric, and Neurodegenerative Disorders?
Sfera A, Thomas KA, Ogunjale IA, Jafri N, Bota PG. Sfera A, et al. J Clin Med. 2024 Mar 15;13(6):1691. doi: 10.3390/jcm13061691. J Clin Med. 2024. PMID: 38541916 Free PMC article. Review.
Orexins and primary headaches: an overview of the neurobiology and clinical impact.
Stanyer EC, Hoffmann J, Holland PR. Stanyer EC, et al. Expert Rev Neurother. 2024 May;24(5):487-496. doi: 10.1080/14737175.2024.2328728. Epub 2024 Mar 22. Expert Rev Neurother. 2024. PMID: 38517280 Free PMC article. Review.
Whole Brain Mapping of Orexin Receptor mRNA Expression Visualized by Branched In Situ Hybridization Chain Reaction.
Tsuneoka Y, Funato H. Tsuneoka Y, et al. eNeuro. 2024 Feb 21;11(2):ENEURO.0474-23.2024. doi: 10.1523/ENEURO.0474-23.2024. Print 2024 Feb. eNeuro. 2024. PMID: 38199807 Free PMC article.

See all "Cited by" articles

References

1. Scammell TE. Narcolepsy. N. Engl. J. Med. 2015;373:2654–2662. doi: 10.1056/NEJMra1500587. - DOI - PubMed
1. Chemelli RM, et al. Narcolepsy in orexin knockout mice: molecular genetics of sleep regulation. Cell. 1999;98:437–451. doi: 10.1016/S0092-8674(00)81973-X. - DOI - PubMed
1. Mochizuki T, et al. Behavioral state instability in orexin knock-out mice. J. Neurosci. 2004;24:6291–6300. doi: 10.1523/JNEUROSCI.0586-04.2004. - DOI - PMC - PubMed
1. Branch AF, et al. Progressive loss of the orexin neurons reveals dual effects on wakefulness. Sleep. 2016;39:369–377. doi: 10.5665/sleep.5446. - DOI - PMC - PubMed
1. Carter ME, et al. Mechanism for Hypocretin-mediated sleep-to-wake transitions. Proc. Natl Acad. Sci. USA. 2012;109:E2635–E2644. - PMC - PubMed

Publication types

Actions

MeSH terms

Actions
Actions
Actions
Actions
Actions
Actions
Actions

Save citation to file

Email citation

Add to Collections

Add to My Bibliography

Your saved search

Create a file for external citation management software

Your RSS Feed

Orexin neurons inhibit sleep to promote arousal

Affiliations

Orexin neurons inhibit sleep to promote arousal

Authors

Affiliations

Abstract

Conflict of interest statement

Figures

Similar articles

Cited by

References

Publication types

MeSH terms

Substances

Grants and funding

LinkOut - more resources

Full Text Sources

Molecular Biology Databases

Abstract

Conflict of interest statement

Figures

Similar articles

Cited by

References

Publication types

MeSH terms

Substances

Related information

Grants and funding

LinkOut - more resources

Full Text Sources

Molecular Biology Databases