Differential responses to UCP1 ablation in classical brown versus beige fat, despite a parallel increase in sympathetic innervation

Biochemical and morphological characteristics of IBAT and ingWAT in wild-type and UCP1-KO mice

Biochemical and morphological characteristics of IBAT and ingWAT from wild-type and UCP1-KO mice acclimated to three different temperatures are presented in Figure 1. In wild-type mice (filled symbols), the weights of both IBAT and ingWAT steadily decreased with acclimation to low environmental temperatures (Fig. 1, A and B) (detailed statistical analysis applying to all Figures is presented in Table S1). Total protein content in each of the tissues increased with acclimation to 18 °C and further increased with acclimation to 4 °C (Fig. 1, C and D, filled symbols). This increase in total protein content was accompanied by an increased protein density in the tissues (Fig. 1, E and F, filled symbols) (in adipose tissue, protein density and adiposity are inversely related). Together, these findings suggest cold-induced thermogenic recruitment in both tissues of wild-type mice.

Open in a separate window

Figure 1

The effect of UCP1 ablation on biochemical and morphological parameters of IBAT and ingWAT. Wild-type (filled symbols) (n = 4–6) and UCP1-KO mice (open symbols) (n = 5–6) were acclimated to the indicated temperatures (as detailed in Experimental procedures). A, IBAT wet weight, (B) ingWAT wet weight, (C) IBAT protein content, (D) ingWAT protein content, (E) IBAT protein density and (F) ingWAT protein density. Each symbol represents one mouse sample. Where not visible, the error bars are smaller than the symbols. Values are means ± SEM. Data were analyzed using two-way ANOVA. (Full ANOVA statistics are given in Table S1). ∗Significant difference between wild-type and UCP1-KO mice for each tissue and temperature using the two-way ANOVA followed by Tukey’s multiple comparison test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. G–R, Histological appearance of IBAT (G, H, K, L, O and P) and ingWAT (I, J, M, N, Q and R) from the animals in (A–F). The tissues were analyzed using hematoxylin and eosin staining, and representative images are presented. Scale bar 100 μm. UCP1-KO, UCP1-knockout.

In UCP1-KO mice (open symbols), general IBAT parameters were relatively similar to those in wild-type mice. However, at 4 °C, the protein density in IBAT from UCP1-KO mice was significantly lower than in the tissue from wild-type mice (Fig. 1E, blue symbols), suggesting increased adiposity of brown fat deficient in UCP1.

ingWAT parameters in wild-type and UCP1-KO mice displayed larger differences with acclimation to low environmental temperatures (Fig. 1, B, D and F). At 4 °C, ingWAT wet weight in UCP1-KO mice was significantly larger (Fig. 1B, blue symbols), and this was accompanied by a marked increase in tissue protein content (Fig. 1D, blue symbols) and protein density (Fig. 1F, blue symbols). This higher wet weight was therefore not a consequence of an increased lipid content in the tissue but was rather an indication of a greater tissue recruitment.

To verify the consequences of UCP1 ablation on IBAT versus ingWAT morphology, the tissues were stained with hematoxylin and eosin. Examples are shown in Figure 1, G–R and are here only qualitatively described as the morphology is principally in agreement with earlier observations as detailed. At thermoneutrality, the majority of brown adipocytes show lipids accumulated in a single large unilocular lipid droplet, as seen earlier (e.g. (14, 15, 16)); no difference was observed between wild-type and UCP1-KO mice (Fig. 1, G and H). The brown adipocytes in wild-type mice acclimated to subthermoneutral temperatures had typical multilocular lipid droplet morphology (Fig. 1, K and O). The majority of brown adipocytes in UCP1-KO mice acclimated to 18 °C and to 4 °C were also multilocular but a number of adipocytes contained only a single large lipid droplet (as shown in (4)) (Fig. 1, L and P). The ingWAT of mice acclimated to thermoneutrality was composed of only unilocular adipocytes (in agreement with e.g. (14, 15, 16)); no difference was observed between wild-type and UCP1-KO mice (Fig. 1, I and J). In wild-type mice acclimated to 18 °C, paucilocular adipocytes emerged only sporadically (Fig. 1M). Paucilocular/multilocular adipocytes in UCP1-KO mice appeared to be more prevalent, appearing in a few islands surrounded by unilocular adipocytes (Fig. 1N). In mice acclimated to stronger cold (4 °C), multilocular adipocytes became even more conspicuous, forming numerous islands surrounded by unilocular adipocytes. Notably, the islands in ingWAT from UCP1-KO mice (Fig. 1R) appeared to be larger compared to those in ingWAT from wild-type mice (Fig. 1Q) (in agreement with e.g. (4, 6, 9)).

These observations at both a histological and a biochemical level essentially confirm previously reported data, such as increased adiposity in brown fat and significantly higher protein content as well as lower adiposity in beige fat from the UCP1-KO mice acclimated to the cold. They thus indicate an apparently regular recruitment of classical brown fat and an enhanced recruitment of brite/beige fat in UCP1-KO mice exposed to cold. An important question relating to this is the nature of proteins that underlie these recruitment processes.

In the cold, the ablation of UCP1 triggers macrophage accumulation in both brown and beige adipose depots

Earlier studies demonstrated that exposure of UCP1-KO mice to subthermoneutral temperatures promotes brown-fat inflammation characterized by increased levels of proteins involved in host defense, macrophage infiltration, and formation of crown-like structures (3, 7). Macrophage recruitment to ingWAT of UCP1-KO mice has not been studied previously but could provide important insights into the mechanisms underlying different responses to cold in IBAT versus ingWAT from these mice. We therefore also examined macrophage recruitment to the beige fat.

The morphology, frequency, and distribution of macrophages in the tissues were first examined qualitatively using immunohistochemistry (Fig. 2, A–L). Macrophages were visualized with an anti-MAC-2 antibody (green); MAC-2, also known as galectin-3 or LGALS3, is a lectin that mediates macrophage phagocytic and inflammatory responses (17, 18). The adipocytes were visualized by staining the lipid-droplet protein perilipin (red). At thermoneutrality, macrophages in brown fat were sporadically found and organized into crown-like structures (Fig. 2, A and B); no difference was observed between the tissues from wild-type (Fig. 2A) and UCP1-KO mice (Fig. 2B). In the brown fat from wild-type mice acclimated to 18 °C and to 4 °C, macrophages could not be visually observed (Fig. 2, E and I). This result is in full agreement with our earlier report demonstrating the presence of macrophages in brown fat from wild-type mice acclimated to thermoneutral temperature and no increase or even downregulation in their abundance upon exposure to cold (19, 20). However, importantly, in clear contrast to the case in the wild-type mice, in the UCP1-KO mice, acclimation to subthermoneutral temperatures resulted in a massive macrophage accumulation into brown fat (Fig. 2, F and J) (in agreement with ((3, 7))). The majority of these macrophages were aggregated into crown-like structures (Fig. 2, F and J), while a few macrophages were scattered within the tissue as solitary macrophages (for details see enlarged version of Fig. 2J (Fig. S1)).

Open in a separate window

Figure 2

A notable accumulation of macrophages in IBAT and ingWAT of UCP1-KO mice housed below thermoneutrality.A–L, representative confocal images of IBAT (A, B, E, F, I and J) and ingWAT (C, D, G, H, K and L) from wild-type and UCP1-KO mice acclimated to 30 °C, 18 °C and 4 °C. The tissues were stained for MAC-2 (green), perilipin (red) and nuclei (blue). Scale bar 50 μm. M and N, representative western blots of MAC-2 in IBAT (M) and ingWAT (N) from animals acclimated to the indicated temperatures. O and P, MAC-2 content in IBAT (O) and ingWAT (P) in the indicated samples. The mean value in IBAT of UCP1-KO mice acclimated to 4 °C was set to 1.0, and the levels in all other samples were expressed relative to this value. Each symbol represents a sample from one mouse. Values are means ± SEM. Where not visible, the error bars are smaller than the symbols. Full ANOVA statistics are given in Table S1. ∗Significant difference between wild-type and UCP1-KO mice for each tissue using two-way ANOVA followed by Tukey’s multiple comparison test. ∗∗∗p < 0.001. To facilitate comparisons of MAC-2 levels between IBAT and ingWAT, the respective graphs were drawn with equal y-axis range. In the right panel of P, the graph was redrawn with optimal y-axis range. UCP1-KO, UCP1-knockout.

In ingWAT, in contrast to the case in IBAT, no macrophages could be seen in mice acclimated to thermoneutrality (Fig. 2, C and D). In mice acclimated to subthermoneutral temperatures, macrophages were found only in ingWAT from UCP1-KO mice (Fig. 2, H and L) (for details see enlarged version of Fig. 2L (Fig. S2)). Thus, similar to the case in brown fat, under cold conditions, the ablation of UCP1 promotes macrophage accumulation in beige adipose depots; the effects seen in ingWAT were qualitatively similar but less robust compared to those in IBAT.

To estimate macrophage content in the tissues, we quantified MAC-2 protein using immunoblot analysis (Fig. 2, M and N). As seen in Figure 2M, in the IBAT, MAC-2 protein could essentially be detected only in tissues from UCP1-KO mice exposed to subthermoneutral temperatures. Immunoblot analysis failed to detect the MAC-2 protein in the IBAT from the mice acclimated to thermoneutrality, likely due to an immense difference in expression levels. In support of this, the analysis of only the samples from wild-type mice (Fig. S3) was fully consistent with our earlier report in which MAC-2 protein had been readily detected in similar samples (19). The total MAC-2 protein in the IBAT (Fig. 2O) was calculated by multiplying the protein levels of MAC-2 expressed per mg tissue protein (Fig. S4) with the total protein content in the tissue (Fig. 1C) and thus was found to display nearly infinite-fold induction upon acclimation of UCP1-KO mice to subthermoneutral temperatures. This observation was further supported by the analysis of a number of macrophage marker genes; the expression levels of all analyzed marker genes (general and M1 and M2 macrophage types) were highest in samples from UCP1-KO mice acclimated to 4 °C (Fig. S5).

Immunoblot analysis of MAC-2 in ingWAT revealed a discrepancy with the immunostaining results. Despite the absence of visually observable macrophages in the tissues from wild-type mice acclimated to subthermoneutral temperatures (Fig. 2, G and K), the MAC-2 protein in these tissues was clearly detectable by immunoblotting (Fig. 2N). This discrepancy may be attributed to the macrophages residing in the small lymph nodes within ingWAT, which may not have been fully removed during the dissection. In support of this notion is the significantly higher expression (approximately 5–10-fold) of several general macrophage marker genes in ingWAT compared to IBAT in cold-acclimated wild-type mice (Fig. S6), both of which contain negligible amounts of macrophages (Fig. 2, I and K). Due to the intricate connection between the adipose and lymphatic compartments in ingWAT, the levels of MAC-2 protein in ingWAT can only be approximately estimated. Despite this limitation, the results of immunoblot analysis of MAC-2 in ingWAT (Fig. 2, N and P) were qualitatively similar to those obtained in IBAT (Fig. 2, M and O), but the differences between wild-type and UCP1-KO mice were less drastic.

We thus confirm here that in the cold, the ablation of UCP1 triggers macrophage accumulation into brown adipose tissue. Importantly, we demonstrate that this is also the case in beige adipose tissue and thus, the phenomenon as such is unlikely to explain the differential response to UCP1 ablation in brown versus beige fat upon exposure to cold, where brown fat undergoes atrophy, while beige fat becomes hyper-recruited. This finding thus raises the question as to whether the macrophages occurring in brown and beige fat differ in their competence to affect adrenergic signaling in the tissue.

No norepinephrine-producing capacity of adipose tissue macrophages

Adipose tissue macrophages (alternatively activated), due to their suggested ability to synthesize and release norepinephrine, have earlier been proposed to directly stimulate adaptive thermogenesis in brown and beige adipose tissues (21). However, in an earlier study (19), we demonstrated that the macrophages accumulating in brown fat during prolonged exposure to thermoneutrality did not express tyrosine hydroxylase, the enzyme catalyzing the first and rate-limiting step in catecholamine biosynthesis (22). This agrees with the observation that the macrophages occurring in brown fat of mice acclimated to subthermoneutral temperatures (21 °C and 4 °C) do not synthesize catecholamines or contribute to adipose tissue adaptive thermogenesis (23). However, importantly, brown fat from those mice contained macrophages only at low densities. In contrast to that study, the brown and beige fat analyzed here from the UCP1-KO mice acclimated to 18 °C or to 4 °C were heavily infiltrated by macrophages. We examined whether these (or some of these) abundantly present macrophages possessed the capacity to produce catecholamines, potentially in a tissue-specific manner.

If macrophages in IBAT and ingWAT from UCP1-KO mice acclimated to subthermoneutral temperatures should be able to synthesize norepinephrine, this would require that they were endowed with tyrosine hydroxylase. We therefore investigated the presence and localization of tyrosine hydroxylase protein in situ using immunohistochemistry. Tyrosine hydroxylase protein was readily detected in both IBAT and ingWAT and appeared as dotted or filamentous staining (green) (Fig. 3, A and B). Importantly, as seen in Figure 3, C and D, no tyrosine hydroxylase immunoreactivity was detected within macrophages (red) (for details see enlarged version of Fig. 3C (Fig. S7)). Visualization of adipocytes with an anti-perilipin antibody (now stained red) (Fig. 3, E and F) demonstrated that these dotted and filamentous tyrosine hydroxylase-positive structures were closely associated with adipocytes but were never found within adipocytes (nor in any other cell type). Therefore, the tyrosine-hydroxylase-positive filamentous structures most likely correspond to sympathetic fibers, whereas the dotted structures represent the boutons-en-passant, the contact sites between the nerves and the adipocytes.

Open in a separate window

Figure 3

Macrophages in IBAT and ingWAT of UCP1-KO mice housed below thermoneutrality do not express tyrosine hydroxylase.A–F, representative confocal images of IBAT (A, C and E) and ingWAT (B, D and F) from UCP1-KO mice acclimated to 4 °C. The tissues were stained for tyrosine hydroxylase (TH) (green), MAC-2 (red) (C and D), perilipin (red pseudocolor) (E and F) and nuclei (blue). Scale bar 50 μm. UCP1-KO, UCP1-knockout.

Taken together, these observations expand previous data (obtained in different mouse models) that the macrophages infiltrating brown and beige adipose tissues do not express tyrosine hydroxylase and are therefore unable to synthesize catecholamines (19, 23)). Consequently, the differences in response to cold in IBAT versus ingWAT from UCP1-KO mice cannot be explained by a tissue-specific macrophage population competent in norepinephrine synthesis. Thus, the norepinephrine found in brown and beige adipose tissues must be of exclusively neuronal origin.

No evidence for norepinephrine-catabolizing capacity in a majority of adipose tissue macrophages

From the above data, it was concluded that the vast number of macrophages accumulated within brown and beige fat of UCP1-KO mice acclimated to cold could not affect the tissues through the synthesis of norepinephrine. However, two other studies have provided a novel aspect of macrophage function with respect to the regulation of adaptive thermogenesis (24, 25). Those studies suggested the existence of a subpopulation of macrophages, found around sympathetic nerves of subcutaneous (sympathetic neuron-associated macrophages (SAMs)) (24) or visceral (nerve-associated macrophages (NAMs)) (25) adipose tissue, that displayed a high expression of the norepinephrine-degrading enzyme monoamine oxidase-A (MAO-A). The SAMs also displayed a high expression of the neuronal norepinephrine transporter (NET) (official gene name: solute carrier family six member 2 (SLC6A2)). Therefore, these specialized macrophages have been ascribed a norepinephrine-lowering and thus thermogenesis-regulatory function.

To reveal whether these specialized macrophages displayed tissue-specific distribution and whether they, by being competent in taking up and catabolizing norepinephrine, could be part of the mechanism underlying the differences between UCP1-deficient IBAT and ingWAT in response to cold, we examined the presence and localization of MAO-A and NET in situ using immunohistochemistry (Fig. 4, A–D). MAO-A (red) was readily detected in a large number of cells of both IBAT and ingWAT (Fig. 4, A and B). Nearly all cells positive for MAO-A were negative for MAC-2 and positive for perilipin (for details see enlarged Fig. 4A (Fig. S8)); only a few cells positive for MAO-A displayed ambiguous staining patterns (they also exhibited positive but faint immunoreactivity for both MAC-2 and perilipin) (Fig. S8, white arrows). Importantly, the NET immunoreactivity was exclusively localized to the sympathetic nerves—it was found within filamentous and punctate structures that were also positive for tyrosine-hydroxylase and was never found within macrophages (Fig. 4, C and D). Thus, no macrophages endowed with NET and MAO-A could be detected either in brown or in beige adipose tissue from the UCP1-KO mice acclimated to cold. Consequently, the differences in response to cold between IBAT and ingWAT from UCP1-KO mice acclimated to cold cannot be explained by a tissue-specific population of macrophages competent in metabolizing norepinephrine. However, we should emphasize that our inability to identify NET- and MAO-A-positive macrophages does not necessarily totally exclude the existence of SAMs/NAMs in these macrophage-rich tissues, but it does suggest that the bulk of the macrophages were not competent to affect the tissues through lowering norepinephrine availability.

Open in a separate window

Figure 4

Macrophages in IBAT and ingWAT of UCP1-KO mice housed below thermoneutrality are not equipped with the enzymes necessary for norepinephrine catabolism.A–D, representative confocal images of IBAT (A and C) and ingWAT (B and D) from UCP1-KO mice acclimated to 4 °C. In (A and B), the tissues were stained for monoamine oxidase-A (MAO-A) (red), MAC-2 (green) and perilipin (blue pseudocolor). In (C and D), the tissues were stained for norepinephrine transporter (NET) (red), MAC-2 (green) and TH (blue pseudocolor). Scale bar 50 μm. Individual antibody staining is shown in the top-inset panels (scale bar 20 μm). Images were acquired with settings in each case allowing the maximum signal detection below the saturation limits of the detectors and are therefore not directly quantitatively comparable. Note that the MAO-A staining (red in A and B) overlapped only with perilipin staining (blue). Similarly, the NET staining (red in C and D) overlapped only with TH staining (blue). UCP1-KO, UCP1-knockout.

Cold exposure specifically induces the expression of MAO-A exclusively in UCP1-KO brown and beige adipocytes

As shown earlier, in brown and beige adipose tissues from the UCP1-KO cold-acclimated mice, MAO-A protein was detectable in the population of cells expressing perilipin, which thus corresponds to the adipocytes. However, in light of a recent report showing that in mice MAO-A was not expressed in adipocytes (26), this observation is unexpected and surprising. Importantly, MAO-A activity is readily measured in rat brown adipose tissue (e.g. (27)), but MAO-A is primarily considered an intraneuronal enzyme (28). To clarify whether MAO-A is innately expressed in brown and beige adipocytes, or its expression is selectively induced by UCP1 ablation, we analyzed the distribution of MAO-A immunoreactivity in tissues from both wild-type and UCP1-KO mice acclimated to cold (Fig. 5, A–L). In the same sections, sympathetic nerves were visualized with an anti-tyrosine hydroxylase antibody (green) and adipocytes with an anti-perilipin antibody (blue).

Open in a separate window

Figure 5

A strong induction and unexpected adipocyte localization of MAO-A in IBAT and ingWAT of UCP1-KO mice housed below thermoneutrality.A–L, Representative confocal images of IBAT (A, B, E, F, I and J) and ingWAT (C, D, G, H, K and L) from wild-type and UCP1-KO mice acclimated to 4 °C. The tissues were stained for MAO-A (red), TH (green) and perilipin (blue pseudocolor). Scale bar 50 μm. Images were acquired with settings in each case allowing the maximum signal detection below the saturation limits of the detectors and are therefore not directly quantitatively comparable. M and N, representative western blots of MAO-A and β-tubulin (cannot be used as a loading control due to the plasticity of the tissues) in IBAT (M) and ingWAT (N) from animals acclimated to the indicated temperatures. O, IBAT MAO-A content and (P) ingWAT MAO-A content in the indicated samples. The mean value in IBAT of wild-type mice acclimated to 4 °C was set to 1.0, and the levels in all other samples were expressed relative to this value. Each symbol represents a sample from one mouse. Values are means ± SEM. Where not visible, the error bars are smaller than the symbols. ∗Significant difference between wild-type and UCP1-KO mice for each tissue using two-way ANOVA followed by Tukey’s multiple comparison test. ∗p < 0.05, ∗∗∗p < 0.001. To facilitate comparisons of MAO-A levels between IBAT and ingWAT, the respective graphs were drawn with equal y-axis range. In the right panel of P, the graph was redrawn with optimal y-axis range. UCP1-KO, UCP1-knockout.

Tissue sections from wild-type mice displayed very scarce MAO-A immunoreactivity, with dotted or filamentous appearance (red) (Fig. 5, A and C), which entirely overlapped with tyrosine hydroxylase staining (Fig. 5, E and G). Thus, in brown and beige fat from wild-type mice, MAO-A displayed only neuronal (sympathetic) localization. In the tissues from UCP1-KO mice, a large number of cells exhibited strong MAO-A immunoreactivity (Fig. 5, B and D) (see also Fig. 4, A and B). However, MAO-A was not confined to adipocytes (Fig. 5, J and L), but was also localized to sympathetic nerves (Fig. 5, F and H). The staining intensity in nerves was weaker and could only be clearly visualized in the overexposed images (see enlarged and overexposed Fig. 5B (Fig. S9)).

Thus, we confirm that mouse adipocytes do not innately express MAO-A (in agreement with (26)), but upon ablation of UCP1, they become competent to express this norepinephrine-degrading enzyme. Importantly, although the brown and beige adipocytes display different anatomical localization and different molecular signatures (8), and are also of different embryological origin (29, 30), they do not differ in their ability to acquire expression of MAO-A in response to the ablation of UCP1 and the exposure to cold.

To quantify MAO-A in IBAT and ingWAT from wild-type and UCP1-KO mice, we quantified the MAO-A protein levels using immunoblot analysis (Fig. 5, M and N). In the mice acclimated to thermoneutrality, MAO-A protein levels (expressed per mg tissue protein) did not differ between the two genotypes in either of the tissues (Figs. 5, M and N and S10). However, under subthermoneutral conditions (18 °C and 4 °C), MAO-A levels were notably increased in both IBAT and ingWAT from UCP1-KO mice (Figs. 5, M and N and S10). MAO-A protein content in the total tissues was calculated by multiplying MAO-A protein levels expressed per mg tissue protein (Fig. S10) with the total protein content in the tissue (Fig. 1, C and D). As shown in Figure 5, O and P and in Table S1, the MAO-A content was only increased in the tissues from the UCP1-KO mice acclimated to 18 °C and 4 °C and was there approximately 10-fold higher than in the tissues from wild-type mice (in agreement with immunostaining data).

Therefore, it may be said that in cold, the brown and beige adipocytes lacking UCP1 acquired what appears to be a distinct neurogenic feature—the ability to express MAO-A—and thus also the potential to catabolize norepinephrine. As MAO-A is a mitochondrial outer membrane-bound enzyme that, by catalyzing deamination of catecholamines generates hydrogen peroxide, it may play a significant metabolic role not only by lowering norepinephrine availability but also by increasing oxidative stress in the tissues (see Discussion).

In the cold, the ablation of UCP1 results in notably enhanced sympathetic innervation in both brown and beige adipose depots

The above-observed recruitment of the norepinephrine-catabolizing enzyme MAO-A in the brown and beige adipose tissues from UCP1-KO mice acclimated to subthermoneutral temperatures would seem to indicate an increased norepinephrine-turnover in these tissues. This implies that the norepinephrine-producing pathways should also be increased. As shown earlier (19, 23), and also in this study (see above), within adipose tissues, tyrosine hydroxylase is exclusively localized to the sympathetic nerves. Thus, in these tissues, norepinephrine can only be of sympathetic origin. To explore whether there was an effect of UCP1 ablation on the level of sympathetic innervation in brown versus beige adipose tissue, we qualitatively (using immunohistochemistry) evaluated the degree of sympathetic innervation in each of the tissues (Fig. 6, A–H).

Open in a separate window

Figure 6

Notably enhanced sympathetic innervation in IBAT and ingWAT of UCP1-KO mice housed below thermoneutrality.A–H, representative confocal images of IBAT (A, B, E and F) and ingWAT (C, D, G and H) from wild-type and UCP1-KO mice acclimated to 4 °C. The tissues were stained for TH (green), perilipin (red) and nuclei (blue). Scale bar 20 μm. Images were acquired with settings in each case allowing the maximum signal detection below the saturation limits of the detectors and are therefore not directly quantitatively comparable. I and J, representative western blots of tyrosine hydroxylase (TH) and FABP4 (was not used as a loading control) in IBAT (I) and ingWAT (J) from animals acclimated to the indicated temperatures. K, IBAT TH content, (L) ingWAT TH content, (M) IBAT TH density, and (N) ingWAT TH density in the indicated samples. The value in IBAT of wild-type mice acclimated to 4 °C was set to 1.0, and the levels in all other samples were expressed relative to this value. Each symbol represents a sample from one mouse. Values are means ± SEM. Where not visible, the error bars are smaller than the symbols. ∗Significant difference between wild-type and UCP1-KO mice for each tissue using two-way ANOVA followed by Tukey’s multiple comparison test. ∗∗p < 0.01, ∗∗∗p < 0.001. To facilitate comparisons of TH levels between IBAT and ingWAT, the respective graphs were drawn with equal y-axis range. In the right panels of L and N, the graphs were redrawn with optimal y-axis range. UCP1-KO, UCP1-knockout.

Sympathetic nerves were visualized with an anti-tyrosine hydroxylase antibody (green). For clearer visualization of nerve fibers, in the upper panels, only tyrosine hydroxylase and nuclei (blue) are presented (Fig. 6, A–D). Counterstaining for perilipin (red) enabled visualization of adipocytes (Fig. 6, E–H). Even with the limitation that the images cannot be directly quantitatively compared (due to the higher exposure times for ingWAT samples, see Experimental procedures), it is evident that in mice acclimated to 4 °C, the tyrosine hydroxylase staining was much less abundant in ingWAT than in IBAT in both of the genotypes (Fig. 6, C, D, G and H). Importantly, the density of TH staining in the tissues from UCP1-KO mice (Fig. 6, B, F, D and H) appeared notably greater than in the corresponding tissues from wild-type mice (Fig. 6, A, E, C and G) and thus indicated an enhanced sympathetic innervation in these tissues.

The degree of sympathetic innervation was then examined quantitatively by measuring the levels of tyrosine hydroxylase protein in IBAT and ingWAT using immunoblot analysis (Fig. 6, I–N). In the mice acclimated to thermoneutrality, protein levels of tyrosine hydroxylase in each of the tissues did not differ between the two genotypes (Figs. 6, I and J and S11). However, under subthermoneutral conditions (18 °C and 4 °C), the levels in both IBAT and ingWAT from UCP1-KO mice were notably increased in comparison to the levels in the corresponding tissues of wild-type mice (Figs. 6, I and J and S11). To estimate the degree of sympathetic innervation in the entire brown and beige adipose depots, we calculated tyrosine hydroxylase protein content in the whole depots by multiplying the tyrosine hydroxylase protein levels expressed per mg tissue protein (Fig. S11) with the total protein content in the tissue (Fig. 1, C and D). As shown in Figure 6, K and L and in Table S1, the tyrosine hydroxylase content in both tissues in each genotype increased with exposure to increased cold stress. Notably, the tyrosine hydroxylase content in the tissues from the UCP1-KO mice was about 6-fold higher than in the corresponding tissues from the wild-type mice; in ingWAT from the UCP1-KO mice acclimated to 4 °C (Fig. 6L, empty blue symbols), the tyrosine hydroxylase content reached values similar to those in IBAT from the wild-type mice (Fig. 6K, filled blue symbols). The remarkably elevated tyrosine hydroxylase content in tissues from UCP1-KO mice acclimated to subthermoneutral temperatures indicates a substantial increase in norepinephrine-producing capacity.

To further characterize the consequence of UCP1 ablation on sympathetic innervation in brown versus beige adipose tissues, we estimated the density of sympathetic innervation in each tissue type. The relative content of tyrosine hydroxylase protein per unit tissue weight (tyrosine hydroxylase density) was much lower in ingWAT than in IBAT regardless of the genotype (Fig. 6, M and N). Importantly, tyrosine hydroxylase protein density in both tissues gradually increased with exposure to decreased environmental temperatures and was also much higher in the tissues from UCP1-KO mice, which agrees with the qualitative observations presented in Figure 6, A–H.

Thus, in the cold, the ablation of UCP1 evoked hyper-innervation of both brown and beige adipose tissues although the levels were much higher in brown fat.

The ablation of UCP1 markedly augments the expression of adrenergically regulated thermogenic marker genes

The enhanced sympathetic innervation may not necessarily be accompanied by an increased sympathetic activity. We therefore sought to estimate the magnitude of sympathetic activity in these tissues. We employed an indirect approach by determining expression levels of adipocyte-specific/selective thermogenic marker genes known to be induced by adrenergic signaling (as discussed in (31)).

Mice of both genotypes were acclimated to thermoneutrality (30 °C) or to cold (4 °C). Gene expression in IBAT and ingWAT was analyzed with real-time quantitative PCR. Reference genes were selected based on their expression, which was not affected by genotype or temperature: 18S rRNA in IBAT (Fig. 7A, top-left panel), and TFIIB in ingWAT (Fig. 7B, top-left panel) (see Experimental procedures). We initially confirmed the absence of UCP1 expression in the tissues from UCP1-KO mice (Fig. 7A, top-right panel) and (Fig. 7B, top-right panel). In line with previous data (e.g. (16)), acclimation to 4 °C led to highly significant cold-induced increases in UCP1 mRNA levels in both IBAT and ingWAT of wild-type mice (filled blue symbols versus filled red symbols). Since UCP1 mRNA could obviously not be utilized to estimate the degree of sympathetic activity in the tissues from UCP1-KO mice, we examined the extent of the induction of thermogenic marker genes known to be induced by adrenergic signaling. The genes analyzed included Fgf21, Elovl3, Pgc1a, glycerol kinase (Gk), type 2 iodothyronine deiodinase (Dio2), and Cidea. Upon acclimation to cold, the induction of practically all examined thermogenesis-related genes was significantly larger in both IBAT and ingWAT from UCP1-KO mice (Fig. 7, A and B, empty blue symbols versus filled blue symbols). This robust increase is strongly in accordance with an elevated sympathetic tone in these tissues. The only exception was Cidea in IBAT (Fig. 7A, bottom-right panel), the expression of which in brown fat is regulated posttranscriptionally and thus remained unaffected by either cold or genotype (16).

Open in a separate window

Figure 7

Significant upregulation of adrenergically regulated genes in both IBAT and ingWAT of UCP1-KO mice housed below thermoneutrality.A and B, expression levels of thermogenesis-related genes in IBAT (A) and ingWAT (B) of wild-type and UCP1-KO mice acclimated to the indicated temperatures. The expression levels were normalized using reference genes: 18 S rRNA for IBAT and TFIIB for ingWAT. Each symbol represents a sample from one mouse. Values are means ± SEM. For each gene, the reported value is the mean Ct value in the condition with the highest expression. Note that Ct values were consistently higher in ingWAT than in IBAT for every examined gene, indicating lower expression levels in ingWAT. ∗Significant difference between wild-type and UCP1-KO mice for each tissue using Student’s unpaired t test. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001. C and D, representative western blots of UCP1, glycerol kinase (GK) and CIDEA in IBAT (A) and ingWAT (B) from animals acclimated to the indicated temperatures. E, IBAT GK content, (F) ingWAT GK content, (G), IBAT CIDEA content, and (H) ingWAT CIDEA content in the indicated samples. The value in IBAT of wild-type mice acclimated to 4 °C was set to 1.0, and the levels in all other samples were expressed relative to this value. Each symbol represents a sample from one mouse. Values are means ± SEM. ∗Significant difference between wild-type and UCP1-KO mice for each tissue using two-way ANOVA followed by Tukey’s multiple comparison test. ∗∗p < 0.01, ∗∗∗p < 0.001. To facilitate comparisons of examined proteins between IBAT and ingWAT, the respective graphs were drawn with equal y-axis range. In the right panels of D and F, the graphs were redrawn with optimal y-axis range. UCP1-KO, UCP1-knockout.

The analysis of thermogenesis-related adrenergically regulated genes at the protein level also revealed an enhanced cold-induced induction in IBAT and ingWAT from UCP1-KO mice (Fig. 7, C–H). The analysis of UCP1 again verified the absence of UCP1 expression in the tissues from UCP1-KO mice (Fig. 7C, top panel and Fig. 7D, top panel). Glycerol kinase and CIDEA protein levels (Figs. 7, C and D and S12, A–D), as well as their total contents (Fig. 7, E–H) (calculated from the data presented in Figs. 1, C and D and S12, A–D), increased in both IBAT and ingWAT of animals acclimated to 18 °C and 4 °C. However, the final protein amounts were much higher in the tissues from UCP1-KO mice compared with the tissues from wild-type mice (Fig. 7, E–H, empty symbols versus filled symbols). The relative increases in both glycerol kinase and CIDEA protein contents, invoked by the ablation of UCP1, were higher in ingWAT than in IBAT (Fig. 7E versus Fig. 7F, right panel and also Fig. 7G versus. Fig. 7H, right panel). It is particularly noteworthy that the amounts of both proteins in ingWAT from UCP1-KO mice acclimated to 4 °C (Fig. 7, F and H, open blue symbols) reached values similar to those in IBAT from wild-type mice also acclimated to 4 °C (Fig. 7, E and G, filled blue symbols), as was essentially the case also for tyrosine hydroxylase (see above, Fig. 6, K and L).

The observed “overinduction” of glycerol kinase and CIDEA proteins in the brown and beige adipose depots of UCP1-KO mice exposed to cold, compared to those in wild-type mice, along with the “overinduced” mRNA levels of genes known to be induced by adrenergic signaling, strongly aligns with an elevated sympathetic tone in these tissues. This suggests that the ablation of UCP1 not only led to significantly enhanced sympathetic innervation in the brown and beige adipose tissues but also resulted in a parallel increase in the activity of the sympathetic fibers that innervate these tissues. The increase in e.g. glycerol kinase may therefore be seen not as a compensatory effect leading to increased thermogenesis due to futile lipid cycling (e.g. (32, 33)) but maybe the unavoidable consequence of enhanced adrenergic stimulation.

The macrophages that massively accumulate in IBAT and ingWAT of the UCP1-KO mice acclimated to cold do not influence the tissues by modulating norepinephrine availability

Macrophages represent the major immune cell types residing in adipose tissues (41, 42). Under obesogenic conditions, macrophages massively infiltrate visceral white adipose tissue (WAT) (43, 44); they aggregate around dead or dying adipocytes and form multinucleate crown-like structures (45, 46). Compared to visceral WAT, subcutaneous WAT displays a much lower propensity to accumulate macrophages (e.g. (47)). Also brown fat can accumulate remarkably high amounts of macrophages but the main factor leading to this is very prolonged exposure to thermoneutrality (19). However, under subthermoneutral conditions, both brown and beige (subcutaneous) fat demonstrate negligible amounts of macrophages. In this study, we detailed a unique (patho)physiological situation: a massive macrophage accumulation into both brown and beige fat that occurs in UCP1-KO mice under subthermoneutral conditions.

Certain macrophages, some of which are referred to as SAMs (sympathetic neuron–associated macrophages), found in brown and beige fat, have been suggested to play an important role in modulating norepinephrine availability and thus also a role in modulating thermogenic activity of these tissues (21, 24). Adipose tissues from the UCP1-KO mice acclimated to cold thus provided us with an unprecedented opportunity to further examine the competence of adipose tissue macrophages to modulate norepinephrine availability. However, we found that these macrophages did not express tyrosine hydroxylase, NET, and MAO-A, and consequently, they were not competent to either synthesize or degrade norepinephrine. Thus, the vast majority of macrophages accumulated within UCP1-deficient brown and beige fat under cold conditions cannot affect the tissues by modulating norepinephrine availability. These macrophages can thus not explain the diametrically opposite response to the cold of UCP1-deficient brown versus beige fat.

The conclusion that the macrophages massively accumulated in UCP1-deficient adipose tissues upon prolonged exposure to cold do not modulate norepinephrine availability in these tissues evidently raises the question of the physiological function of these macrophages. We suggest that they execute their conventional (but probably not their only) function: phagocytosis and degradation of dead cells, dying cells, and cellular debris. This scenario implies a high degree of cell death or a high degree of cell damage in adipose tissues from the UCP1-abated mice exposed to subthermoneutral conditions (supported by unpublished results of Kramarova L, Shabalina I and Boulet N). The nature of the signal(s) that triggers macrophage accumulation is not known, but some hypothetical signals are discussed below.

The possible role of adipose tissue-resident macrophages in the control of tissue innervation (48), the possible presence of cholinergic adipose macrophages (ChAMs) (49), and the ‘identity’ of macrophages—M1 versus M2 subtype at a single-cell level – have not been investigated here.

A possible physiological explanation for the high abundance of sympathetic nerves

Thermogenic recruitment—the increase in the capacity of thermogenesis-competent adipose depots to perform thermogenesis upon acclimation to cold—is mediated by norepinephrine (50). In this study, we demonstrated that the adipose tissue norepinephrine is exclusively of sympathetic origin. Exposure to subthermoneutral temperatures results in signals from the hypothalamus to increase sympathetic activity to brown (50) and also to beige adipose tissue (51, 52), in order to initiate recruitment. With time, this recruitment will result in an enhanced capacity for heat production, through enhanced cell proliferation and differentiation. A certain degree of sympathetic tone, proportional to the cold intensity, will maintain the recruited state once the necessary capacity has been reached. We suggest that the same process occurs in UCP1-KO mice—a given cold stress will activate the sympathetic outflow to brown and beige adipose tissues. As these tissues are unable to produce the heat required, the sympathetic activity will become further intensified, in a vain attempt by the central mechanisms of the animal to generate a heat-producing tissue. Thus, at any degree of cold stress, the tissues are more adrenergically stimulated than they are in the corresponding tissues of the wild-type mice.

In addition to the central mechanisms, adipose afferent sensory neurons have been also proposed to play a role in modulating sympathetic outflow to adipose tissues (53, 54). Similarly, sensory neurons within ingWAT may act as an inhibitory brake on the local sympathetic function (55). Therefore, also through these mechanisms, the lack of heat production in the UCP1-deficient adipose tissues would likely result in an augmentation of the local sympathetic activity.

Assessment of sympathetic activity

The remarkably increased expression levels of the norepinephrine-synthesizing enzyme tyrosine hydroxylase in both brown and beige fat of UCP1-KO mice, along with the unexpectedly acquired ability of brown and beige adipocytes to express the NE-degrading enzyme MAO-A, suggest an increased NE-turnover and thus also an increased adrenergic stimulation in IBAT and ingWAT from the UCP1-KO mice acclimated to cold. To estimate the magnitude of sympathetic activity in these tissues, we used an indirect approach: we measured mRNA and protein levels of adipose-selective molecular marker genes known to be induced by adrenergic signaling. The expression levels of these adrenergically induced genes correlate positively with the expression levels of tyrosine hydroxylase and thus strongly suggest a parallel increase in sympathetic activity.

The norepinephrine turnover rate and thus the sympathetic tone has been assessed in brown fat from the wild-type and UCP1-KO mice acclimated to 20 to 22 °C (56) (beige fat has not been examined). Sympathetic activity, even under these mild cold conditions, was about 8-fold higher in the IBAT of the UCP1-KO mice than that of the wild-type mice. Thus, our conclusion regarding the sympathetic activity in adipose tissues from UCP1-KO versus wild-type mice, based on an indirect approach, is in full agreement with the data obtained by measuring norepinephrine turnover rate.

Hyper-recruitment of beige fat

Higher adrenergic innervation in beige adipose tissue from UCP1-KO mice would be expected to result in hyper-recruitment of the tissue. In agreement with this expectation, the beige adipose depot from the UCP1-KO mice acclimated to the cold displayed enhanced recruitment (Fig. 1, B, D and F). This enhanced recruitment may therefore be understood not as a compensatory effect leading to UCP1-independent thermogenic responses, but rather as an unavoidable consequence of the enhanced adrenergic stimulation to the tissue because of the absence of thermogenesis.

In recent years, a number of alternative adaptive thermogenic mechanisms independent of UCP1 have been suggested. These have included futile cycles in a broad sense, such as a lipolysis/re-esterification cycle (e.g. (5, 32, 33)), creatine cycling (57), and calcium cycling (58). All these suggested mechanisms may be classified as being ATP-dependent, since they require that ATP is formed and then used in an ‘unproductive’ way, leading to ADP generation and consequently to stimulation of substrate oxidation/oxygen consumption in the mitochondria (i.e., thermogenesis). Thus, thermogenic mechanisms that are independent of UCP1 can only be accomplished in tissues that are endowed with considerable phosphorylation capacity, and their presence would be best discerned in tissues devoid of UCP1. The UCP1-KO mice thus represent an excellent tool to examine both UCP1-dependent and UCP1-independent thermogenic mechanisms. In the beige adipose depot deficient in UCP1, the oxidative capacity, compared with that of wild-type mice, is not diminished, although still much lower than in brown fat (Fig. 8 and (34)). Even having considered the increase in ATP synthase in the beige fat from UCP1-KO mice (Fig. 8), beige mitochondria generally have a remarkably low ATP synthase capacity (Fig. 8I and (34)). This would then indicate that ATP-consuming mechanisms are not likely to be responsible for major thermogenic processes in beige adipocytes, not even in UCP1-KO mice.

Why does sympathetic hyper-innervation not result in hyper-recruitment of brown fat?

In UCP1-KO mice acclimated to cold, sympathetic innervation was remarkably increased in both brown and beige adipose tissue. However, in contrast to the beige adipose tissue that, as expected, had undergone a hyper-recruitment, brown fat had become atrophied. Even though the increase in the total tissue protein in UCP1-KO mice acclimated to cold was not different from that in wild-type mice and thus appeared to indicate proper tissue recruitment (Fig. 1C), the content of mitochondrial respiratory chain proteins was strongly reduced (in agreement with (3, 4, 5)), the degree of adiposity was apparently higher and the tissue was massively infiltrated with macrophages. This may be interpreted to indicate that the recruitment of brown fat is not primarily regulated by sympathetic activity. However, as the magnitude of sympathetic innervation and also the expression levels of genes known to be induced by adrenergic signaling (Fig. 7) and MAO-A were much higher in brown than in beige fat, a more plausible alternative would be that the brown fat from UCP1-KO mice acclimated to cold is under such strong adrenergic stimulation that it could be considered to be beyond physiological.

Under strong cold stimulation, adrenergically induced lipolysis results in the release of fatty acids. In wild-type mice, these fatty acids are mostly shuttled towards the mitochondria where they initiate thermogenesis by activating UCP1 and are also utilized as a substrate for thermogenesis. Since the brown adipocytes lacking uncoupling activity have a limited capacity to combust fatty acids, it may be hypothesized that in brown fat deficient in UCP1, the free fatty acids reach such a high level that this results in cytotoxic effects. Whether this scenario, alone or in combination with oxidative stress caused by MAO-A (see below), could underlie the remarkable decrease in respiratory chain proteins in UCP1-KO tissues, remains to be investigated. Immune cells that massively accumulate into UCP1-deficient brown fat upon exposure to cold (Fig. 2 and (3, 7)) may be an indication of the tissue’s defense mechanism in response to e.g. lipolysis-(cold)-induced cytotoxicity or to damage caused by ROS produced due to MAO-A activity.

Protein analysis

The authors would like to thank the Experimental Core Facility staff for breeding the mice and the Imaging Facility at Stockholm University for help with confocal microscopy.