Animals were anesthetized with isoflurane (2% induction; 1–1.5% maintenance doses) and placed in a David Kopf stereotaxic apparatus (Tujunga, CA, USA). Body temperature was set at 37°C through a water-heated pad (Gaymar T/Pump, Orchard Park, NY, USA). The sagittal midline of the scalp was sectioned and retracted. A small craniotomy was drilled over the HDB nucleus (coordinates from Bregma: 0.5 mm anterior 0.8 mm lateral, 5.5 mm–6 mm depth from the cortical surface).

Multiunit activity (MUA) was recorded in the HDB nucleus through a tungsten microelectrode (1–2 MΩ, World Precision Instruments, WPI, Sarasota, FL, USA). Electrocorticogram (ECoG) was also recorded in the primary somatosensory (S1) cortex (from Bregma −2 mm posterior, 3 mm lateral, 1 mm depth) through tungsten macroelectrodes (<1 MΩ, WPI). MUA and ECoG were filtered between 0.3 and 3 kHz and 0.3 Hz-100 Hz, respectively, and amplified using a DAM80 preamplifier (WPI). Signals were sampled at 10 or 1 kHz, respectively, through an analog-to-digital converter (Power 1401 data acquisition unit, Cambridge Electronic Design, Cambridge, UK) and fed into a PC for offline analysis with Spike 2 software (Cambridge Electronic Design).

Whisker deflections were evoked by brief air pulses using a pneumatic pressure pump (Picospritzer, Hollis, NH, USA; 1–2 kg/cm², 20 ms duration), delivered through a 1 mm inner diameter polyethylene tube. All whiskers were first trimmed to a length of 5 mm to avoid complex responses due to multiple whiskers’ deflections. The experimental protocol consisted of 60 air pulses delivered to the principal whisker at 0.5 Hz after a control period of 1 min for ECoG basal recording.

Optogenetic experiments were performed to identify cholinergic neurons (ChAT+ identified neurons) through blue-light activation, using the transgenic mice (see above). Animals were anesthetized and prepared as indicated above. Optical stimulation of ChR-expressing neurons was achieved with light pulses from a light-emitting diode (LED; Thomas Recording, Germany) delivered through an optrode, composed of a tungsten microelectrode 0.5–0.8 MΩ attached to an optical fiber (core diameter 120 μm; Thomas Recording), stereotaxically positioned in the HDB nucleus. Optogenetic stimulation was applied by long-lasting pulses of 473 nm light (27 stimuli with a duration of 300 ms, each stimulus was repeated every 3 s), with an illumination intensity of <30 mW/mm². Unit recordings were performed through the optrode. They were filtered (0.3–3 kHz) and amplified using a DAM80 preamplifier (WPI). Single-unit activity was sampled at 10 kHz and extracted with the aid of Spike2 software.

IGF-I was locally delivered in the HDB nucleus (10 nM; 0.2 μl; coordinates as above) employing a 1 μl Hamilton syringe. Besides, IGF-I was injected systemically (1 μg/g; i.p.) in other experiments. The muscarinic receptor antagonist atropine (1 mg/Kg in 0.9% NaCl i.p.) was administered 15 min before IGF-I systemic injection to assess whether IGF-I effects were due to activation of muscarinic receptors. Orexin A (Tocris, Spain) was also injected into the HDB nucleus (10 nM; 0.2 μl; coordinates as above).

In a set of experiments, we administrated human IGF-I (hIGF-I) through Alzet osmotic mini-pumps (Model 1004; USA) for chronic administration (Pre-Protech, USA; 50 g/kg/day) in ChAT-ChR2-YFP animals or the vehicle (saline solution). Pumps were implanted subcutaneously between the scapulae, following the manufacturer’s instructions. Treatment lasted 28 days. After that, animals were submitted to electrophysiological recordings as described above.

Animals were euthanized with an overdose of pentobarbital (50 mg/kg) and perfused transcardially with saline 0.9% followed by 4% paraformaldehyde in 0.1 N phosphate buffer (PB), pH 7.4. Coronal 50-μm-thick brain sections were cut in a vibratome and collected in PB 0.1 N. Sections were incubated in permeabilization solution (PB 0.1N, Triton X-100, NHS 10%), followed by 24/48 h incubation at 4°C with primary antibody (1:500) in blocking solution (PB 0.1N, Triton X-100, NHS 10%). After washing three times in PB, Alexa-coupled goat/rabbit polyclonal secondary antibodies (1:1,000, Molecular Probes, USA) were used.

Finally, a 1:1,000 dilution in PB of Hoechst 33342 was added for 5 min. Slices were rinsed several times in PB, mounted with gerbatol mounting medium, and allowed to dry. The omission of the primary antibody was used as a control. Confocal analysis was performed in a Leica (Germany) microscope. For double-stained ChAT/c-fos counting, five sections per animal were scored using Matlab and Imaris software. The antibodies used in this study include rabbit polyclonal c-Fos (Abcam, UK, ab190289), rabbit polyclonal IGF-I Receptor-β (Santa Cruz, USA, 713/AC), and rabbit anti-IGF-I receptor β XP (Cell Signaling Technology, USA, 9750), anti-Choline acetyltransferase polyclonal antibody (Merck-Millipore, USA, AB144P), and anti-pAkt (Cell Signaling Technology, USA, 9271).

In some experiments, mice were processed for immunocytochemistry 2 h after ip IGF-I injection to allow the expression of c-fos (see above). IGF-I (1 mg/ml) was first dissolved in acetic acid 1N and then prepared with saline to a final dose of 1 μg/g body weight.

Images (coordinates from Bregma: 0.5 mm anterior 0.8 mm lateral, 5.5 mm–6 mm depth from the cortical surface) were taken from the HDB in a confocal microscope (Leica, Germany). The analysis was carried out using the Imaris 9.4 software (the thickness of the images taken was 20 μm). In the ChAT/c-fos analysis, we used 20× magnification; images were taken from the HDB area. Spots of this area were selected with an estimated XY size of 20 μm to count ChAT+ neurons in the red channel. Then, spots were filtered with the c-fos channel in green to obtain the amount of spots ChAT+/c-fos+.

In the ChAT/IGF-1R analysis, we used 40× magnification to measure IGF-IR+ / ChAT+ cells by the signal intensity (%) after identifying the ChAT neurons in the red IGF-1R in the green channel. First, we create a 3D surface (taking the entire thickness) in the red channel using the “surface mode.” In this way, a threshold was set (it was the same for all pictures) to remove the background, and all ChAT neurons above it were selected. Then, we obtained a total number of neurons in the desired area. From the total number of ChAT neurons obtained, we select only those which coincide with the green channel’s maximum intensity, which is the corresponding color of the IGF-1R secondary antibody. To do that, we add a selection filter called “maximum intensity selection in the channel green.” In this way, we were able to distinguish only those neurons that colocalize with the green channel maximum intensity, using an automatic threshold (corresponding to IGF-1R). Finally, we calculated the relationship between ChAT neurons colocalizing with the maximum intensity of IGF-1R.

ECoG segments of 1 min every 5 min were analyzed by Spike 2 software, using the Fast Fourier Transform algorithm to obtain the power spectra. The mean power density was calculated for five different frequency bands that constitute the global ECoG: delta (0.3–4 Hz), theta (4–8 Hz), alpha (8–12 Hz), beta (12–30 Hz), and gamma (>30 Hz) bands. The total power of the five frequency bands was considered 100%, and the percentage of each frequency band was calculated. Gamma frequencies were excluded from the results because their power was negligible in anesthetized mice.

Somatosensory evoked potentials (SEPs) were elicited in the S1 cortex by whisker deflections. The area under the curve of the first negative wave was calculated in each case. A control period of 10 min was recorded, and the mean SEP area was considered 100%. Recordings were also performed every 5 min for 30 min after local IGF-I injection in the HDB nucleus. The mean firing rate of MUA recordings was also calculated before (10 min) and after drug injections (every 5 min for 30 min). ChAT+ neurons were identified and included in this study when blue-light pulses induced an increase in their firing rate of at least 10%.

Statistical analysis was performed using Graph Pad Prism 6 software (San Diego, CA, USA). Depending on the number of independent variables, normally distributed data (Kolmogorov–Smirnov normality test), and the experimental groups compared, we used either Student’s t-test or two-way ANOVA followed by Sidak’s multiple comparison test. For non-normally-distributed data, we used the Mann–Whitney U test to compare two groups, Kruskall-Wallis or Friedman test, with Dunn’s multiple comparisons and a Post Hoc analysis such as Scheirer–Ray Test, a non-parametric alternative to multi-factorial ANOVA. The sample size for each experiment was chosen based on previous experience and aimed to detect at least a p < 0.05 in the different tests applied, considering a reduced use of animals. Results are shown as mean ± standard error (SEM) and p values coded as follows: *p < 0.05, **p < 0.01, ***p < 0.001.

Immunohistochemical studies were focused on the HDB area because it projects abundantly to the S1 cortex (e.g., Chaves-Coira et al., 2016, 2018; Figure 1A). Immunohistochemistry of the IGF-I receptor (IGF-IR) showed abundant staining in ChAT + cells in the HDB nucleus of young and old mice. The co-localization of IGF-IR signal to ChAT+ identified neurons strongly suggested that they expressed IGF-IR (Figure 1B). The expression of IGF-IR was also observed in the neuropil and in non-identified cells of HDB. We compared the number of ChAT+ cells per area in young and old mice. There was a slight increase of ChAT+ cells in old animals but differences were not statistically significant (n = 3 mice per group, p > 0.05; Unpaired t-test; Figure 1C). However, we found that the mean expression of IGF-IR in ChAT+ cells of old mice was reduced (p = 0.0419; Unpaired t-test; Figure 1D), suggesting a reduction of sensitivity to IGF-I in ChAT+ cells of old animals.

Then, we determined the expression of the immediate early gene protein c-fos to test if IGF-I activates HDB neurons. We studied double-labeled ChAT+/c-fos+ cells in young and old animal groups after saline or IGF-I i.p. injections (Figures 2A–D). The plot of the percentage of ChAT+/c-fos+ cells showed a significant increase in old-saline over young-saline animals (p = 0.0087; Two-Way ANOVA, and Fisher’s LSD comparison test; Figure 2E), suggesting a basal increase in neuronal activity in old animals. Furthermore, IGF-I injection increased the expression of c-fos in young animals (p = 0.045; Two-Way ANOVA, and Fisher’s LSD comparison test). However, IGF-I reduced the number of ChAT+/c-fos+ cells in old animals although differences did not reach statistical significance (Figure 2E).

The above results suggest a reduced response of HDB neurons to IGF-I in old animals. Consequently, we studied HDB neuronal response to local injection of IGF-I (10 nM; 0.2 μl) through MUA recordings in young and old mice (Figure 3A). We used local injections to avoid a possible reduction of IGF-I entry in old mouse brains through the endothelium. At basal conditions, the spontaneous activity was similar in young and old mice (22.1 ± 6.4 spikes/s and 23.8 ± 4.6 spikes/s, respectively; p > 0.05, Unpaired t-test; Figure 3B). Local IGF-I injection induced an immediate increase of their firing rate in HDB neurons of young mice (Figure 3C). However, the response to IGF-I was significantly lower in old animals. Five minutes after IGF-I injection, young mice showed an increase of the firing rate of 61.4%, while old animals displayed a reduction of 7.4%. Spontaneous activity continued to increase for at least 30 min (278.1% vs. 48.6% in young and old animals, respectively; young = 14, old = 6; p = 0.0044; Ordinary Two-way ANOVA; Figure 3C).

Since cortical projections of HDB neurons induce cortical activation (Duque et al., 2000; Zaborszky et al., 2015; Chaves-Coira et al., 2018), we tested if the injection of IGF-I in HDB could induce changes in the ECoG recorded in the S1 area of young and old anesthetized mice. We studied the IGF-I effect on the ECoG analyzing its power spectrum and the proportion of the frequency bands. In basal conditions, delta waves were the predominant frequency band due to anesthesia. In these conditions, there were no differences between young and old mice (n = 9 per group; F_(4,32) = 0.04674; p = 0.9957, Two-way RM ANOVA; Figure 4A). However, local IGF-I injection in the HDB nucleus (10 nM; 0.2 μl) increased fast ECoG activities in young but not in old mice (Figures 3B–D). θ waves gradually increased their power after IGF-I injection in young animals while old animals only showed a rapid increase at 5 min after injection and returned to basal values, showing a statistically significant difference throughout the 30 min of recording (young = 13, old = 9; F_(1,138) = 6,436; p = 0.0123; Ordinary Two-way ANOVA; Figure 4B). Similarly, α waves increased their power after IGF-I application in young but not in old animals (young = 13, old = 9; F_(1,138) = 11,85; ***p = 0.0008; Ordinary Two-way ANOVA; Figure 4C). β waves also slightly increased after IGF-I injection in young animals, but differences in this frequency band were small, probably due to anesthesia, and did not reach statistical significance (young = 13, old = 8; F_(1,145) = 1,833; p = 0.1779; Ordinary Two-way ANOVA; Figures 4D,E).

Different studies have demonstrated that BF neurons facilitate cortical sensory responses by activating muscarinic receptors (Golmayo et al., 2003; Fournier et al., 2004; Chaves-Coira et al., 2016, 2018). Here, we want to test if local IGF-I injection in the HDB nucleus also facilitates cortical tactile responses. We analyzed the SEP induced in the S1 cortex by whisker deflections. The SEP area was calculated before and after local injection of IGF-I (10 nM; 0.2 μl) in the HDB nucleus (Figure 5A). We did not find differences in the SEP area between young and old mice at basal condition (young = 14; 0.1520 ± 0.0379 mV²; old = 13; 0.1402 ± 0.060 mV²; p = 0.8682, Unpaired t-test; Figure 5B). However, IGF-I injection in the HDB nucleus induced fast facilitation of tactile S1 responses in young mice that increased their response up to 41% over basal response at 5 min after IGF-injection, while IGF-I did not modify tactile responses in old mice at that time (young = 15, old = 13; p = 0.0176, Unpaired t-test; Figures 5C,E). The time course of tactile responses showed a rapid increase in young mice. Nevertheless, tactile responses slowly increased in old mice, reaching a percentage similar to young animals at 15–25 min after IGF-I application (young = 15, old = 13; F_(1,177) = 5.886, p = 0.0163; Ordinary Two-way ANOVA; Figure 5D).

It is well known that cholinergic projections from the BF are responsible for the cortical activation and facilitation of sensory responses (see above references). Consequently, we determine the specific effect of IGF-I on ChAT cells using an optogenetic approach. We used mice expressing channelrhodopsin (ChR) specifically in cholinergic neurons (ChAT-ChR2-YFP animals; see “Materials and Methods” section). We used optogenetic stimulation for selective activation of cholinergic neurons in the HDB nucleus through an optrode that performed unit recordings and optical stimulation simultaneously in the same place. Long-lasting blue-light pulses (300 ms duration) were applied, which stimulated a small volume of tissue (about 200 μm in radius; Figure 6A). Unit activity was measured at 100 ms time intervals after the light onset (0–100 ms, 100–200 ms, and 200–300 ms intervals) and compared with the previous 100 ms time interval before light onset (basal condition). Neurons were considered cholinergic (ChAT neurons) when blue-light pulses induced an increase of at least 10% of their unit activity.

Blue-light pulses induced similar responses in ChAT neurons of either young or old mice (Figures 6B,C, respectively). In young animals, blue-light pulses increased 33.3% the firing rate over the basal level at 100 ms after light onset (p = 0.0001), and 22.2% at 200 ms over the basal level (F_{(2.040,42.85)} = 9.216; p < 0.0004; n = 26 neurons; One-way Repeated Measure ANOVA; at basal vs. 0–100 ms: p < 0.0001, or 100–200 ms: p = 0.0078, Dunnett’s Multiple Comparison test; Figure 6B). Old animals also displayed a blue-light response, which increased at 100 ms (25.4%; p = 0.0010) and 200 ms (14.9%; p = 0.0277) over basal levels (F_{(1.908,40.06)} = 6.702, p = 0.0035, n = 24; One-way Repeated Measure ANOVA; p < 0.0010 at 0–100 ms vs. basal activity, p < 0.0277 at 100–200 ms; n = 24 neurons; Dunnett’s Multiple Comparison test; Figure 6C). The increment of the firing rate evoked by the blue-light pulse returned to basal levels at 300 ms after light onset in both young and old animals.

After neuronal identification, we studied the effect of local application of IGF-I (10 nM; 0.2 μl) in the HDB nucleus on those neurons. The response pattern of ChAT neurons to IGF-I was substantially different in both animal groups. IGF-I induced an immediate 95.7% increase in the firing rate in young animals at 5 min, whereas only a 4.5% increase was seen in old animals (n = 13 neurons per group; young = 19.57 ± 2.63 spikes/s; old = 10.45 ± 2.28 spikes/s; p = 0.0151; Unpaired t-test; Figure 6D). The time course showed a rapid increase of the ChAT firing rate in young animals while the firing rate increased slowly in old animals (n = 13 neurons per group; p = 0.0163; Ordinary Two-way ANOVA; Figure 6E).

There is some evidence that IGF-I may interact with muscarinic receptors in various brain systems (Batty et al., 2004; Granja et al., 2019). Thus, we explore if the muscarinic cholinergic receptor antagonist atropine could prevent IGF-I effects in young animals. We firstly identified ChAT cells through optogenetic stimulation, as above. Blue-light pulses increased their firing rate in 16 recorded neurons at 100 ms (53.3%; p < 0.001) as well as at 200 ms (34.3%; p = 0.0345; Figure 7A). After that, we compared the IGF-I effect on ChAT neurons in control conditions (saline solution i.p.; n = 7 neurons) or when atropine was i.p injected 15 min before the local injection of IGF-I in the HDB nucleus (n = 9 neurons). In both cases, IGF-I induced a fast increase in the firing rate. However, the long-lasting effect evoked by IGF-I was blocked (saline/IGF-I = 7 neurons, atropine/IGF-I = 9 neurons, F_(1,151) = 4.878, p = 0.0287; Two-Way ANOVA; Figure 7B). This local effect of atropine on the activity of ChAT neurons may be due to the existence of local collaterals within the BF, as has been observed previously (Zaborszky and Duque, 2000; Zaborszky, 2002). Thus, atropine may affect ChAT neurons, reducing spontaneous activity in the HDB as well as the response to IGF-I local application.

In addition, atropine also blocked the SEP facilitation evoked by IGF-I (Figure 7C). IGF-I increased SEP area in control animals (saline-injected animals), while tactile facilitation was blocked in atropine-injected animals (saline/IGF-I = 13 animals, atropine/IGF-I = 9 animals; F_(1,160) = 26.38, p < 0.0001; Ordinary Two-Way ANOVA).

The above findings showed a reduction of IGF-I effects in old animals. To study if a chronic IGF-I application may restore these effects, we administered IGF-I or saline solution for 28 days through Alzet^® mini-pumps in old animals. We compared the spontaneous activity of ChAT+ cells in young and old mice and with ChAT+ cells recorded in IGF-I-treated old mice. Although no differences were observed between the spontaneous firing rate of young and old mice according to the MUA recordings (see Figure 3B), old mice had a higher firing rate (11.19 ± 0.89 spikes/s n = 10) than young mice when ChAT+ neurons were specifically recorded (7.12 ± 0.71 spikes/s; n = 14 neurons p = 0.0119; Figure 8A). Furthermore, ChAT+ neurons in IGF-I-treated old animals showed a greater spontaneous firing rate (21.17 ± 1.19 spikes/s; n = 16 neurons) than old untreated (p = 0.0001) or young mice (p = 0.0001; One-Way ANOVA, p < 0.0001, Holm–Sidak’s multiple comparisons test).

Local IGF-I injection in the HDB area (10 nM; 0.2 μl) increased the firing rate in both untreated and IGF-I-treated-old mice. The effect of IGF-I was faster in treated animals; however, differences did not reach statistical significance (n = 15 neurons, F_(1,182) = 0.6303; p = 0.4283; Ordinary Two-Way ANOVA; Figure 8B). The IGF-I-facilitatory effect on tactile responses was also faster in IGF-I-treated old mice than in old untreated animals (old = 13 neurons, old treated = 15 neurons, F_(1,180) = 5.868, p = 0.0164, Ordinary Two-Way ANOVA; Figure 8C). In addition, IGF-I-treated old mice show greater fast oscillatory ECoG than untreated old mice after local IGF-I injection in the HDB nucleus (Figures 8D–F). The θ, α, and β frequency bands increased; power of θ band (old = 13, old treated = 15; F_(1,180) = 8.075; p = 0.005; Figure 8D), α band (old = 13, old treated = 15; F_(1,166) = 6,790; p = 0.001; Figure 8E), and β band (young = 15, old = 13; F_(1,182) = 14.08; p < 0.0002, Ordinary Two-Way ANOVA; Figure 8F) over 30 min recording. In conclusion, chronic IGF-I treatment improves basal cholinergic activity and overall cortical activity in old mice.

Finally, we wanted to know whether reduced IGF-I action on ChAT neurons of old animals is a general phenomenon. We tested the effect of orexin in the HDB nucleus because orexin neurons contribute to ECoG activation in the awake state (Sakurai and Sasaki, 1998; de Lecea et al., 1998; Zegarra-Valdivia et al., 2020). Orexinergic receptors are expressed in the BF region and can activate cholinergic and non-cholinergic neurons (Marcus et al., 2001; Arrigoni et al., 2010; Villano et al., 2017). In agreement with that, we observed orexin-A receptors in ChAT+ neurons (Supplementary Figure 1A) and non-identified neurons of the HDB nucleus. Recordings in ChAT neurons of the HDB nucleus of young and old animals (Supplementary Figure 1B) showed an increase in the firing rate in both animal groups after local orexin-A injection (10 nM; 0.2 μl), indicating that orexin responses were not affected by aging (young = 12, old = 8, p = 0.6680, Supplementary Figure 1C).