Molecular changes associated with migratory departure from wintering areas in obligate songbird migrants Free

Twice a year, billions of birds undertake to-and-fro long-distance nocturnal journeys between breeding grounds in the north and wintering grounds in the south. Day length regulates these predictable seasonal migratory movements and the subsequent reproductive state in avian migrants (Ramenofsky and Wingfield, 2007; Rani et al., 2017). In particular, increasing photoperiod induces the development of spring migratory and subsequent reproductive phenotypes in avian migrants. This is faithfully reproduced under laboratory conditions. For instance, when exposed to appropriate photoperiods, captive birds exhibit a migratory phenotype (body fattening and weight gain, and intense nocturnal activity – the Zugunruhe used as a proxy for the propensity of birds to migrate; Berthold and Querner, 1988) and gonadal maturation, used as a proxy for reproduction (Lofts, 1975). Several lines of evidence suggest the mediobasal hypothalamus (MBH) as the key brain area for the photoperiodic induction. The MBH contains receptors and signal transduction molecules for photoperiod perception and phototransduction, a time-keeping mechanism to assess the photoperiod change, and effector (output) pathways that convert the photoperiodic message into a biological response (Surbhi and Kumar, 2014; Cassone and Yoshimura, 2015; Kumar and Mishra, 2018). It is therefore the site for controlling the photoperiodic induction of multiple processes, including the photostimulated development of spring migratory and reproductive phenotypes in migratory songbirds (Sharp, 2005; Yoshimura, 2006; Nakao et al., 2008; Rastogi et al., 2011; 2013; Stevenson and Kumar, 2017).

Both field and controlled laboratory studies demonstrate that migratory preparedness is linked to fat accumulation and changes in the activity of enzymes associated with metabolism (Odum, 1960; Newton, 2008; Weber, 2009). In fact, the association of accumulated fat stores and enzyme activity with departure decision from stopover sites has been reported in several songbird migrants (Bairlein, 1985; Ramenofsky, 1990; Fusani et al., 2013; Lupi et al., 2016). In particular, changes in fatty acid (FA) synthesis (lipogenesis) in the liver and its transport to skeletal (flight) muscles serve as reliable indices for readiness for the departure and subsequent prolonged nocturnal flights in migratory songbirds (McFarlan et al., 2009; Banerjee and Chaturvedi, 2016).

A key question is whether the information about peripheral fat accumulation is integrated into the photoperiodic control mechanism(s) in the MBH. To address this, we compared gene and protein expression in migratory redheaded buntings (Emberiza bruniceps) that were long-day (LD) photostimulated for 4 days (early stimulated: LD-eS) or ∼3 weeks (stimulated: LD-S), and with that in birds maintained under non-stimulatory short-day (SD) conditions. Buntings are a Palearctic-Indian obligate migrant songbird with predictable yearly migrations between breeding and wintering areas (Ali and Ripley, 1999) and exhibit a spring migratory phenotype and testicular maturation under LD (Sharma et al., 2018a).

Here, we first examined the expression of genes coding for enzymes involved in FA synthesis (acetyl CoA carboxylase, acc; FA synthase, fasn; diacylglycerol O-acyltransferase 2, dgat2) in the liver, and those coding for proteins/enzymes involved in FA transport in both liver (apolipoprotein 1, apoa1) and flight muscles (FA translocase, cd36/fat; FA binding protein 3, fabp3; carnitine palmitoyltransferase1, cpt1). This was done to show changes in the transcription of genes associated with FA synthesis and transport with the photoperiod-induced transition of buntings into the migratory state, based on a previous study in which these genes were found to be upregulated in expression in both liver and muscle during the photostimulated state (Sharma and Kumar, 2019).

Then, we measured the expression of genes associated with a few key physiological pathways in the MBH that might show alteration with the photoperiodic induction of the migratory state (Boss et al., 2016; Sharma et al., 2018a,b). We chose candidate genes associated with neural processes, cellular stress, metabolic status and migratory behaviour in view of the role of photoperiod in controlling multiple processes at the brain level (Sharp, 2005; Nakao et al., 2008; Rastogi et al., 2013; Cassone and Yoshimura, 2015; Stevenson and Kumar, 2017). For example, the activation of the Ca²⁺-signalling and transport pathway (camk1 and camk2a, encoding calcium/calmodulin-dependent protein kinase 1 and II alpha; and atp2a2, encoding ATPase sarcoplasmic Ca²⁺ transporting 2 protein) has been reported in photostimulated migratory willow warblers (Phylloscopus trochilus; Boss et al., 2016) and blackheaded buntings (Emberiza melanocephala; Sharma et al., 2018a,b). Although not shown in birds, the alteration in Ca²⁺-signalling has been reported to correlate with key neural processes in some animals; for example, the neurogenesis and maintenance of synaptic connections (Berridge et al., 2000; Bouron, 2020). Additionally, there is evidence for neurogenesis and neuronal recruitment in different brain regions during the migratory season in avian migrants, such as white-crowned sparrows (Zonotrichia leucophrys gambelii; LaDage et al., 2011) and migratory reed warblers (Acrocephalus scirpaceus; Barkan et al., 2014). Increased neuronal activity and neurogenesis may indicate an enhanced capacity for the integration of sensory inputs and the ability to use acquired information for migration and navigation (Åkesson and Hedenström, 2007; Muheim et al., 2006; Warren et al., 2010). It is, therefore, likely that genes associated with Ca²⁺ signalling and transport serve as important molecular correlates of enhanced hypothalamic activity during the photostimulated state.

Further, we reasoned that changes in metabolic status and oxidative stress as a consequence of heightened activity during the photostimulated state would be reflected at the brain level. To test this, we measured hspa8 (a member of heat shock protein family), sod1 (encoding superoxide dismutase) and nos2 (encoding inducible nitric oxide synthase, iNOS) expression to show the effects on the cellular oxidative stress (Jones et al., 2008; Raja-Aho et al., 2012; Sharma et al., 2018b; Valek et al., 2019). Concurrently, we measured apoa1 (encoding apolipoprotein A1) and sirt1 (encoding sirtuin 1) expression, levels of which correspond to migratory fattening and serve as indices of the metabolic status (energy reserve) in songbird migrants (Trivedi et al., 2014; Frias-Soler et al., 2020). In addition, we measured adcyap1 (encoding pituitary adenylate cyclase activating polypeptide, PACAP) and vps13a (encoding vacuolar protein sorting-associated protein 13A) expression. PACAP is light sensitive, and is involved in circadian clock phase shifts (Harrington et al., 1999; Nowak and Zawilska, 2003); hence, the change in adcyap1 gene expression, if any, could be associated with circadian clock-controlled physiological and behavioural shifts that occur with photoperiodic induction of the migratory phenotype in migratory songbirds including buntings (Bartell and Gwinner, 2005; Rani et al., 2006; Singh et al., 2015). There is also a reported association of the allele size of the adcyap1 gene with Zugunruhe as well as with migratory distance in migratory blackcaps (Sylvia atricapilla; Mueller et al., 2011). Likewise, vps13a gene expression has been linked with migration directionality in Vermivora warblers (Toews et al., 2019).

Finally, we sought to examine changes at the proteome level rather than at the more commonly used transcriptome level, as it can reveal the complex hypothalamic processes involved in migratory preparedness and subsequent gonadal maturation. For this, we used a MBH-specific proteome-wide assay, and identified key proteins that were differentially expressed with the photostimulated transition from the LD-eS to the LD-S state. We used a mass spectrometry-based quantitative proteome approach that facilitates protein profiling and assessment of protein interactions involved in biological processes (Yugandhan et al., 2019).

All procedures were approved and carried out in accordance with guidelines of the Institutional Animal Ethics Committee (IAEC) of Department of Zoology, University of Delhi, India (Institutional Ethical Approval number: DU/ZOOL/IAEC-R/2015/04). This study was done on adult male redheaded buntings (Emberiza bruniceps Brandt 1841), which were procured from an overwintering flock in late February, and acclimated to captive conditions for ca. 2 weeks in an outdoor aviary (3×2.5×2.5 m) under the natural photoperiod (∼11.3 h, sunrise to sunset) and temperature (21–25°C) conditions. The birds were then brought indoors and maintained under SD conditions (8 h light and 16 h dark), and were singly housed in activity cages (42×30×52 cm; one bird per cage) placed in independent photoperiodic boxes (70×50×70 cm; one cage per photoperiodic box) for a period of 4 weeks, until the experiment began. Each cage was equipped with two perches and mounted with a passive infrared (PIR) motion sensor (DSC, LC100 PI digital PIR detector, Concord, ON, Canada) to monitor the activity of the bird inside the cage. Photoperiodic boxes were lit by compact fluorescent lamps (CFL, Phillips) providing a light period of 300 lx and a dark period of <1 lx. HOBO temperature loggers monitored 24 h temperature of each box, and this was maintained over 24 h at 24.7±0.07°C (mean±s.e.m.). All birds had ad libitum access to food and water, which were replenished only during the light phase. Under SD conditions, buntings maintain small testes and retain responsiveness to the stimulatory effects of long photoperiod; we called them SD-photosensitive birds.

The experiment used a total of 21 adult male redheaded buntings maintained under SD conditions (SD photosensitive birds: body mass 21–24 g, volume of unstimulated testes 0.33–0.52 mm³). Fig. S1 describes the experimental protocol. Briefly, 16 birds were exposed to a stimulatory 12.5 h photoperiod (12.5 h light:11.5 h dark) for 4 days (n=8; LD-eS) or until each bird had shown 4 successive nights of Zugunruhe (∼3 weeks; n=8; LD-S), which was verified by the videographs recorded using the night vision camera. A 12.5 h light period per day approximated the natural daylight (sunrise to sunset) available in the wild at the time of spring migration in early April. We called birds exposed to LD conditions for 4 days ‘early stimulated’ based on our earlier bunting studies in which exposure to LD conditions for 1 or 2 days triggered photoperiodic induction at the molecular level in the hypothalamus (Majumdar et al., 2015; Mishra et al., 2018). The remaining 5 birds were retained for the next 3 weeks under SD conditions, and served as controls (SD state). The temperature over 24 h was maintained at 24.7±0.07°C.

We monitored the 24 h activity–rest pattern, body mass and testis size as faithful markers of behavioural and physiological changes associated with migration and gonadal maturation (Berthold and Querner, 1988; Trivedi et al., 2014; Sharma et al., 2018a).

The activity–rest pattern over 24 h reflects circadian clock effects, which mediate the phase transition from the non-migratory to migratory state in songbirds (Bartell and Gwinner, 2005; Rani et al., 2006). Photostimulated Zugunruhe (nocturnal migratory restlessness) also reflects faithfully the temporal pattern of actual migration in the wild (Gwinner, 1972). The activity of each bird was recorded by a PIR motion sensor mounted in front of the activity cage. The PIR sensor detected and recorded general movements including perch hopping activity of the bird in its cage, and transferred and stored these in 5 min bins into designated channels of the computerized data acquisition system. The collection, analysis and graphic presentation of the 24 h activity–rest pattern were done by using The Chronobiology Kit software (Stanford Software Systems). To compare the overall distribution of activity over 24 h between groups, we first averaged hourly activity for 4 consecutive days for every bird in all three groups, and then normalized the average value relative to the hour with the maximal value, which was given a score of 100 (Sharma and Kumar, 2019). The normalized individual scores were plotted in a circular polar plot using a MS-Excel based add-in application, showing the magnitude in concentric circles and the phase in radial lines for each bird.

Each bird was weighed on a top pan balance to an accuracy of 0.1 g on day 1 of the experiment (initial value) and 1 day before the bird was used for harvesting the tissue samples (final value). From these values, we calculated change in body mass (Δbody mass=final value−initial value). Further, using a subjective criterion on a 0–5 scale (see below), we also recorded subcutaneous fat accumulation in each bird 1 day before tissue harvesting in order to assess body fattening as a result of photoperiodic exposure. This scheme of fat scoring, which is routinely done in our laboratory for buntings (Malik et al., 2004; Budki et al., 2009), is based on previous studies in songbird migrants (Helms and Drury, 1960; Wingfield and Farner, 1978; Biebach et al., 1986). The scores run as follows: 0=no visible subcutaneous fat deposits, 1=light fat deposits overlying musculature (musculature remains clearly visible), 2=heavier fat deposits overlying musculature (vasculature still visible), 3=fat deposits overlying the entire region, 4=areas filled with whitish, bulging fat deposits, and 5=copious fat deposits all over three regions.

The size of the left testis was measured to an accuracy of 0.5 mm at the beginning of the experiment by laparotomy and at the end of the experiment, prior to tissue samples being taken. Laparotomy is a routine procedure carried out in our laboratory. Here, the testes were located through a small incision on the left flank between the last two ribs in birds under general anaesthesia (a mixture of ketamine/xylazine solution: 67.5 mg ketamine+7.5 mg xylazine per kg body mass). The dimensions of the left testis were measured to an accuracy of 0.5 mm, and from these, testis volume was calculated using the formula 4/3πab² (where a is half the length and b is half the width; Budki et al., 2009). The incision was sutured, and an antibacterial skin cream (Soframycin, Aventis Pharma Ltd) was applied. The bird was returned to its cage on to a warm pad for ∼30 min, and the resumption of normal perch hopping in ∼1 h indicated full recovery of the bird from anaesthesia effects.

At the end of the experiment, half an hour before lights off (SD: hour 7.5 h; LD-eS and LD-S: hour 12 h; where hour 0=lights on), birds were killed by decapitation (an unanticipated, quick procedure lasting ∼10 s), which was used to preclude possible anaesthesia effects on mRNA expression (Staib-Lasarzik et al., 2014). The liver and pectoral muscle, which is a dominant avian flight muscle, and MBH were harvested. To obtain the MBH, the brain was placed with its ventral side up, and two coronal incisions were made to isolate the diencephalon; the MBH was then removed in the shape of an inverted V by placing longitudinal incisions on either side of the third ventricle (Olkowicz et al., 2016; Mishra et al., 2018). Tissues were snap-frozen on dry ice and stored at −80°C until use.

Using RT-qPCR, we measured the expression of genes associated with FA synthesis and transport in the liver (acc, fasn, dgat2 and apoa1), and with FA transport in the muscle (cd36, fabp3 and cpt1). In the MBH, we measured the expression of genes involved in calcium signalling and transport (camk1, camk2a and atp2a2), metabolism (apoa1 and sirt1), cellular stress/energy demand (hspa8, sod1 and nos2), and migratory behaviour (adcyap1 and vps13a). These genes were chosen based on microarray, transcriptome and gene expression studies in migratory white-crowned sparrows (Z. leucophrys gambelii; Jones et al., 2008), willow warblers (P. trochilus; Boss et al., 2016), blackheaded buntings (E. melanocephala; Sharma et al., 2018b) and redheaded buntings (E. bruniceps; Sur et al., 2020).

For each gene expression assay (n=5 samples per tissue), total RNA was extracted from each sample using Trizol solution (Ambion, 15596-018), as per the manufacturer’s protocol. The concentration and purity of RNA was checked by NanoDrop™ 2000C (ThermoFisher Scientific, Wilmington, DE, USA); an A₂₆₀/A₂₈₀ ratio of 1.9–2.0 was considered as pure for RNA. To assess the integrity of the RNA, we proceeded with cDNA preparation and then used PCR to amplify the cDNA from each sample using primers (beta-actin) that generate a greater amplicon size (∼750 bp). Only those cDNA samples that could amplify beta-actin (∼750 bp) were used for the measurement of gene expression by qPCR (Gong et al., 2006). For reverse transcription, 1 μg pure RNA was treated with RQ1 RNase-free DNase (Promega, M610A) for 30 min at 37°C. The reaction was stopped by adding 1 µl of the stop solution (50 mmol l⁻¹ EDTA) and incubating samples at 65°C for 10 min. The samples were then reverse transcribed to synthesize cDNA using Revert Aid First Strand cDNA synthesis kit (ThermoFisher Scientific, K1622) by incubation of DNase-treated RNA first with random hexamers for 5 min at 25°C, and then with 10 mmol l⁻¹ dNTP mix, reverse transcriptase and RNase inhibitor enzyme for 60 min at 42°C. The product of this reaction (post-assessment of RNA integrity) was used for qPCR. Gene-specific primers were designed from gene sequences available with us (acc, fasn, dgat2, cd36, fabp3, cpt1, atp2a2, sirt1, hspa8, sod1, nos2 and adcyap1; Trivedi et al., 2014; Sharma et al., 2018a,b; Sharma and Kumar, 2019; Sur et al., 2019, 2020) or from those we cloned for this study (apoa1, vps13a, camk1 and camk2a). We cloned a gene sequence by using degenerate primer sequence based on a conserved region of the gene, and the amplified products (>500 bp) of bunting hypothalamus cDNA templates were commercially sequenced (Eurofins, Bangalore, India) and subjected to a NCBI database nucleotide BLAST search to ascertain gene identity. Identified gene sequences were submitted to NCBI GenBank (for the accession numbers, see Table S1). qPCR was carried out in duplicate with SYBR green chemistry and a reaction volume of 6 μl (3 μl 1× SYBR Green, 1 μl each primer and 1 μl cDNA). We ran 2-step qPCR for 40 cycles, with each cycle lasting for 75 s (denaturation at 95°C for 15 s and annealing at 60°C for 60 s). Each PCR plate included beta-actin as a control (reference) gene (Sharma et al., 2018a). Consistent with guidelines for the use of qPCR (Bustin et al., 2009), relative mRNA expression was determined by the ΔΔCt method (Livak and Schmittgen, 2001), as in our previous studies on redheaded buntings (Sharma et al., 2018a). The cycle threshold (Ct) determined by fluorescence surpassing the background noise was used to calculate ΔCt (Ct[gene of interest]−Ct[reference gene]). The Ct values normalized against the Ct value of the pooled cDNA from all samples gave ΔΔCt, and the negative value of ΔΔCt powered to 2 (2^−ΔΔCt) gave the relative mRNA expression level.

Mass spectrometry (LC-MS/MS) was carried out for global profiling and differential protein expression in the MBH using three samples each from LD-eS and LD-S. Each MBH sample was homogenized in protein extraction buffer (20 mmol l⁻¹ Tris buffer pH 8, 8 mol l⁻¹ urea, 4% w/v Chaps, 0.5% v/v Triton X-100, 10 mmol l⁻¹ DTT and protease inhibitor cocktail). The lysate was centrifuged at 15,142 g at 4°C for 1 h, the supernatant collected in a fresh tube and protein concentration determined by using BCA Protein Assay Kit (ThermoFisher Scientific, 23225). An equal amount (150 µg) of total protein from each sample was reduced and alkylated using 10 mmol l⁻¹ DTT for 1 h followed by treatment with 40 mmol l⁻¹ iodoacetamide for 1 h at room temperature. The samples were cleaned up by acetone precipitation and the precipitated proteins were centrifuged at 15,871 g at 4°C for 1 h. The pellet was washed with ice-cold acetone and then re-suspended in 1 ml of 100 mmol l⁻¹ tetraethylammonium tetrahydroborate (TEAB). Protein digestion was performed by incubating the samples overnight at 37°C with trypsin (enzyme: total protein, 1:50; Promega Corporation). The peptides thus obtained were labelled with TMTsixplex label reagent (ThermoFisher Scientific, 90061) as per the manufacturer's protocol. Briefly, 0.8 mg TMT label (brought to room temperature) was added to 41 µl of anhydrous acetonitrile, and the reagent was allowed to dissolve for 5 min with occasional vortexing. To this, a total of 100 µl of protein sample was added and incubated for 1 h at room temperature. For reaction quenching, 8 µl of 5% hydroxylamine was added to each tube and incubated for 15 min at the room temperature. TMT-labelled samples were then mixed in equal amounts in a new centrifuge tube, vacuum dried and then used for LC-MS/MS analysis. A brief outline of the procedure for protein purification is given in Fig. S1.

The prepared samples (digested and labelled peptides) were analysed using an Orbitrap Velos Pro mass spectrometer coupled with a nano-LC 1000 (ThermoFisher Scientific). For this, the peptide mixtures were loaded onto a reverse-phase C-18 pre-column (Acclaim PepMap, 75 μm×2 cm, particle size 3 μm, pore size 100 Å, ThermoFisher Scientific), in line with an analytical column (Acclaim PepMap, 50 μm×15 cm, particle size 2 μm, pore size 100 Å). The peptides were separated using a gradient of 5% to 50% of solvent B (0.1% formic acid in 95/5 acetonitrile/water). The eluted peptides were injected into the MS and spectra were acquired in the Orbitrap at a resolution of 60,000. A minimum of 1000 counts was needed to trigger the MS/MS using HCD (higher-energy collisional dissociation) and the daughter ions were recorded in the Orbitrap at a resolution of 7500. The charge-state screening of the precursor and monoisotopic precursor selection was enabled, and an unassigned charge state and singly charged ions were rejected. The acquired spectra were analysed using the SEQUEST algorithm in the Proteome Discoverer (PD, v.1.4) software, with a precursor tolerance of 20 ppm and tolerance of 0.1 Da for MS/MS against the Ficedula albicollis database (Fic_Alb1.5) and two missed cleavages were allowed. The resultant identified peptides were validated using Percolator at 5% false discovery rate (FDR; q-value<0.05), which uses PEP (posterior error probability) and q-value for the validation. The proteins with fold-change (LD-S/LD-eS) of <0.5 (down-regulated) or >1.5 (up-regulated) were considered to be differentially expressed. We also performed functional analysis, and annotated proteins and pathways based on GO enrichment using RefSeq protein ID, protein family identification (PFAM) and KEGG using the DAVID database (Dennis et al., 2003). A pathway with P<0.05 was considered as significantly enriched.

To decipher a functional linkage of differentially expressed proteins, we performed STRING network analysis (string-db.org, v.11.0; a database of known and predicted protein–protein interactions), with data in the background from migratory F. albicollis. The network was drawn based on the confidence values, i.e. the strength of the data support was based on text mining, experiments, databases, co-expression, neighbourhood and co-occurrence. Based on the nature and quality of the support evidence, each interaction was given a score between 0 (lowest) and 1 (highest). The minimum interaction score was set to a medium confidence value of 0.400. The network STRING diagram is composed of nodes (which represent proteins) and edges (which represent protein–protein interactions). Whereas filled nodes represent known or predicted 3D structure, the empty nodes represent proteins with unknown 3D structure.

We used GraphPad Prism v.6.0 (GraphPad Software Inc., San Diego, CA, USA) for statistical analysis and data plots. One-way ANOVA was used to analyse significant differences, except for fat score data, for which we used the Kruskal–Wallis test. If these tests revealed a significant difference, we used the Newman–Keuls post hoc test or Dunn's post hoc test (for fat score only) for the group comparison. Pearson's correlation coefficient was calculated to show the relationship of genes involved in fatty acid synthesis with body mass gain, and of genes involved in the transport of fatty acids to flight muscle with Zugunruhe. The proteome data were analysed by the SEQUEST algorithm in PD v.1.4 software. For statistical significance, alpha was set at 0.05.

In contrast to diurnal activity patterns under SD and LD-eS, buntings under LD-S showed a transition to predominantly night-time activity (Fig. 1A; Fig. S1). This was nocturnal Zugunruhe, as confirmed by videographs (data not shown). We found a significant difference between photoinduced states (F_2,18=7.936, P=0.0034, one-way ANOVA), with night-time activity higher in the LD-S than in the SD and LD-eS states (P<0.05, Newman–Keuls post hoc test; Fig. 1B). Consistent with this, we also found significant differences between states in fat score (KW statistics=13.62, P=0.0001, Kruskal–Wallis test; Fig. 1C), Δbody mass (F_2,18=7.536, P=0.0042, one-way ANOVA; Fig. 1C) and testis size (F_2,18=11.67, P=0.0006, one-way ANOVA; Fig. 1D). Buntings were fatter, gained body mass and had larger testes in the LD-S than in the SD and LD-eS states (P<0.05, Newman–Keuls or Dunn's post hoc test).

Both hepatic and muscular gene expression supported enhanced FA synthesis and transport in the LD-S state. There was a significant difference in mRNA levels of genes associated with FA synthesis (acc: F_2,12=7.76, P=0.007; dgat2: F_2,12=15.47, P=0.0005) and transport (apoa1: F_2,12=9.22, P=0.004, one-way ANOVA) in the liver, with higher mRNA levels in the LD-S than in the SD and LD-eS state (not for acc; P<0.05, Newman–Keuls post hoc test; Fig. 1E–H). However, we did not find a change in fasn expression (F_2,12=0.331, P=0.725). There was also a positive correlation of expression of genes involved in FA synthesis (both acc: r=0.557, r²=0.311, P=0.031; and dgat2: r=0.587, r²=0.345, P=0.0213) mRNA levels with weight gain (Δbody mass; Fig. 1I).

Likewise, there was a significant difference in gene expression associated with FA transport (cd36: F_2,12=56.58, P<0.0001; fabp3: F_2,12=59.03, P<0.0001; cpt1: F_2,12=54.33, P<0.0001; one-way ANOVA) in the muscle, with significantly higher mRNA levels in the LD-S than in the SD and LD-eS state (P<0.05, Newman–Keuls post hoc test; Fig. 1J–L). The mRNA levels of cd36, in particular, were also higher in the LD-eS than in the SD state (Fig. 1J). The overall mRNA levels of cd36, fabp3 and cpt1 were positively correlated with night activity (Zugunruhe) levels (cd36: r=0.568, r²=0.323, P=0.027; fabp3: r=0.646, r²=0.417, P=0.0092; cpt1: r=0.694, r²=0.481, P=0.0041; Fig. 1M).

We found significant changes in the expression of genes associated with calcium ion signalling and transport, cellular stress and metabolic status, but not with migratory behaviour. There was a significant difference in mRNA levels of both Ca²⁺-signalling kinases (camk1: F_2,12=39.50, P<0.0001; camk2a: F_2,12=44.64, P<0.0001) and ATPase sarcoplasmic transport protein (atp2a2: F_2,12=13.82, P=0.0008, one-way ANOVA), but with subtle differences (Fig. 2A–C). As compared with the SD state, camk1 mRNA levels were higher in the LD-S state and camk2a levels were higher in the LD-eS state, whereas atp2a2 mRNA levels were higher in both the LD-eS and LD-S state (P<0.05, Newman–Keuls post hoc test; Fig. 2A–C).

Likewise, we found significant differences in the expression of genes associated with cellular and metabolic (oxidative) stress (hspa8: F_2,12=20.03, P=0.0002; sod1: F_2,12=17.82, P=0.0003; however, no difference in nos2: F_2,12=0.702, P=0.515; one-way ANOVA; Fig. 2D–F). The overall hspa8 and sod1 mRNA levels were higher in the LD-S state than in the SD and LD-eS state (P<0.05, Newman–Keuls post hoc test; Fig. 2D–F).

Consistent with differential energy demands, there were significant changes in the expression of genes that reflect metabolic status (apoa1: F_2,12=38.34, P<0.0001; sirt1: F_2,12=26.05, P<0.0001; one-way ANOVA). Expression of both apoa1 and sirt1 was increased consistently during the LD conditions, with mRNA levels in the order SD<LD-eS<LD-S state (P<0.05, Newman–Keuls post hoc test; Fig. 2G,H).

However, we found no difference in adcyap1 (F_2,12=1.539, P=0.254) and vps13a (F_2,12=2.155, P=0.158; one-way ANOVA) expression, which were measured as markers of effects on migratory behaviour (Fig. 2I,J).

We assayed and compared MBH-specific proteome-wide changes between the LD-eS and LD-S states; exclusion of SD precluded any photoperiod effect. We identified a total of 503 proteins (Table S2). These enriched 27 functional pathways including the highly enriched carbon metabolism, TCA cycle, glycolysis and oxidative phosphorylation (Fig. 3A). A small number of proteins (28) were found to be differentially expressed between the LD-eS and LD-S states, with 21 upregulated and 7 downregulated proteins in the LD-S compared with the LD-eS state (Figs 3B, 4A). STRING network analysis revealed a mutual interaction of 11 out of the 28 differentially expressed proteins (Fig. 3B), and the enrichment of five functional pathways, namely astrocyte development (P=1.37E−04; PLP1, TSPAN2, VIM), regulation of neurotransmitter secretion (P=0.0160; SNCG, CAMK2A), glycolytic process (P=0.0337; PGAM1, GAPDH), intermediate filament protein (P=0.0460; VIM, NEFL) and PPAR signalling (P=0.0451; APOA1, FABP4) (Fig. 4B).

The present results show the molecular basis of FA biosynthesis and its transport in parallel with the photostimulated indices of the readiness for vernal migration in buntings. Here, we report elevated mRNA expression of acc and dgat2 (acc mRNA levels in LD-S>SD, and dgat2 mRNA levels in LD-S>SD and LD-eS) genes that code for ACC and DGAT2 enzymes involved in the hepatic FA biosynthesis pathway at the initial and final steps, respectively (Lu et al., 2015). At the same time, a similar fasn mRNA expression in all three states is inconsistent with the reported association of FASN enzyme activity with pre-migratory weight gain in buntings (Banerjee and Chaturvedi, 2016). Differences in acc and fasn gene expression in buntings might indicate their differential roles in lipogenesis: whereas acc-encoded ACC acts as the main rate-limiting enzyme and promotes lipogenesis at higher rates as required during the migratory state, the fasn-encoded FASN enzyme controls lipogenesis at a lower rate to fulfil short-term energy needs (Donaldson, 1979).

Changes in the hepatic apoa1 and muscular cd36 and fabp3 paralleled the acc and dgat2 mRNA levels, suggesting protein-mediated transport of FAs to the flight muscles, concurrently with their synthesis in the liver in the migratory state. FABP levels were also found to be elevated in the migratory state of western sandpipers (Calidris mauri; Guglielmo et al., 2002), white-throated sparrows (McFarlan et al., 2009) and blackheaded buntings (Srivastava et al., 2014). Elevated cpt1 mRNA levels in flight muscles further support increased FA supply in the LD-S state. We suggest that the rate-limiting CPT1 enzyme mediates the mitochondrial uptake and β-oxidation of fatty acids (Shriver and Manchester, 2011) in ‘working’ muscles to meet energy demands of the migratory state in buntings.

We associate changes in gene expressions in MBH with activation (or inhibition) of hypothalamic molecular switches controlling the photoperiodic induction of the seasonal spring migration and gonadal maturation. For example, augmented Ca²⁺ transport, as indicated by upregulated camk2a, camk1 and atp2a2 expression, could be indicative of the overall alteration in neural processes, including the neurogenesis and maintenance of synaptic connections in the LD-S state (Berridge et al., 2000; Bouron, 2020; Boss et al., 2016). Although it is purely speculative at this time, we associate upregulated Ca²⁺ transport with enhanced neurogenesis and high neuronal activity in the MBH in order to integrate different sensory inputs and use the acquired information for navigation (Åkesson and Hedenström, 2007; Muheim et al., 2006; Warren et al., 2010). Interestingly, camk2a and camk1 showed increased expression in LD-eS and LD-S, respectively, whereas atp2a2 mRNA levels were elevated in both LD-eS and LD-S states. This might mean differential activation of molecular gears (enzymes) of the Ca²⁺-signalling and transport cascade. Whereas calcium/calmodulin-dependent kinases were activated at different times, the ATPase sarcoplasmic Ca²⁺ transporting protein was activated consistently during the LD exposure. Elevated atp2a2 mRNA levels were also found in photostimulated migratory blackheaded buntings (Sharma et al., 2018b). Perhaps, calcium also plays a role in switching on the necessary genomic changes (Bading, 2013) with the photostimulated transition from the non-migratory to migratory state in avian migrants. However, we would like to emphasize that the suggested Ca²⁺-signalling and transport pathway does not preclude the involvement of alternative pathways including the release of hunger hormones, PPAR signalling, etc., in the photoperiodic induction of seasonal responses in migratory birds.

We also found change in hspa8 and sod1 gene expression, suggesting concurrent effects on physiological processes that include cellular stress and energy homeostasis in the MBH (Jones et al., 2008; Raja-Aho et al., 2012; Sharma et al., 2018b; Sharma and Kumar, 2019). We interpret that hspa8-encoded ATP-dependent molecular chaperones play a role in the protein quality check, and regulate protein transport and sorting via de-assembly of clathrin-coated vesicles (Stricher et al., 2013). Similarly, the enhanced sod1 mRNA levels in the LD-S state are consistent with SOD1 playing a protective role against free superoxide radicals, which are probably produced in larger amounts during the photostimulated state. At the same time, the lack of change in nos2 mRNA levels could be a neuroprotective response, as a prolonged cellular stress as might happen during the photostimulated migratory state could eventually increase nitric oxide, as well as nitric oxide-dependent post-translational redox modifications such as S-nitrosylations (SNO), which may interfere with neuronal functions and longevity (Valek et al., 2019). However, we caution that further experiments may be necessary to identify many more candidate molecules that constitute the multifaceted oxidative stress pathway system.

Further, changes in apoa1 and sirt1 expression are consistent with the idea of the involvement of a periphery–central feedback molecular circuit in the metabolic regulation (Cakir et al., 2009). We suggest that the MBH actively assesses the nutritional status and integrates appropriate responses at the whole-body level in order to respond to energy needs of the migratory state. Consistent with this, elevated apoa1 and sirt1 expression was found in the brain of northern wheatears and blackheaded buntings, respectively, when they were photostimulated and accumulated fat stores under stimulatory long days (Trivedi et al., 2014; Frias-Soler et al., 2020). This suggests the importance of apoa1-encoded ApoA1 (a major cholesterol transporter protein) in the photoperiodic induction of the migratory phenotype (Frias-Soler et al., 2020), although possible mechanisms remain unclear. Likewise, we suggest the role of sirt1 in the regulation of appetite and metabolism, which are key components of the migratory state, based on evidence from mammals (Yamamoto and Takahashi, 2018). SIRT1 overexpression in pro-opiomelanocortin (POMC) neurons enhances energy expenditure, and its overexpression in agouti-related protein (AgRP) neurons suppresses food intake; this further suggests the neuron-specific role of SIRT1 in energy homeostasis (Yamamoto and Takahashi, 2018).

The lack of change in adcyap1 and vps13a expression reinforces the view that they are not part of the hypothalamic gene switches that are involved in the photoperiodic induction of the spring migratory phenotype. This is not surprising as this study did not test the migratory behaviour per se. Notably, variation in the allele size of adcyap1, not its expression level, was related to the intensity of Zugunruhe in European blackcaps (birds with longer adcyap1 alleles displayed higher Zugunruhe; Mueller et al., 2011), but not in migratory juncos (Junco hyemalis; Peterson et al., 2013). Similarly, the association of vps13a expression was reported to relate to the migration directionality, not migratory state, of genetically closely related golden-winged warblers (Vermivorachrysoptera) and blue-winged warblers (Vermivoracyanoptera) migrating to winter in Central and South America, respectively (Toews et al., 2019). However, we will not completely discount an indirect role of the adcyap1 gene, via its role as a light-sensitive molecule in circadian clock effects (Nagy and Csernus, 2007). The circadian clock regulates the phase transition in daily activity behaviour, when diurnal songbirds become predominantly night active with the onset of the migratory state (Bartell and Gwinner, 2005; Rani et al., 2006). Both adcyap1 and vps13a could still be among candidate molecules of a ‘migratory gene package’ that via epistatic interactions determines/modifies the migration-associated behaviour and physiology in songbirds (Liedvogel et al., 2011).

The 503 annotated proteins were clustered in 27 physiological pathways, indicating a large coverage of the functional hypothalamus proteins by LC/MS-MS. The 28 differentially expressed proteins support the overall gene expression results on the role of Ca²⁺ signalling and transport, cellular stress and metabolic status, and protein-mediated transport of FA in the LD-S state. For example, higher CAMK2A protein levels suggest an activated calcium signalling pathway, perhaps for neuroadaptation in the migratory state. Similarly, low PRDX4 levels are consistent with high sod1 and unchanged nos2 mRNA levels in the LD-S state, suggesting an overall modulation of the anti-oxidant defence system at the MBH level (Schulte, 2011). Likewise, the downregulated PLP1 (proteolipid protein) and TSPAN2 (tetraspanin-2) of axon development and astrocyte development pathways probably played an activational role in shaping of the neural processes in the LD-S state. Although unknown in birds, single PLP or TSPAN2 knockout showed moderate, while double (PLP and TSPAN2) knockouts showed an enhanced activation of the astrocytes in mice (de Monasterio-Schrader et al., 2013).

There were significant protein-level changes, suggesting both re-modelling and transport of proteins at the cellular level, as might be required for an enhanced metabolic homeostasis in the LD-S state. In particular, we found upregulated expression of alpha and beta subunits of ATP-synthase (ATP5A1 and ATP5B), indicating enhanced ATP synthesis, and increased synthesis of ATP-dependent molecular chaperone HSPA8 and HSP90AA1 regulating protein transport (Stricher et al., 2013), PPIA involved in protein folding, and NME7 responsible for de novo formation of the microtubules (Liu et al., 2014). Upregulated TUBA1A, NEFL and TPPP, and concurrently downregulated VIM (vimentin) and SNCG (gamma-synuclein) proteins, further suggest cytoskeleton-dependent cellular remodelling of the proteins in the LD-S state. A similar differential hypothalamic cytoskeletal protein expression has been found between non-migratory and migratory states in Swainson's thrushes (Catharus ustulatus; Johnston et al., 2016). It may be noted that in Swainson's thrushes the characterization of both migratory and non-migratory states was based on fat accumulation and night activity, similar to that in the present bunting study.

We further propose that peripheral fat accumulation may serve as a physiological predictor of the vernal migratory state. Upregulated APOA1 and FABP4, which are FA transport proteins and enrich the PPAR signalling pathway, may allow easy FA access through the blood–brain barrier either by passive diffusion or by using the ATP-dependent transporter proteins in the LD-S state (Tracey et al., 2018). This remains purely speculative at this time, however. Also, high GAPDH and PGAM1 protein levels, which catalyse the reversible steps of the glycolysis/gluconeogenesis metabolic pathway, suggest the overall increase of cellular metabolism in buntings; this can be important for flight-associated aerobic exercise in the migratory state.

In this first MBH-specific proteome-wide study of a migratory songbird, we have identified a subset of hypothalamic proteins that are potentially important and differentially expressed between LD-eS and LD-S states, as could be predicted from the gene expression results. In photostimulated buntings, we show that changes in the expression of cytoskeletal proteins and of those associated with astrocyte and axon development and PPAR signalling paralleled with the photostimulated migratory phenotype and testis maturation. These important results fill some gaps in our understanding of how the peripheral accumulation of fat could possibly serve as a physiological predictor of the migratory preparedness and readiness for vernal migration in obligate migratory songbirds.

A key question is what all the suggested mechanistic relationships based on molecular changes in the MBH mean for the physiology and/or ecology of a migratory bird. Although, the current study was not designed to answer this, we would argue that the activation (or inhibition) of molecular switches controlling the photoperiodic induction of migratory phenotype reinforces the view that hypothalamic substrates act as key integrators of the cascade of mechanisms that govern anticipatory energy intake in migrants in order to prepare them physiologically, so that they begin flying in time and arrive at their destination when conditions are most favourable. It may be noted further that the present molecular changes could account for the effects on metabolic processes that are linked to both the migratory and reproductive phenotypes, as captive buntings show body fattening and weight gain, as well as testis maturation when they are exposed to stimulatory LD conditions. This seems to have relevance to the overall ecology of avian migration, as the success of migration depends on the appropriate timing across multiple physiological systems that govern various sub-cycles of the annual life history, such as moult and reproduction, of a migratory species.

Finally, we show a series of molecular changes that include various physiological and metabolic pathways, but how these molecular pathways have functional connectivity and coordinate migration in songbirds remains unclear. This requires further study, perhaps in a field setting, to better address on the molecular mechanism(s) controlling the development of seasonal migration in avian migrants.

We thank Dr Pawan Malhotra, International Centre for Genetic Engineering and Biotechnology, New Delhi, India, for the help with quantitative proteomics.