Transplanted hypothalamic neurons restore leptin signaling and ameliorate obesity in db/db mice

Abstract

Evolutionarily old and conserved homeostatic systems in the brain, including hypothalamus, are organized into nuclear structures of heterogeneous and diverse neuron populations. To investigate whether such circuits can be functionally reconstituted by synaptic integration of similarly diverse populations of neurons, we generated physically chimeric hypothalami by micro-transplanting small numbers of embryonic enhanced green fluorescent protein-expressing leptin-responsive hypothalamic cells into hypothalami of postnatal leptin receptor-deficient (db/db) mice that develop morbid obesity. Donor neurons differentiated and integrated as four distinct hypothalamic neuron subtypes, formed functional excitatory and inhibitory synapses, partially restored leptin responsiveness, and ameliorated hyperglycemia and obesity in db/db mice. These experiments serve as proof of concept that transplanted neurons can functionally reconstitute complex neuronal circuitry in the mammalian brain.

To functionally repair complex circuitry in the central nervous system (CNS), newly incorporated neurons should be electrophysiologically and functionally integrated, whether transplanted or derived from endogenous progenitors. Newly integrated neurons might optimally be a single neuronal subtype in some “point-to-point” sensori-motor CNS systems, or instead a diverse and nucleus-specific population of neuronal subtypes for some critical, homeostatic systems organized into distributed nuclear structures. These systems often signal polymodally via neuropeptides, and are regulated by paracrine and endocrine mechanisms via multiple neuronal subtypes operating in parallel. One such critical homeostatic system is that of leptin signaling in hypothalamus, regulating energy balance, glucose, food intake, and body weight.

The adipocyte-derived hormone leptin signals through a variety of hypothalamic neuropeptide producing neurons; defective leptin signaling in the CNS due to gene mutations in either leptin (ob/ob) or its receptor (db/db) results in severe obesity and diabetes in both humans and rodents (1, 2). Deletion or rescue of leptin receptors specifically in the CNS is sufficient to fully reproduce or rescue, respectively, the obesity and diabetic phenotypes (3, 4). Recent studies link enhanced adult neurogenesis in hypothalamus to persistent body weight reduction (5, 6); a significant number of these newborn neurons were leptin-responsive. Leptin receptors are highly expressed by diverse populations of hypothalamic neurons; these include anorexogenic proopiomelanocortin (POMC) neurons and orexogenic neuropeptide Y (NPY) neurons. Cell type-specific deletion of leptin receptors selectively in POMC neurons or in both POMC and NPY neurons only partially replicates the phenotype of db/db mice (7). This demonstrates that additional neuronal populations in hypothalamus or other brain regions are required for this modulatory system to function correctly, and suggest that local hypothalamic restoration of leptin receptor signaling by diverse neuron subtypes might lead to partial phenotypic rescue.

To investigate whether partial circuit reconstitution with the diverse hypothalamic neuronal populations involved in leptin signaling might substantially reverse energy balance defects in db/db mice, we generated physical hypothalamic chimeras via micro-transplantation of small numbers of dissociated immature, wild-type, eGFP-expressing, leptin-responsive, hypothalamic cells (~1.5 × 10⁴ per hypothalamus) into medial hypothalamus of postnatal day 0.5 (P0.5) to P5.5 db/db mice under high-resolution ultrasound guidance (8, 9) (Fig. 1A). Donor cells were developmentally appropriate E13.5 immature neurons and progenitors, poised to develop into leptin-responsive hypothalamic neurons, able to integrate in circuitry (10–18); (fig. S1; S text 1).

Fig. 1. Transplanted E13.5 eGFP+ hypothalamic cells mature into appropriate hypothalamic neuron subtypes.

(A) Ultrasound-guidedmicrotransplantation of dissociated E13.5 hypothalamic cells into early postnatal db/db mouse hypothalamus in sagittal view. The ~100-mm-diameter pulled glass micropipette (MP) appears on the ultrasonogram much larger than its real size due to ultrasound shadow. (B) Schematic of a coronal brain section magnified in C. (C) Location and survival of transplanted eGFP+ cells in the representative recipient hypothalamus 20 weeks aftermicrotransplantation. (D and E) Representative confocal three-dimensional (3D) reconstructions of donor-derived eGFP+ neurons in the VMH of db/db mice 20 weeks after microtransplantation, showing (D) a POMC neuron identified by the cleavage product b-endorphin and (E) a NPY neuron. For each confocal reconstruction, the x-z plane is shown at the bottom and the y-z plane is shown at the right. (F to I) Four distinct electrophysiological donor-derived eGFP+ neuron subtypes. (F) The firing behavior of a representative newly integrated eGFP+ donor-derived NRRS neuron. “Type A” and “type C” neurons can be distinguished by the absence or presence, respectively, of delayed firing (bottom traces). Note the delayed firing in “type C” neurons (arrow) when hyperpolarizing current is withdrawn. (G) Representative donor-derived eGFP+ RRS neuron. An apparent voltage sag (dashed line) and a single rebound action potential were typical. (H) Representative donorderived eGFP+ BS neuron. A low-threshold calcium spike and cluster of rebound action potentials were typical in response to hyperpolarizing current injection. (I) Representative donor-derived eGFP+ FS neuron. A high firing frequency and absent firing adaptation are two typical characteristics of this neuron type. DS, dorsum sellae (bone); ST, sella turcica (bone); AC, anterior commissure; S, skin and fur on the head; 3V, third ventricle. Asterisks demarcate the rostrocaudal extent of the transplantation tracks. Scale bars in (A), 1 mm; in (C), 100 mm; in (D) to (I), 10 mm.

Twenty weeks after micro-transplantation, clusters of eGFP⁺ cells were located primarily in the arcuate nucleus (ARC), ventromedial hypothalamic nuclei (VMH), dorsomedial hypothalamic nuclei (DMH, not shown), and lateral hypothalamic nuclei (LH, not shown) (Figs. 1B, 1C; fig. S2). Most displayed typical polarized hypothalamic neuron morphology (fig. S3). Fewer eGFP⁺ cells were astroglia (figs. S3, S4). Double-labeling immunocytochemistry revealed 45 ± 3% of eGFP⁺ cells expressing the mature neuronal marker NeuN, while 4 ± 4% expressed the mature astroglial marker S100β (fig. S3). Transplanted cells differentiated into several neuronal subtypes, distinguished using specific markers: pan neuronal HuC/D; GABA; GABAergic neuron calbindin; and dopaminergic synthetic tyrosine hydroxylase (TH) (fig. S5). Notably, some newly incorporated eGFP⁺ cells expressed neuronal subtype markers linked to regulation of energy balance. Seven ± 5% of transplanted eGFP⁺ cells in representative sections developed and matured as POMC neurons (identified by immunocytochemistry directed against the POMC cleavage product β-endorphin) (Fig. 1D). Fewer eGFP⁺ cells differentiated and survived as NPY neurons (Fig. 1E). These morphologic and neurochemical results demonstrate that transplanted eGFP⁺ cells mature and survive in recipient brains for at least 5 months, and develop into multiple distinct neuron types that normally reside in the hypothalamus.

To further classify the diverse neuronal subtypes of the newly incorporated neurons via their distinct electrophysiological properties, and to determine whether they functionally integrate into recipient hypothalamic circuitry, we performed whole-cell patch clamp recordings of eGFP⁺ cells in acute brain slice from the hypothalamus of recipient db/db mice, using reported methods (19). We identified four different neuronal subtypes with distinct electrophysiological properties (Figs. 1F–I, table S1). The first and most frequent subtype, non-rebound regular-spiking neurons (NRRS, n = 17) (Fig. 1F), showed an adapting, regular firing response to depolarizing current injections, with no voltage sag and no rebound action potentials fired following a strong hyperpolarizing current injection. These properties are typical of “type A” and “type C” neurons found in ARC (20) and VMH (21), and type 1 neurons found in LH (22). The second subtype, rebound regular-spiking neurons (RRS, n = 7) (Fig. 1G), displayed both an adapting firing response to depolarizing current injections, and a depolarizing sag during hyperpolarizing current injections (table S1) followed by a single rebound action potential, suggesting the presence of hyperpolarization-activated cation currents (Ih). This is typical of hypothalamic POMC neurons (23) and type 2 neurons of LH (22). The third subtype, burst-spiking neurons (BS, n = 7) (Fig. 1H), displayed a cluster of rebound sodium action potentials overlying a slower calcium spike in response to hyperpolarizing current injections, indicating the presence of a low threshold T-type Ca²⁺ conductance. This firing behavior is similar to that of type B neurons found in both ARC (20) and VMH (21). The fourth subtype, fast-spiking neurons (FS, n = 5) (Fig. 1I), displayed non-adapting, high frequency firing of narrow sodium spikes (table S1) with long depolarizing current injections. These properties are typical of fast spiking GABAergic neurons and the type 3 neurons in LH (22). These results indicate that transplanted immature wild-type hypothalamic neurons develop and mature electrophysiologically into a specific and diverse set of neuronal subtypes typical of the normal hypothalamic neuron populations (tables S1, S2).

To investigate whether these newly incorporated eGFP⁺ neurons are functionally integrated into the recipient db/db hypothalamic circuitry, we examined whether synapses are formed between native and donor eGFP⁺ neurons. We first used immunocytochemical analysis of synaptophysin (a presynaptic marker) for synaptic localization. Synaptophysin immunopositive puncta developed on the somas and processes of eGFP⁺ neurons (Figs. 2A, 2B). NPY-positive neurons had extensive contacts with the transplanted eGFP⁺ cells (fig. S6). We then examined functional synaptic transmission by assessing for post-synaptic currents in the transplanted neurons. Importantly, all recorded transplanted neurons that fired action potentials also displayed spontaneous inhibitory (sIPSC) and spontaneous excitatory (sEPSC) postsynaptic currents under voltage clamp conditions (Fig. 2C, fig. S7). The four neuronal subtypes showed similar mean sEPSC and sIPSC amplitudes and frequencies (fig. S7). Electron microscopy employing anti-GFP immunogold labeling to unequivocally identify transplanted neurons further confirmed that synapses form between native and donor-derived eGFP⁺ neurons (fig. S8). The presence of synaptophysin immunopositive puncta and spontaneous synaptic events indicate that eGFP⁺ neurons receive synaptic inputs in the chimeric hypothalamus. To investigate whether eGFP⁺ donor neurons form functional synapses onto native neurons, we performed dual whole-cell patch clamp recordings of eGFP⁺ and native neurons. Fast synaptic connections from eGFP⁺ to native neurons were observed (n = 3; Fig. 2D). Firing of different eGFP⁺ neurons (green tracings) elicited either an EPSP (upper trace) or IPSP (lower trace) in adjacent native neurons (Fig. 2D). Together, the immunofluorescence labeling, electrophysiological, and ultrastructural results strongly indicate that incorporated eGFP⁺ neurons form reciprocal synaptic connections with native hypothalamic circuitry for as long as 5 months following micro-transplantation.

Fig. 2. Transplanted eGFP+ hypothalamic neurons form excitatory and inhibitory synaptic connections and response to the energy state signals leptin, glucose, and insulin.

(A and B) Donor-derived eGFP+ neurons in db/db recipients bear extensive appositions with synaptophysin-positive puncta (red). (B) Higher-magnification confocal image of the boxed region of a dendrite in the merged confocal image in (A), showing synaptophysin-positive puncta indicating synaptic contacts. (C) Spontaneous EPSCs and IPSCs were observed in eGFP+ neurons. (D) Representative dual whole-cell recording showing that stimulation of donor-derived eGFP+ neurons (green traces) elicits EPSPs (upper) or IPSPs (lower) in native neurons (black traces). (E) Representative confocal 3D reconstruction of donor-derived eGFP+ neuron in the VMH of db/db mice 20 weeks after microtransplantation with leptin-induced signaling indicated by phosphorylated STAT3 in response to leptin administration. (Bottom) x-z plane; (right) y-z plane. (F) Representative voltage tracing from a leptindepolarized eGFP+ neuron. Leptin depolarized the membrane potential and substantially increased the firing rate; insulin had the opposite effect. (G) Representative voltage tracing from a leptin-hyperpolarized eGFP+ neuron. Both leptin and insulin hyperpolarized the membrane potential and substantially decreased the firing rate; low glucose had the opposite effect. Scale bars in (A), 8 mm; in (B), 2 mm; in (E), 10 mm.

To investigate potential leptin responsiveness of newly integrated eGFP⁺ neurons, we both examined a known leptin-activated signaling pathway using immunocytochemistry, and measured the electrophysiological response to leptin in these neurons. The phosphorylation and nuclear translocation of STAT3 is a well characterized leptin signaling response (24–26). After intraperitoneal administration of leptin to hypothalamically chimeric db/db mice, 8 ± 3% of eGFP⁺ neurons exhibited STAT3 phosphorylation (Fig. 2E), compared with 0% in the recipient (eGFP⁻) db/db hypothalamic neurons lacking leptin receptors. We further investigated the effects of leptin on action potential firing and membrane potential in 17 eGFP⁺ transplanted neurons and in endogenous neurons in hypothalamically chimeric db/db mice. Leptin (100 nM) depolarized the membrane potential (4.5 ± 1.2 mV, n = 4) and increased the spontaneous firing rate (96.0 ± 51.6 %, n = 4) in 4 of the 17 neurons tested (Fig. 2F, table S3). This response is typical of hypothalamic POMC neurons (27). Consistent with this conclusion, 3 of 4 leptin-depolarized neurons also displayed the RRS firing behavior typical of POMC neurons. In contrast, leptin hyperpolarized the membrane potential (−3.1 ± 0.6 mV, n = 10) and decreased the spontaneous firing rate (−62.3 ± 6.4 %, n = 10) in the majority of NRRS and BS neurons in the ARC, VMH, and LH, as well as in FS neurons in the mammillary nucleus (n = 10; Fig. 2G, table S3). This inhibitory response to leptin is typical of NPY neurons (28). Only three of the 17 eGFP⁺ neurons were insensitive to leptin. In contrast, all recordings from neighboring eGFP⁻ neurons in the same hypothalamically chimeric db/db brain slices displayed no response to leptin (n = 3; fig. S9). These results indicate that newly incorporated eGFP⁺ neurons reconstitute the typical diversity of excitatory and inhibitory leptin responses found in the normal hypothalamus.

We next investigated whether chimerically-incorporated and synaptically-integrated neurons exhibit electrophysiological responses to glucose and insulin typical of neurons in these hypothalamic nuclei. Hypothalamic neurons with leptin responses typically display distinct matching insulin and glucose responses (27). Like native neurons, all leptin-depolarized eGFP⁺ donor-derived neurons examined hyperpolarized in response to 100 nM insulin (Fig. 2F, table S3) (n = 3). In response to reduction of glucose from 10 mM to 5 mM, two of four leptin-depolarized eGFP⁺ neurons exhibited membrane hyperpolarization, and decreased their firing rates (table S3), typical of normal leptin-depolarized hypothalamic neurons (23, 29, 30). Two of three leptin hyperpolarized neurons also hyperpolarized in response to insulin (Fig. 2G, table S3), while one neuron was strongly excited by insulin (table S3) (29). Of seven leptin-hyperpolarized eGFP⁺ neurons with glucose reduction, four were significantly depolarized (Fig. 2G, table S3), normal for hypothalamic neurons (31). These results indicate that newly integrated neurons exhibit responses to glucose, insulin, and leptin typical of native neurons of medial hypothalamus.

We next investigated whether these chimeric, leptin-responsive, hypothalamic neuronal circuits can functionally regulate peripheral energy balance and body weight. To investigate this question, we analyzed physiological effects of these donor-derived neurons on body weight, comparing db/db mice with integrated hypothalamic neuron transplants vs. four relevant control groups (SOM text 2). Db/db mice with hypothalamic neuron transplants displayed significantly reduced body weight compared to non-operated or sham operated db/db mice (Fig. 3A), intermediate between db/db and control non-obese mice. Sham experiments, in which cell-free medium was injected into the same hypothalamic location of db/db mice, revealed no effect on body weight compared to db/db non-operated littermates (Fig. 3A; fig. S10). Similarly, micro-transplantation of E13.5 wild-type hypothalamic eGFP⁺ cells into hypothalami of control non-obese mice did not result in body weight differences compared to non-operated non-obese littermates (Fig. 3A). To further investigate the specificity of the effect of transplanted immature hypothalamic neurons and progenitors, we micro-transplanted immature neurons from the neocortex isolated during an equivalent developmental stage (E16.5–E18.5) (11, 12, 32) to control for potential non-specific incorporation of non-leptin-responsive, non-hypothalamic neurons (SOM text 3). Twenty weeks after transplantation, almost all surviving neocortical donor cells were S100β positive glia that did not fire action potentials (fig. S11). These results with neocortical donor cells confirm prior results in the neuronal transplantation literature that the specific sources and subtypes of donor neurons are critical for successful neuronal circuit integration and survival (11). The body weight of db/db mice transplanted with cortical cells was unaltered (fig. S10A), reinforcing the specific ability of E13.5 hypothalamic donor neurons and progenitors to reconstitute functional hypothalamic circuitry. Taken together, these results demonstrate that leptin-responsive donor neurons derived from wild-type E13.5 hypothalamus are able to ameliorate obesity in hypothalamically chimeric db/db mice, following functional integration and partial circuit reconstitution. This decrease in body weight is equivalent in magnitude to the increase observed when leptin receptors are selectively deleted from the ventromedial hypothalamic nucleus (33), reinforcing the physiological level of this repair.

Fig. 3. Transplanted E13.5 wild-type immature hypothalamic neurons reduce body weight, fat mass, serum leptin, and glucose levels in leptin receptor–deficient db/db mice.

(A) Hypothalamically chimeric db/db mice (db/db mice microtransplanted with E13.5 hypothalamic cells into the medial hypothalamus) have significantly lower body-weight gain than nonoperated db/db mice or sham-operated db/db mice (F2,361 = 120.75; chimeric db/db versus db/db nonoperatedmice, P < 0.001; chimeric db/db versus db/db sham-operated mice, P < 0.001). Microtransplantation of E13.5 hypothalamic cells into nonobese control mice did not change body weight compared with nonobese, nonoperated littermates. (B) Analysis of body composition using DEXA demonstrates significant fat mass reduction in transplanted compared with nontransplanted obese (db/db) mice. (C and D) Analysis of blood parameters demonstrates statistically significant reductions in (C) serum leptin levels at 13 weeks of age and (D) blood glucose at 9 and 13 weeks of age. *P < 0.05; **P < 0.01; ***P < 0.001. Data reported asmean T SEM.

To further investigate tissue-specific contributions to body weight reduction in transplanted db/db mice, post-mortem body composition was assessed using dual energy X-ray absorptiometry (DEXA). There was a significant decrease in the fat mass (Fig. 3B), but not in lean mass (fig. S12A). Of note, food intake was not significantly altered in these mice, suggesting that changes in fat mass were likely due to increased energy expenditure (fig. S12B). Consistent with the body weight reduction, serum leptin was also significantly reduced (Fig. 3C).

We further investigated the efficacy of central hypothalamic cell transplantation in restoring peripheral metabolic parameters of blood glucose and serum insulin, two central biophysiological parameters that are dysregulated in db/db mice. Hypothalamically chimeric db/db mice displayed a large and significant reduction of blood glucose levels at both 9 and 13 weeks after transplantation (Fig. 3D), consistent with the known central role of hypothalamic leptin receptors in peripheral glucose regulation (34). This marked rescue of peripheral glucose homeostasis occurred despite the lack of difference in serum insulin relative to controls (fig. S12C). Taken together, these analyses indicate that newly synaptically integrated hypothalamic neurons significantly ameliorate energy balance and metabolic dysregulation in db/db mice.

In summary, these results demonstrate that transplanted immature, newborn hypothalamic neurons and progenitors of appropriate developmental stage can chimerically integrate into hypothalamic circuitry as functional neurons with subtype diversity typical of the normalhypothalamus, and that this relatively small number of functionally integrated wild-type neurons is sufficient to ameliorate obesity, hyperleptinemia, and hyperglycemia in db/db mice.

At least two possible network mechanisms might explain how this relatively few newly integrated neurons partially rescues the db/db phenotype. Rescue might be solely due to the new neurons themselves. Alternatively, the relatively few new, cell-autonomously leptin-responsive neurons might modulate their firing in response to systemic leptin levels, secondarily regulating other neurons in hypothalamic circuitry. Thus, transplanted neurons might act as leptin sensors whose output is then transmitted by endogenous, hypothalamic circuitry still unresponsive to direct leptin activation. Either way, functional synaptic integration and leptin responsiveness by newly incorporated wild-type neurons imparts important organism-level behavioral rescue. These experiments serve as a “proof of concept” for cellular repair of critical modulatory homeostatic systems that are often organized with diverse neuronal subtypes contributing in parallel “push-pull” circuitry. Substantial restoration of appropriate regulatory function is possible by functional integration of relatively few appropriately diverse neuronal subtypes via micro-transplantation of developmentally appropriate immature neurons and progenitors.

Supplementary Material

supplement 1

Supporting Online Material

www.sciencemag.org

Materials and Methods

Supporting texts

Reference (35)

Supporting figures S1 to S12

Supporting tables S1 to S3