Sexual dimorphism in serum leptin levels in ALS patients and SOD1G93A mouse model
Circulating levels of leptin in ALS patients have been somehow controversial, but there are studies that point out that the huge variation found might be partly due to the sex differences in ALS patients19. In order to elucidate whether leptin levels in ALS are associated with sex, we analyzed the serum of ALS patients and age-sex-matching healthy controls. As previously reported, the levels of leptin differ between men and women in ALS (Fig. 1A) with lower levels of leptin in the blood of men ALS patients and no significant alterations in the women suffering ALS (Fig. 1C). Thus, it supported the idea that the treatment strategy of lowering the levels of leptin in ALS could work in the women but not in the men. Coincidently, previous studies measuring circulating levels of leptin found sexual dimorphism in SOD1G93A mice, with normal levels of leptin in females and much lower levels in males compared to wild-type littermates, at early (90 days, P90) and late (120 days, P120) disease stages23. To corroborate this finding, we measured circulating leptin levels by enzyme-linked immunosorbent assay (ELISA) in the blood of females at P90. Female SOD1G93A mice had double the levels of leptin than their wild-type (Fig. 1E).
Since leptin is mainly produced in the subcutaneous white adipose tissue, the circulating levels are normally proportional to the amount of subcutaneous fat depots. We checked whether the levels of leptin in blood were also correlated with these fat depots in ALS patients, and in SOD1G93A mice, despite weighing less than their controls. We used the skinfold thickness measurement as a noninvasive method of body fat estimation in ALS patients. Men suffering from ALS had less amount of subcutaneous fat tissue and this reduction was not similarly observed in women with ALS. As expected, leptin levels mirrored the subcutaneous fat tissue (Fig. 1B, D). Inguinal white adipose depots (iWAT) from SOD1G93A and WT littermate female mice at 90 days of age were dissected and weighed (Fig. 1F). Despite of having more body weight, the iWAT depot of SOD1G93A female mice weighed more than those of their control WT littermates, which correlated with the higher levels of leptin in the blood, even when corrected by body weight (data not shown). These results showed that the reduction in body weight, at least in the first stages of disease in the female SOD1G93A mice, was not totally due to a major loss of adipose tissue, which explained the higher levels of circulating leptin in females SOD1G93A mice.
Leptin haploinsufficiency restores normal levels of circulating leptin and maintains the weight in SOD1 G93A females mice
A genetic strategy to lower leptin levels has been previously used in SOD1G93A mice, aiming at ameliorating the disease burden and progression by reducing the systemic hypermetabolism and preserving the body weight23. They found that lowering the systemic levels of leptin had beneficial effects in females but not in males SOD1G93A. Thus, we aimed to identify the pathways that might be operating to exert beneficial effects on the disease of SOD1G93A - Lepob/+ females, as they could be interesting therapeutic targets. We selected two tissues in which studying those potential pathways: i/ the lumbar spinal cord, to evaluate the direct effect of the leptin haploinsufficiency on the main primary affected tissue in the SOD1G93A mice, and ii/ in the iWAT, as the main producer and receptor tissue of leptin in the body responding to systemic metabolism.
We replicated the previously published strategy23 and generated SOD1G93A female mice with leptin haploinsufficiency (SOD1G93A - Lepob/+), by crossing female Lepob/+ with males SOD1G93A (Fig. 2A). We first validated the effectiveness of the leptin-haploinsufficiency background in our mice by measuring serum leptin levels in the four different groups of the study, all females at 90 days of age (P90), before the onset of weight loss. As expected, the amount of leptin in blood was given by their genetics, with reduced levels in the Lepob/+ mice compared to their Lep+/+ littermates (Fig. 2B). We next weighed the dissected subcutaneous iWAT depots in these mice and found that the iWAT depots were not very different between the groups (Fig. 2C), although there were some reductions in the double mutant compared to the SOD1G93A mice. These results suggest that reducing the circulating levels of leptin had the expected effect in wild-type mice, increasing the fat depots, but the effect seems less evident in the fat depots of SOD1G93A - Lepob/+ mice. Next, we measured the effect of leptin deficiency in the body weight of the mice along the disease progression. The deficiency of leptin seems to maintain the weight of the SOD1G93A female mice, similar to those of the wild-type littermates, especially at early stages of the progression of the disease (Fig. 2D). These results evidenced that the iWAT depot weights in the SOD1G93A female mice were not proportional to their body weights, supporting previously reported alterations in the fat tissues of ALS patients, and how the observed weight loss at early disease stages are not necessarily explained by a reduction in the fat tissue in the body4.
Leptin deficiency had a major impact in the iWAT transcriptome in SOD1 G93A mice.
The iWAT regulation looked altered in SOD1G93A female mice, and the leptin deficiency seemed to be able to preserve or maintain their body and iWAT weights. Thus, we aimed to identify the genes and pathways that are operating in the iWAT and spinal cord of the SOD1G93A in comparison to the SOD1G93A-Lepob/+ mice, in order to understand both the pathological mechanism and how those might change by the leptin deficiency on each tissue.
First, we run a transcriptomic analysis of the iWAT of the four groups of interest female mice (n = 4) at early symptomatic disease stages (P90) in order to identified differentially expressed genes (DEGs) with a false discovery rate (FDR) < 0.05. The transcriptional changes that the leptin haploinsufficiency induced directly on the iWAT tissue, by comparing the iWAT transcriptome of Lepob/+ mice to the WT mice, showed no significant alterations (Fig. 3A, B), even though the Lepob/+ mice, and marginally their iWAT, weighed more than their wild-type littermates (Fig. 2C, D). Only five DEGs were identified comparing the iWAT of the Lepob/+ to WT mice. Of those, four were downregulated (Aqp5, Itagv, Gm128, Gm4613) and one upregulated (Ddit4).
Next, we evaluated the transcriptional alterations in the iWAT of SOD1G93A mice, compared to WT, and identified 14 DEGs (11 up and 3 down) with threshold FDR 0.05 (Fig. 3C) and 710 DEGs with p value < 0.05 (Supplementary Fig. 1A). In order to identify the pathways altered by SOD1 mutation in the iWAT, we run pathway analysis by ORA and GSEA. The analysis of biological processes enriched by the ORA technique with DEGs showed that 9 of the top 10 deregulated processes are related to RNA splicing (Supplementary Fig. 1B). Furthermore, the pathway analysis performed by the GSEA technique, showed an upregulation of several processes related to the immune system: "immunoglobulin production”, “lymphocyte differentiation”, "production of molecular mediator of immune response", “T-cell activation” (Fig. 3D). These data suggest a mild activation of the immune response in the iWAT of SOD1G93A mice even at early disease stages, which could partly explain why these mice had more iWAT than their wild-type littermates, despite weighing less.
Lastly, we evaluated the transcriptional effect of lowering the levels of leptin in the iWAT of SOD1G93A mice. The comparison between the transcriptome of SOD1G93A and SOD1G93A-Lepob/+ mice identified 1793 DEGs with FDR < 0.05, with a clear predominance of inhibited genes (398 up and 1394 down) (Fig. 3E) and 5000 DEGs with p value < 0.05 (Supplementary Fig. 1C). The ORA analysis showed that the top 10 dysregulated pathways are processes related to the immune system "T-cell activation", "adaptive immune response", and "lymphocyte differentiation" (Supplementary Fig. 1D). The GSEA analysis showed these same processes to be the most dysregulated and determined that they were inhibited (Fig. 3F). Up to 483 of the 1793 identified DEGs in the iWAT of SOD1G93A-Lepob/+ mice were associated with the GO term immune system process (GO:0002376), and up to 204 of them were involved in lymphocyte activation (GO:0046649). Most of these genes were strongly inhibited, with up to 61 genes having a log2 fold change < -5.
Leptin deficiency countered the lymphocyte activation processes in the iWAT of SOD1 G93A mice.
We next looked deeper into the specific transcriptional activation of the immune response caused by the SOD1G93A transgene and whether those might be countered by the leptin deficiency, iWAT of SOD1G93A mice. We selected commons immune processes deregulated in SOD1G93A mice and SOD1G93A-Lepob/+ in the iWAT (“immunoglobulin production”, “lymphocyte differentiation” and “T-cell activation”) and plotted the distribution of the fold change of the genes involved in these pathways (Fig. 4A). The enrichment score plot indicates that most of these pathways were activated in the iWAT of SOD1G93A mice, and the opposite inhibition was observed when reducing the levels of leptin in the context of SOD1G93A (SOD1G93A-Lepob/+) (Fig. 4B).
Among all the immune system processes found, the “T-cell activation” process was the most notable. Thus, we identified the highly altered genes involved on these processes and looked for their profile distribution, represented in a hierarchical clustering (Fig. 4C). The analysis evidenced that the effect of the two mutations combined had a stronger effect on those genes analyzed than any of the mutations alone (SOD1G93A vs WT, or Lepob/+ vs WT). Leptin deficiency alone (Lepob/+ vs WT) had a mild but evident inhibition of the genes involved in the T cell activation pathway in the iWAT. The SOD1G93A transgene induced an upregulation (shown in red colour) of several of those genes. The combination of the two mutations had a stronger inhibition of the genes related to lymphocyte T activation pathway in the iWAT (SOD1G93A vs SOD1G93A- Lepob/+). Among the genes identified were chemokines (Ccl19, Ccl21a, Ccl21b, Ccr6, Ccr7, Ccr9), surface antigens (Cd3d, Cd3e, Cd3g, Cd5, Cd6, Cd8a, Cd40lg), histocompatibility antigen (H2-Eb2, H2-M2), interleukins (Il12a, Il12b, Il4i1b, Il7r), immunoglobulins (Btla, Ctla4, Ighd, Ighg1, Igkj5) and tumor necrosis factor (Tnfrsf13C, Tnfrsf4).
The transcriptional effect of SOD1 G93A mutation is higher in the spinal cord than in the iWAT.
Since the spinal cord is the tissue primarily affected by the SOD1G93A mutation, we evaluated if the systemic effect of lowering leptin levels would have a direct impact on the spinal cord. Thus, in parallel, we run a transcriptomic analysis of the spinal cord on the same mice as in the transcriptomic analysis of the iWAT. The expression of the leptin receptor in the spinal cord is considerably low (https://www.proteinatlas.org/ENSG00000116678-LEPR/tissue). We corroborated the low expression of leptin receptor in the SPC, specially compared to the iWAT, and checked that the expression was not modified by any of the genetic mutations (Lepob/+ vs WT: log2FC = 0.13, FDR = 0.99; SOD1G93A vs WT: log2FC = 0.32, FDR = 0.99). These results were consistent with previous detection of the leptin receptor in the spinal cord by immunostaining23. They found that the leptin receptor was only present in the glial cells of the ventral horn of the spinal cord, and none in the alpha motor neurons, concluding that any of the positive effect of lowering leptin levels in the spinal cord should be indirect and not directly on the motor neurons.
As with the iWAT, we first evaluated the effect of the leptin deficiency alone in the transcriptome of the lumbar spinal cord (SPC) (Lepob/+ vs WT) and found very mild effects, with 21 DEGs identified (16 up and 5 down) at FDR < 0.05 (Fig. 5A). The GSEA pathway analysis showed a down regulation of processes related to metabolism, such as “fatty acid metabolic process", “monocarboxylic acid metabolic process” and “cholesterol metabolic process” (Fig. 5B). The ORA analysis revealed other altered processes, mainly in the PI3K signaling pathway. Interestingly, the "negative regulation of cytokine production" pathway was among the top 10 GO terms altered, which is consistent with the anti-inflammatory effect of leptin deficiency (Supplementary Fig. 2A).
The transgene SOD1G93A had a very strong effect on the SPC gene expression profile, as we previously shown (from Fernández-Beltrán LC et al. 2021). We identified 1174 DEGs (639 up and 535 down) (Fig. 5C, adapted from Fernández-Beltrán LC et al. 2021). Similar to the iWAT transcriptome of SOD1G93A mice, the SOD1G93A transgene induced more upregulation of genes than inhibition. In the GSEA analysis we also found, among the top dysregulated biological pathways, several pathways related to an activation of immune system, such as: “positive regulation of defense response”, “positive regulation of cytokine production”, “inflammatory response” (Fig. 5D). On other hand, top downregulated processes were related to metabolic processes such as: fatty acid, steroids, and alpha amino acids. Similar results were found with the ORA analysis in relation to immune system activation: leukocyte activation, tumor necrosis factor, and phagocytosis (Supplementary Fig. 2B).
In the spinal cord of the SOD1G93A the effect of lowering leptin levels was very mild. Compared to the SOD1G93A alone, the transcriptome of the SPC of SOD1G93A -Lepob/+ identified 18 DEGs with FDR > 0.05 (9 up and 9 down) (Fig. 5E). These results were expected since the effect of lowering the leptin levels might be mostly indirect on the spinal cord, and from systemic metabolic changes. Looking at the most relevant biological processes and pathways altered by the combination of the two mutations (SOD1G93A-Lepob/+), the GSEA analysis identified an activation of transcription-translation processes (“Demethylation”, “ribonucleoprotein complex assembly” and “Cytoplasmic translational initiation”) and metabolic processes (“Fatty acid metabolic process” and “Monocarboxylic acid metabolism process”) (Fig. 5F). These metabolic processes were downregulated in both Lepob/+ and SOD1G93A mice when compared with WT.
Leptin deficiency had a mild effect on maintaining the regulation of fatty acid metabolism and no effect on inflamation in the spinal cord of SOD1 G93A mice.
Lowering the levels of leptin had minimum effect on the transcriptional inflammatory response in the SPC in the SOD1G93A background (comparing SOD1G93A - Lepob/+ vs SOD1G93A) (Supplementary Fig. 3). Interestingly, other metabolic pathways seemed to be changed by the approach. The pathway analyses in the SPC of the different groups also revealed that metabolism processes were dysregulated not only by SOD1G93A but also when combined with leptin deficiency (Fig. 5B, D, F). The “fatty acid metabolism” (GO: 0006631) was the common metabolic process dysregulated in SOD1G93A mice and in SOD1G93A-Lepob/+ in the SPC. We identify the genes altered on this process and plotted them in a distribution based on their up or down regulation by fold change (FC). The genes involved in fatty acid metabolism tend to be inhibited in the SPC of SOD1G93A mice (Fig. 6A), while leptin deficiency when placed in SOD1G93A background, showed increased expression of these same fatty acid genes (Fig. 6B). Finally, we performed a hierarchical clustering analysis on those genes involved in fatty acid metabolism comparing the expression profile among the four different groups. Leptin deficiency in the spinal cord of SOD1G93A mice corrected the inhibition of some of the genes involved in fatty acid metabolism (Fig. 6C).
These results support the concept that lowering systemic leptin levels affected the transcriptional profile of the spinal cord, although milder than in the iWAT, and through different pathways. These differences between the tissues could be reflecting the fact that the effect of leptin haploinsufficiency might be direct in the iWAT and seemed indirectly in the SPC.