Deficiency in ZMPSTE24 and resulting farnesyl–prelamin A accumulation only modestly affect mouse adipose tissue stores

Running ZMPSTE24 deficiency has only modest effects on mouse adipose tissue Abstract Zinc metallopeptidase STE24 (ZMPSTE24) is essential for the conversion of farnesyl–prelamin A to mature lamin A, a key component of the nuclear lamina. In the absence of ZMPSTE24, farnesyl–prelamin A accumulates in the nucleus and exerts toxicity, causing a variety of disease phenotypes. By ~4 months of age, both male and female Zmpste24 –/– mice manifest a near-complete loss of adipose tissue, but it has never been clear whether this phenotype is a direct consequence of farnesyl–prelamin A toxicity in adipocytes. To address this question, we generated a conditional knockout Zmpste24 allele and used it to create adipocyte-specific Zmpste24 – knockout mice. To boost farnesyl–prelamin A levels, we bred in the “prelamin A–only” Lmna allele. Gene expression, immunoblotting, and immunohistochemistry experiments revealed that adipose tissue in these mice had decreased Zmpste24 expression along with strikingly increased accumulation of prelamin A. In male mice, Zmpste24 deficiency in adipocytes was accompanied by modest changes in adipose stores (an 11% decrease in body weight, a 23% decrease in body fat mass, and significantly smaller gonadal and inguinal white adipose depots). No changes in adipose stores were detected in female mice, likely because prelamin A expression in adipose tissue is lower in female mice. Zmpste24 deficiency in adipocytes did not alter the number of macrophages in adipose tissue, nor did it alter plasma levels of glucose, triglycerides, or fatty acids. We conclude that ZMPSTE24 deficiency in adipocytes, and the accompanying accumulation of farnesyl– prelamin A, reduces adipose tissue stores, but only modestly and only in male mice.


Introduction
ZMPSTE24, an integral membrane zinc metalloprotease (1), is required for the biogenesis of mature lamin A, a key component of the nuclear lamina (2,3). Lamin A is produced from a precursor protein, prelamin A, by four enzymatic processing steps (4). The cysteine in prelamin A's carboxyl-terminal CaaX motif (-CSIM) is farnesylated by protein farnesyltransferase. Next, the last three amino acids of the protein (-SIM) are clipped off by RCE1 (Ras-converting enzyme 1) or ZMPSTE24. The newly exposed farnesylcysteine is then methylated by ICMT (Isoprenylcysteine methyltransferase). Finally, the last 15 amino acids of prelamin A (including the farnesylcysteine methyl ester) are clipped off by ZMPSTE24, releasing mature lamin A.
Prelamin A-to-mature lamin A processing is normally very efficient, such that prelamin A is virtually undetectable in cells and tissues. However, prelamin A-to-lamin A processing is blocked by ZMPSTE24 deficiency (2,3). In the absence of ZMPSTE24, farnesyl-prelamin A accumulates in the cell nucleus, and the biogenesis of mature lamin A is completely abolished.
The accumulation of farnesyl-prelamin A in Zmpste24 -/mice is toxic, resulting in a variety of disease phenotypes (e.g., reduced growth, nonhealing bone fractures, sclerodermatous changes in the skin, and loss of adipose tissue) (2,5). The extent of disease depends on the level of prelamin A expression. When farnesyl-prelamin A production in Zmpste24 -/mice is reduced by 50% (by introducing a single knockout allele for Lmna), disease phenotypes are completely abolished (5).
The loss of adipose tissue in Zmpste24 -/mice is profound, such that white adipose tissue (WAT) is nearly undetectable in both male and female Zmpste24 -/mice by ~5 months of age (2,5). However, the mechanism for the loss of adipose tissue has been unclear. One possibility is that the loss of adipose tissue is a direct consequence of farnesyl-prelamin A toxicity in adipocytes.
Such a mechanism is plausible-for several reasons. Missense mutations in LMNA cause partial lipodystrophy in humans (6)(7)(8). Also, patients with mandibuloacral dysplasia Type B, a disease resulting from loss-of-function mutations in ZMPSTE24, have reduced adipose tissue stores (9,10). Finally, HIV protease inhibitors (HIV-PI) that have been linked to the side effect of partial by guest, on May 7, 2020 www.jlr.org Downloaded from lipodystrophy (e.g., lopinavir) inhibit ZMPSTE24 in cultured fibroblasts, resulting in an accumulation of farnesyl-prelamin A (11,12). Darunavir, an HIV-PI that is largely free of the lipodystrophy side effect, does not inhibit ZMPSTE24 or lead to an accumulation of farnesylprelamin A in fibroblasts (12). Despite these observations, there are ample reasons to be cautious about ascribing the loss of adipose tissue to the toxic effects of farnesyl-prelamin A. First, no one has actually tested the impact of farnesyl-prelamin A accumulation in adipocytes, and it is entirely conceivable that adipose tissue is resistant to the toxicity of farnesyl-prelamin A. For example, Zmpste24-deficient mice are free of liver disease despite a substantial expression of prelamin A in hepatocytes (2,5). Also, Zmpste24 -/mice have nonhealing bone fractures, most prominently in the ribs and the zygomatic arch (2,5), and it is conceivable that the loss of adipose tissue is secondary to these bone fractures (and reduced food intake) rather than being a direct result of farnesyl-prelamin A accumulation in adipose tissue.
In the current study, our goal was to determine if the loss of adipose tissue is a direct consequence of ZMPSTE24 inactivation in adipocytes (and the resulting accumulation of farnesyl-prelamin A). To pursue this goal, we created a conditional knockout allele for Zmpste24 (Zmpste24 fl ) and used it to create mice lacking ZMPSTE24 specifically in adipocytes. To minimize the possibility of overlooking a small effect of farnesyl-prelamin A on adipocyte biology, we generated adipocyte-specific Zmpste24 knockout mice that were homozygous for the "prelamin A-only" Lmna allele (Lmna PLAO ) (13,14). Prelamin A production from the Lmna PLAO allele is approximately twice-normal; thus, we were able to examine whether an exaggerated accumulation of farnesyl-prelamin A in adipocytes alters adipose tissue stores in mice. Echo-MRI. Body composition in live mice was measured using an EchoMRI 3-in-1 analyzer (EchoMRI Corp., Houston, TX), which assesses lean mass, fat mass, free water (mostly urine), and total water.
Plasma glucose, triglyceride, and free fatty acid levels. A blood sample (100 µl) was collected from anesthetized mice by retro-orbital puncture with a heparinized capillary tube (Kimble Chase).
Plasma was separated from red blood cells by centrifugation (13,000 ´ g for 30 sec) and stored at -80° C until analysis. Plasma triglycerides (Sigma, TR0100), free fatty acids (Abcam, ab65341), and glucose (Cayman Chemical, 10009582) were measured according to kit instructions.

Measurement of macrophage content in WAT by fluorescence activated cell sorting (FACS).
The stromal vascular fraction from gonadal WAT was prepared as described (16). Briefly, gonadal WAT was minced on ice, digested with collagenase type II (Worthington, LS004176) in PBS containing 0.5% BSA at 37° C, and filtered through a 100-µm filter. After centrifugation at 300 ´ g for 10 min at 4° C, the cell pellet was incubated with RBC lysis solution (Caprico Biotech), centrifuged, and the resuspended cell pellet filtered a second time through a 100-µm filter. The Western blotting. Urea-soluble protein extracts from tissues were prepared as described (5,17).
Targeted mouse embryonic stem (ES) cells were identified by long-range PCR and used to generate chimeric mice, which were bred with C57BL/6 females to create Zmpste24 fl/+ mice.
Adipocyte-specific Zmpste24 knockout mice. To inactivate Zmpste24 in adipocytes, we bred Zmpste24 fl/fl mice harboring a Cre transgene driven by the adiponectin promoter (Adipoq-Cre) (24,25). Quantitative (q)RT-PCR studies revealed that the Adipoq-Cre transgene was expressed in adipose tissue but not in liver. Also, fluorescence microscopy studies on tissues of Rosa mT/mG transgenic mice (26) carrying the Adipoq-Cre revealed recombination in adipose tissue but not in kidney or peritoneal macrophages (supplemental Figure S1A-C). In Zmpste24 fl/fl Adipoq-Cre + mice, Zmpste24 transcript levels were reduced by ~50% in WAT ( Figure 1B) and ~70% in brown adipose tissue (BAT) (Figure 1C), whereas transcript levels were not altered in liver and kidney and reduced by only 9% in peritoneal macrophages (supplemental Figure S1D-F).
We were uncertain whether the levels of farnesyl-prelamin A accumulation in adipocytes of Zmpste24 fl/fl Adipoq-Cre + mice would be sufficient to elicit disease phenotypes. For that reason, we bred Zmpste24 fl/fl Adipoq-Cre + mice homozygous for the prelamin A-only allele (Lmna PLAO ) (14,15). All of the output from the Lmna PLAO allele is channeled into prelamin A (rather than into both lamin C and prelamin A) (14,15), resulting in an ~twofold increase in prelamin A expression.
We showed previously that prelamin A in Lmna PLAO/PLAO mice is fully processed to mature lamin A and that twice-normal amounts of lamin A have no effect on the vitality of mice or body weight (14,15). Also, the levels of farnesyl-prelamin A in Zmpste24 -/-Lmna PLAO/PLAO mice are double those in Zmpste24 -/-Lmna +/+ mice (15). Not surprisingly, the disease phenotypes in Zmpste24 -/-by guest, on May 7, 2020 www.jlr.org
Western blots of BAT and WAT extracts from Lmna PLAO/PLAO Zmpste24 fl/fl Adipoq-Cre + mice revealed an accumulation of farnesyl-prelamin A (Figure 1D). There was no prelamin A accumulation in the liver of these mice. The accumulation of prelamin A in the WAT and BAT of Lmna PLAO/PLAO Zmpste24 fl/fl Adipoq-Cre + mice was located in adipocytes, as judged by immunohistochemistry (Figure 2). As expected, the prelamin A was located at the nuclear rim ( Figure 2). No prelamin A accumulation was observed in littermate mice lacking the Adipoq-Cre transgene (Figure 2). Also, no prelamin A was detected in the endothelial cells of adipose tissue (supplemental Figure S2).  (Figure 4A-B). There was a trend for lower BAT weights in male mice, but the difference did not achieve statistical significance ( Figure 4C) (P = 0.24). There were no differences in kidney weights ( Figure 4D).

Fat pad weights in female
Lmna PLAO/PLAO Zmpste24 fl/fl Adipoq-Cre + mice were no different than in littermate controls ( Figure   4A-C). The adipose tissue phenotypes in male Lmna PLAO/PLAO Zmpste24 fl/fl Adipoq-Cre + mice did not appear to be explained by changes in food consumption (supplemental Figure S3).
Because the severity of disease phenotypes in conventional Zmpste24-deficient mice depends on the level of prelamin A expression (5, 15), we hypothesized that the more prominent adipose tissue findings in male Lmna PLAO/PLAO Zmpste24 fl/fl Adipoq-Cre + mice might relate to greater amounts of farnesyl-prelamin A accumulation in adipose tissue. Indeed, as judged by western blotting, the prelamin A levels in BAT and WAT extracts were ~60% greater in male mice than in female mice (P < 0.02) (Figure 5A-B). The higher prelamin A protein levels in the male mice are likely due to increased expression of the Lmna gene. Prelamin A transcripts in adipose tissue were 46% higher in male mice than in female mice (P < 0.05) (Figure 5C). Prelamin A transcripts in the liver were similar in male and female mice ( Figure 5D). The modest decrease in adiposity in male Lmna PLAO/PLAO Zmpste24 fl/fl Adipoq-Cre + mice was not accompanied by perturbations in free fatty acid, triglyceride, or glucose levels (supplemental Figure S4).
We examined gene expression related to adipocyte differentiation, triglyceride metabolism, extracellular matrix synthesis, and the p53 pathway in the adipose tissue of  Figure 6C); however, we found no differences in the macrophage content of adipose tissue in the two groups of mice ( Figure   6D). Consistent with these findings, we did not observe changes in the expression of macrophage-

Discussion
We used a newly developed Zmpste24 conditional knockout allele and the Adipoq-Cre transgene to create adipocyte-specific Zmpste24 knockout mice. Our goal was to examine the impact of farnesyl-prelamin A accumulation in adipocytes. We created adipocyte-specific Zmpste24 knockout mice that were homozygous for the Lmna PLAO allele, reasoning that twice-normal amounts of prelamin A expression would make farnesyl-prelamin A toxicity more pronounced and easier to detect.
Our a priori expectation was that we would encounter substantial loss of adipose tissue in We were initially puzzled by the fact that changes in adipose tissue mass were evident only in male Lmna PLAO/PLAO Zmpste24 fl/fl Adipoq-Cre + mice, but we uncovered a likely explanation. The expression of prelamin A transcripts was ~40% higher in male mice than in female mice, and by western blotting the level of prelamin A accumulation in adipose tissue was ~60% higher in male mice. These 40-60% differences are obviously not enormous, but it is important to note that modest differences in farnesyl-prelamin A accumulation have a huge effect on disease phenotypes.
Reducing prelamin A expression levels by 50% in Zmpste24 -/mice completely eliminates disease phenotypes (5), whereas doubling prelamin A expression levels with the Lmna PLAO allele markedly increased the severity of disease (15). However, we cannot exclude the contribution of other mechanisms. For example, differences in genetic background have been suggested to explain the more severe lipodystrophy in male R482Q-lamin A transgenic mice (28), whereas androgen synthesis has been suggested to account for the earlier onset of cardiomyopathy in male Lmna H222P/H222P mice (29).
The fact that a deficiency of ZMPSTE24 in adipocytes and the accompanying accumulation of farnesyl-prelamin A did not induce lipodystrophy and lipodystrophy-related metabolic abnormalities will likely raise doubts about the relevance of ZMPSTE24 inhibition to the lipodystrophy observed in patients treated with HIV-PIs (e.g., lopinavir). The fact that therapeutic concentrations of lopinavir bind to ZMPSTE24 (30) and inhibit ZMPSTE24 activity in cultured fibroblasts is well documented (11,12,31), but it is important to note that the level of inhibition is far from complete. Even in the presence of high levels of lopinavir, more than half of the prelamin A in fibroblasts is cleaved by ZMPSTE24 and processed to mature lamin A (11,12).
Also, it is not clear that the lopinavir-induced accumulation of farnesyl-prelamin A observed in cultured fibroblasts occurs in the tissues of patients. In one study (32), prelamin A was detected by western blot in the adipose tissue of patients undergoing treatment for HIV, but the amount of prelamin A, relative to mature lamin A, was extremely low. Another study failed to detect any prelamin A in leukocytes from HIV-PI-treated patients (33). Those observations, combined with