Research ArticleACNE VULGARIS

Human sebum requires de novo lipogenesis, which is increased in acne vulgaris and suppressed by acetyl-CoA carboxylase inhibition

See allHide authors and affiliations

Science Translational Medicine  15 May 2019:
Vol. 11, Issue 492, eaau8465
DOI: 10.1126/scitranslmed.aau8465

Getting the skinny on sebum

Sebum, an oily material secreted by glands in the skin, has a physiological role, but abnormally high secretion of sebum can be associated with acne. Through a detailed investigation of sebum production in human skin samples and volunteer individuals, Esler et al. determined that most of the sebum in human skin is generated through de novo lipogenesis rather than recycled from circulating lipids. With this knowledge, the authors designed an inhibitor of acetyl-CoA carboxylase, a key enzyme in de novo lipogenesis, and tested it in cells, in rats, and in humans. This compound was well tolerated and successfully suppressed sebum production in human individuals.

Abstract

Sebum plays important physiological roles in human skin. Excess sebum production contributes to the pathogenesis of acne vulgaris, and suppression of sebum production reduces acne incidence and severity. We demonstrate that sebum production in humans depends on local flux through the de novo lipogenesis (DNL) pathway within the sebocyte. About 80 to 85% of sebum palmitate (16:0) and sapienate (16:1n10) were derived from DNL, based on stable isotope labeling, much higher than the contribution of DNL to triglyceride palmitate in circulation (~20%), indicating a minor contribution by nonskin sources to sebum lipids. This dependence on local sebocyte DNL was not recapitulated in two widely used animal models of sebum production, Syrian hamsters and Göttingen minipigs. Confirming the importance of DNL for human sebum production, an acetyl-CoA carboxylase inhibitor, ACCi-1, dose-dependently suppressed DNL and blocked synthesis of fatty acids, triglycerides, and wax esters but not free sterols in human sebocytes in vitro. ACCi-1 dose-dependently suppressed facial sebum excretion by ~50% (placebo adjusted) in human individuals dosed orally for 2 weeks. Sebum triglycerides, wax esters, and free fatty acids were suppressed by ~66%, whereas non–DNL-dependent lipid species, cholesterol, and squalene were not reduced, confirming selective modulation of DNL-dependent lipids. Last, individuals with acne vulgaris exhibited increased sebum production rates relative to individuals with normal skin, with >80% of palmitate and sapienate derived from DNL. These findings highlight the importance of local sebocyte DNL for human skin sebaceous gland biology and illuminate a potentially exploitable therapeutic target for the treatment of acne vulgaris.

INTRODUCTION

Sebum oils play a number of critical functions in the human integument, including thermoregulation, control of evaporation, moisturization, and antimicrobial defense (1). Increased rates of sebum production, however, correlate with severity of acne vulgaris (2), which presents as a spectrum of skin lesions including comedones, inflammatory papules, pustules, nodules, and cysts. The metabolic pathways underlying increased sebum production in humans with acne vulgaris remain incompletely understood.

Acetyl–coenzyme A (CoA) carboxylase (ACC), which catalyzes the conversion of acetyl-CoA to malonyl-CoA, plays a key role in regulating lipid metabolism. ACC is the rate-limiting step in the de novo synthesis of fatty acids [de novo lipogenesis (DNL)]. It also regulates oxidation of long-chain fatty acids. There are two closely related isoforms, ACC1, which predominates in the liver and adipose tissue, and ACC2, which predominates in skeletal muscle and heart. Inhibitors of ACC have been sought as potential treatments for metabolic disease (3) and cancer (4), and these agents inhibit DNL and stimulate fatty acid oxidation in cellular and animal models (58) and in human individuals (9). As part of a metabolic disease drug discovery program, PF-05175157 (ACCi-1), a potent inhibitor of ACC1 and ACC2 (9), was evaluated in nonclinical rat toxicity studies. ACCi-1 produced alterations in the morphology of cutaneous sebaceous glands consistent with reduced sebum content. We hypothesized that this was a consequence of DNL inhibition suppressing sebum biosynthesis. Although sebaceous glands contain functional ACC and are capable of DNL (10, 11), the importance of this pathway in humans for sebum biosynthesis relative to the utilization of circulating lipids derived from external sources, including diet, adipose tissue, or liver, is not known. In the present study, we evaluated the quantitative importance of local flux through the DNL pathway for sebum biosynthesis in cultured sebocytes, animal models, and humans.

RESULTS

ACC inhibition alters sebaceous gland morphology in rats

Rat toxicity studies of ACCi-1 (Fig. 1A) identified histological changes in cutaneous sebaceous glands consistent with reduced sebum lipid content, as well as likely secondary changes in the epidermis from the ventral abdomen (Fig. 1, B to E, and table S1). Sebaceous glands were atrophied, composed of smaller sebocytes with decreased amounts of cytoplasm and less lipid vacuolation (Fig. 1C) relative to sebocytes of vehicle-treated control animals (Fig. 1B). The infundibular hair follicle epithelium and interfollicular epidermis increased in thickness (epidermal hyperplasia), and there was thickening and compaction (hyperkeratosis) of the stratum corneum (Fig. 1E) relative to vehicle-treated control animals (Fig. 1D). Other less frequent changes observed in skin histology included epidermal and dermal inflammation, ulceration/erosion, and/or superficial bacterial colonization (table S1), thought to be secondary to a change in the amount or nature of the sebaceous secretion. However, the possibility of a primary effect of ACC inhibition on normal epidermal maturation and keratinization could not be entirely excluded. Similar findings of sebaceous gland atrophy and epidermal hyperplasia were also observed in toxicity studies with other ACC inhibitors (tables S2 and S3). Sebocyte necrosis or apoptosis was not observed in any of these studies, which were up to 4 months in duration. On the basis of these observations, we hypothesized that sebum production in humans may also be dependent on local flux through the DNL and may consequently be suppressed by ACC inhibition.

Fig. 1 ACC inhibition reduces sebaceous gland lipid content in rats.

(A) The chemical structure and potency of PF-05175157 (ACCi-1) as an inhibitor of human ACC isoenzymes 1 and 2 (hACC1 and hACC2, respectively). Geometric mean inhibitory concentration values are reported ±95% confidence bounds. Representative sections of pilosebaceous units in the dermis from rats treated orally once daily with vehicle or ACCi-1 (200 mg/kg) for 7 days. Comparison of sebaceous gland morphology (asterisk) in normal (B) and treated animal (C) in skin of ventral abdomen: ACCi-1 produced mild sebaceous gland atrophy consistent with reduced sebum content. Scale bars, 20 μm. Comparison of hair follicle epithelium (arrows) and interfollicular epidermis (arrowheads) in normal (D) and treated animal (E) in skin of ventral abdomen: Note increased thickness (increased cell layers) of follicular epithelium and epidermis (epidermal hyperplasia) and thickening and compaction of the stratum corneum (SC; arrow) (hyperkeratosis) in compound-treated animal. Scale bars, 50 μm.

ACC inhibition suppresses DNL and blocks sebocyte triglyceride production in cultured SZ95 human sebocytes

To characterize the importance of DNL for sebum biosynthesis, we evaluated the impact of pharmacological DNL inhibition on sebum production in cultured human sebocytes. SZ95 cells are an immortalized human sebocyte cell line that retains the major characteristics of sebocytes including lipid production (12). This cell line has been used to evaluate multiple agents that modulate sebum production (1316). ACCi-1 dose-dependently inhibited formation of malonyl-CoA with a half-maximal efficacious concentration (EC50) of 320 nM (Fig. 2A). The effect of ACC inhibition on sebaceous gland DNL was characterized in SZ95 cells treated with 14C-acetate in the presence of ACCi-1 or vehicle (Fig. 2B). ACCi-1 suppressed the incorporation of 14C-acetate into lipid species in these cells with an EC50 of 710 nM (Fig. 2B), comparable to the EC50 for malonyl-CoA inhibition (320 nM). Thin-layer chromatography analysis confirmed that ACCi-1 blocked incorporation of 14C into fatty acids and into lipid species containing fatty acids (glycerides, phospholipids, and cholesterol esters; Fig. 2C). Although 14C was also incorporated into free sterols including cholesterol, this incorporation was not inhibited by ACCi-1 (Fig. 2C). ACCi-1 did not produce any detectable changes in sebocyte viability in mock experiments in which 14C-acetate was omitted (fig. S1A). These observations are consistent with the mechanism of action of the compound because ACC inhibition would be expected to suppress incorporation of 14C-acetate into lipid species containing fatty acids, but not into free sterols, which proceed through the mevalonate pathway (3-hydroxy-3-methylglutaryl-CoA from acetyl-CoA and acetoacetyl-CoA) that does not involve ACC. The impact of DNL inhibition on sebaceous gland lipid production was quantified in cultured SZ95 sebocytes treated for 48 hours with ACCi-1 or vehicle. ACCi-1 suppressed sebocyte triglyceride concentrations, the major lipid species in human sebum (Fig. 2D), by up to about 90%, with an EC50 of 182 nM, comparable to the EC50 for inhibition of malonyl-CoA (320 nM) and DNL (710 nM). In contrast, concentrations of phosphatidylcholine (fig. S1B) and sphingomyelin (fig. S1C), major components of cell membranes but not of sebum, were unchanged with ACCi-1 treatment although it blocked the incorporation of DNL-derived fatty acids into phospholipids (Fig. 2C). These findings suggest that the fractional contribution of DNL to these phospholipid classes is relatively low or that non–DNL-derived fatty acids can be used as an alternative source by human sebocytes when DNL is inhibited. ACCi-2 produced a similar pattern of DNL inhibition (figs. S2 and S3).

Fig. 2 ACC inhibition suppresses malonyl-CoA, DNL, and sebum lipids in SZ95 sebocytes.

(A) The effect of ACCi-1 on ACC activity as assessed by intracellular malonyl-CoA concentrations (n = 5 to 6). (B) The effect of ACCi-1 on 14C-acetate incorporation into sebum lipids (DNL; n = 4) (C) The effect of ACCi-1 on conversion of 14C-acetate to individual lipid classes was assessed by thin-layer chromatography. CE, cholesterol esters; TG, triglycerides; FFA, free fatty acids; FS, free sterols (including cholesterol); DAG, diacylglycerol; MAG, monoacylglycerol; PL, phospholipids; DMSO, dimethyl sulfoxide. (D) The effect of ACCi-1 on sebocyte triglyceride biosynthesis. Triglyceride concentrations are reported as nanomoles per million cells seeded (n = 6). For all graphs, data are presented as means ± SEM. Error bars that are not visible are smaller than the symbol.

Sebum biosynthesis in humans requires local flux through the DNL pathway

To determine whether the DNL pathway plays an important physiological role in human sebum production, healthy individuals (n = 22; Fig. 3) received heavy water (2H2O) daily by oral administration for 2 weeks to measure isotope incorporated from enriched body water into de novo synthesized fatty acids (17, 18). Deuterium labeling of sebum palmitate and circulating very-low-density lipoprotein (VLDL)–triglyceride palmitate represents fatty acid–synthesized de novo from acetyl-CoA (Fig. 3A) (17, 18). Correction for the observed time course of isotopic enrichment of body water (18) was used to accurately quantify the fractional contribution from DNL relative to other sources of fatty acids.

Fig. 3 Sebum biosynthesis in humans depends on local DNL.

(A) Mechanism of incorporation of deuterium into palmitate through DNL. Deuterium from labeled water through the metabolic action of glycolysis, the tricarboxylic acid (TCA) cycle, malic enzyme, or direct hydrogen exchange may be incorporated into acetyl-CoA, malonyl-CoA, or NADPH [reduced form of nicotinamide adenine dinucleotide phosphate (NADP+)]. Flux through the DNL pathway can result in incorporation of the deuterium from these labeled precursors into newly synthesized palmitate. The measurement of the isotopic enrichment of palmitate relative to the precursor pool can be used to quantify the fractional contribution of DNL to palmitate. The hydrogen atoms depicted in red represent potential sites for deuterium incorporation from tissue 2H2O (not all sites are fully exchanged with body H2O, depending on metabolic conditions). FAS, fatty acid synthase. (B) Potential routes for incorporation of DNL-derived palmitate into sebum. DNL-derived palmitate can be incorporated into sebum by two distinct routes. Palmitate synthesized in the liver through hepatic DNL can be secreted into the circulation as VLDL-triglyceride (adipose DNL makes a minor contribution to nonesterified fatty acids released into the circulation). Sebaceous glands using palmitate from circulating lipids for sebum biosynthesis can incorporate this DNL-derived palmitate synthesized in the liver into sebum (green pathway). (C) Schematic for isotopic labeling study conducted in healthy individuals. Individuals were dosed with deuterated water (D2O) for 2 weeks. Plasma, sebum, and meibum samples were collected at the indicated times. (D) The percent contribution of DNL to plasma VLDL-triglyceride palmitate (n = 20 at all time points), sebum lipid palmitate (n = 5 for days 4, 7, and 11; n = 20 for day 14), and meibum lipid palmitate (n = 5 for days 4, 7, and 11; n = 20 for day 14) was quantified in human individuals. (E) The percent contribution of DNL to sebum sapienate (C16:1n10) was quantified (n = 5 for days 4, 7, and 11; n = 20 for day 14). For all graphs, data are presented as means ± SEM. Error bars that are not visible are smaller than the symbol.

If local sebaceous gland DNL is the major source of sebum fatty acids, rather than sources from outside of skin (such as lipids derived from diet, adipose tissue, or liver), then the fractional contribution of DNL to sebum lipids should be higher than the DNL contribution to the circulating lipids (Fig. 3B). In contrast, if circulating lipids (fatty acids esterified as glycerolipids or nonesterified fatty acids) are the principal metabolic sources for sebum fatty acids, then the fractional contribution of DNL to sebum and circulating lipids should be quantitatively similar (Fig. 3B).

Sebum was collected from facial skin using Sebutape (19), to establish a time course for isotopic enrichment and to assess the fractional contribution of DNL (17, 18) at steady state (Fig. 3C). Plasma and/or saliva samples were also collected to follow deuterium enrichment of body water and to measure fractional contribution from DNL to circulating lipids (VLDL-triglyceride). The quantitative contribution of DNL to sebum was compared to that for a similar lipid-based secretion by another specialized, epithelial, sebaceous organ, the meibomian gland of the eye. Accordingly, its product, meibum, was also collected.

Deuterium enrichment of body water reached about 1% by day 2 and remained at that enrichment throughout the study. Consistent with previous studies (17), the fractional contribution from DNL to VLDL-triglyceride palmitate (Fig. 3D) varied considerably among individuals (coefficient of variation, 61%) and within individuals day to day (see fig. S4 for individual subject data). This variability reflects the sensitivity of DNL to nutritional status, diet, and amount of physical activity (17). In aggregate, DNL contributed 20% of the total palmitate pool in circulating VLDL-triglyceride at steady state (Fig. 3D), consistent with the proportion reported by other studies of humans without metabolic disease, under nonfasted conditions (2022). Fractional labeling of sebum lipids reached steady state by day 11 of 14 days of dosing with 2H2O. Eighty percent of total sebum palmitate was derived from de novo synthesis (Fig. 3D). The substantially higher contribution of DNL-derived flux to sebum palmitate than to circulating triglyceride palmitate implies that the human sebaceous gland preferentially synthesizes fatty acids de novo rather than using circulating lipids for sebum biosynthesis. Label incorporation into circulating nonesterified fatty acids remained minimal after 2 weeks of 2H2O administration (fig. S5), suggesting that circulating nonesterified fatty acids do not account for the high fractional contribution of DNL to sebum palmitate.

To confirm the high contribution of DNL to sebum lipid, we quantified the incorporation of deuterium into sapienate, a monounsaturated fatty acid (C16:1n10) found exclusively in human sebum (23). As observed for human sebum palmitate, >80% of sebum sapienate was derived from DNL (Fig. 3E). In contrast to sebum, the fractional contribution of DNL to meibum palmitate mirrored the fractional contribution of DNL to circulating palmitate of ~20% (Fig. 3D), implying that human meibomian glands primarily use palmitate from circulating lipids for the biosynthesis of meibum.

ACCi-1 inhibits DNL in human individuals

To confirm that ACCi-1 inhibits DNL in humans, we assessed the suppression of hepatic DNL in a randomized, double-blinded, placebo-controlled, parallel-group study in healthy individuals. DNL was measured as incorporation of [1-13C]-acetate into VLDL-triglyceride palmitate. Labeled acetate was infused intravenously at a constant rate to isotopically enrich the precursor pool of hepatic, cytosolic acetyl-CoA (17, 18). Each individual had a baseline measurement of DNL in the absence of study treatment (Fig. 4A), followed 1 week later by a second measurement of DNL after a single oral dose of either ACCi-1 (100, 250, 600 mg) or placebo (Fig. 4B). To reduce variability in rates of hepatic DNL, individuals received a 10-hour oral fructose load throughout the period of plasma sampling for DNL measurement, which has been shown to ensure reproducible rates of DNL during repeat measurements in human individuals (24).

Fig. 4 ACC inhibition suppresses DNL and reduces sebum production in healthy human individuals.

(A to C) The effect of ACCi-1 versus placebo on hepatic DNL. (A) Fractional contribution of DNL to VLDL palmitate measured in the fructose-stimulated state in healthy individuals during the baseline period using isotopic labeling. Arms A to C, n = 8; arm D, n = 6. Arms A to D were treated identically during the baseline period. Data are presented as means ± SEM. (B) Fractional contribution of DNL to VLDL palmitate measured in the fructose-stimulated state in healthy individuals after treatment with ACCi-1 or placebo. Arm A (placebo), arm B (100 mg), and arm C (250 mg), each had n = 8; Arm D (600 mg), n = 6. Data are presented as means ± SEM. (C) Dose response for ACCi-1 DNL inhibition in human individuals. The ratio of fractional contribution of DNL during the treatment period relative to the baseline period is shown. The arithmetic mean (●) and individual values (●) are depicted. Box plot provides median and 25%/75% quartiles with whiskers to the last point within 1.5 times the interquartile range. (D to L) Healthy individuals were dosed with ACCi-1 [200 mg twice daily (BID), n = 10] or matched placebo (n = 5) for 2 weeks. Sebum lipids were collected at baseline and end of treatment using Sebutape and analyzed using gas chromatography with flame ionization detection. Total sebum production was assessed using Sebumeter. (D) Summary of the relative change in total sebum from baseline to after treatment, as assessed using Sebumeter. (E) Time course for relative change in sebum triglyceride from baseline (defined as the mean of values from days −3 and −2) for placebo (○)– and ACCi-1 (●)–dosed individuals after treatment (days 13 and 14). Individual subject level data are shown. Summary of the relative change from baseline to after treatment for sebum triglyceride (F), sebum free fatty acids (G), sebum wax esters (WE) (H), sebum cholesterol (I), sebum squalene (J), sebum triglyceride sapienate (TG 16:1n10) (K), and sebum wax ester sapienate (WE 16:1n10) (L). For (D) and (F) to (L), all individual subjects are depicted as circles. Box plots provide median and 25%/75% quartiles with whiskers to the last point within 1.5 times the interquartile range. *P < 0.05, **P < 0.01, ***P < 0.001. For all graphs, error bars that are not visible are smaller than the symbol. n.s., not significant.

During the baseline DNL assessment, oral fructose increased the fractional DNL contribution to VLDL-triglyceride palmitate over the 10-hour loading period, representing a change in peak fractional DNL contribution of about 30%, after subtracting the fractional DNL contribution during the fasted state (Fig. 4A). A similar baseline DNL response to fructose was observed for individuals across each of the four treatment arms (Fig. 4A). During the treatment phase (Fig. 4B), the measurement of DNL in placebo-treated individuals showed a response similar to baseline (ratio of posttreatment to baseline peak DNL is 0.99). In contrast, ACCi-1 dose-dependently inhibited fructose-stimulated DNL (Fig. 4B) by 15, 45, and 75%, respectively, after 100-, 250-, and 600-mg doses (Fig. 4C). Plotting the average unbound plasma drug concentration versus the average DNL inhibition for each individual over the 10-hour duration of the study (fig. S6) resulted in an EC50 of 321 nM. The robust DNL inhibition confirmed utility of ACCi-1 in studying the biology of this pathway in humans.

DNL inhibition reduces sebum secretion and ACC-dependent lipids in human sebum

ACCi-1 is a systemically acting compound that freely distributes from plasma to tissue compartments, allowing distribution to and pharmacological activity in skin. ACCi-1 (200 mg BID, n = 10) or placebo (n = 5) was administered to healthy individuals for a period of 2 weeks (Fig. 4, D to L). A dose of 200 mg BID at steady state was expected to produce about comparable DNL inhibition to a single dose of 600 mg. The full effect on sebum production was expected within 2 weeks because the fractional contribution of DNL to sebum palmitate plateaued in 11 days (Fig. 3D), indicating that the sebum lipid pool turns over completely within 11 days. Treatment with ACCi-1 was generally well tolerated. In individuals treated with 200 mg BID of ACCi-1, six adverse events (three diarrheas, two nauseas, and one decreased appetite) that were considered treatment related were reported. All of these adverse events were mild except for one event of diarrhea in one individual, which was classified as moderate in intensity.

Total sebum excretion rate assessed by Sebumeter was suppressed 49% from baseline (placebo corrected) by ACCi-1 treatment (Fig. 4D). Comparison of sebum lipids from samples collected before (days −3 and −2) and after treatment (days 13 and 14) demonstrated that DNL inhibition robustly suppressed sebum biosynthesis. Triglycerides, the predominant lipid species in sebum (25), were reduced 66% (placebo adjusted) from baseline in the ACCi-1–dosed individuals (Fig. 4, E and F). Free fatty acids (Fig. 4G) and wax esters (Fig. 4H), which are expected to be dependent on DNL, were reduced by a similar magnitude. In contrast, sebum cholesterol (Fig. 4I) was not suppressed. Further, sebum squalene increased twofold over baseline (Fig. 4J), suggestive of compensation. The observation that sebum cholesterol and squalene were not suppressed is consistent with (i) their synthesis by ACC- and DNL-independent pathways and (ii) the specificity of ACC inhibition by ACCi-1. ACCi-1 did not affect the availability of alternative sources of fatty acid because circulating triglycerides were unchanged (fig. S7). Further evidence that ACCi-1 reduces secretion of sebum by directly inhibiting DNL in sebocytes is the lower observed sapienate (16:1n10) in the triacylglycerol (Fig. 4K) and wax ester (Fig. 4L) fractions of sebum samples from treated patients.

Syrian hamsters and Göttingen minipigs rely on circulating lipids for sebum biosynthesis

The Syrian hamster ear skin model is widely used to evaluate pharmacological modulators of sebum production because the sebaceous glands of these animals are structurally similar to those of humans (26). The contribution of DNL to sebum lipids and circulating lipids using in vivo 2H2O metabolic labeling was evaluated in Syrian hamsters. The fractional contribution of DNL to circulating triglyceride palmitate and to sebum was equivalent to about 55% (fig. S8A). Because the fractional contribution of DNL-derived fatty acids to sebum and to circulating lipids was the same in hamsters, this suggests that hamster’s sebocytes use predominantly circulating lipids, rather than synthesizing fatty acids de novo for sebum production.

To test this interpretation, oral (100 mg/kg) administration of ACCi-1 was compared with topical application to the Syrian hamster ear skin (10% solution, 5 μl/cm2) of an ACC inhibitor (ACCi-2; fig. S2) with properties more suitable for topical administration. ACCi-2 had higher intrinsic clearance in human hepatocytes than ACCi-1 (13 μl/min per million versus <2 μl/min per million, respectively). ACCi-2 also had better aqueous solubility at pH 6.5 than ACCi-1 (860 and 81 μM, respectively). If DNL-derived fatty acids in sebum were synthesized within the sebaceous glands of hamster ear skin, then oral and topical administration of ACC inhibitor would be expected to suppress label incorporation into sebum to a similar extent. Both routes of exposure suppressed ACC activity in skin to the same extent, as indicated by decreased malonyl-CoA (fig. S8B). Oral ACCi-1 administration, but not topical ACCi-2 inhibitor administration, suppressed 14C-acetate incorporation into lipid within the liver (fig. S8C), demonstrating that orally administered but not topically administered ACC inhibitor could inhibit ACC and DNL systemically. Topically administered ACCi-2 did not inhibit incorporation of DNL-derived fatty acids into sebum (fig. S8D). Conversely, orally administered ACCi-1 did suppress incorporation of DNL-derived fatty acids into sebum (fig. S8D).

Together, these data indicate that DNL-derived fatty acids incorporated into hamster sebum were synthesized remotely (liver) and delivered via circulation, rather than synthesized locally in the sebocyte. Consequently, topical ACC inhibitor treatment would not be expected to reduce sebum biosynthesis in hamsters. Further, oral ACCi-1 administration would also not be expected to suppress sebum biosynthesis in hamsters unless it lowered circulating lipids sufficiently to reduce sebocyte uptake. Consistent with this hypothesis, oral administration of ACCi-1 for 19 days did not reduce sebum production in Syrian hamsters (fig. S8E). This suggests that circulating lipids derived from diet or adipose were sufficient for sebum biosynthesis in this species. These findings indicate that sebum secretion by skin of Syrian hamster’s ears does not recapitulate human biology.

The Göttingen minipig is another widely used animal model because the morphology and physiology of its skin resemble those of human skin (27). The fractional contribution of DNL to sebum lipids and circulating lipids was evaluated in Göttingen minipigs. The fractional contributions of DNL to sebum and to circulating triglyceride palmitate mirrored each other (fig. S7F), suggesting that the Göttingen minipig relies predominantly on circulating lipids rather than local DNL within the sebaceous gland for sebum biosynthesis.

Elevated DNL flux underlies increased sebum production rate in humans with acne vulgaris

Rates of sebum production are markedly higher in patients with acne vulgaris than non-acne control subjects. This increased sebum production plays a causal role in acne pathophysiology (23). The metabolic pathways responsible for sebum overproduction in patients with acne have not been characterized. Accordingly, 2H2O labeling studies and sebum sampling were carried out in patients with healthy skin and in patients with a diagnosis of acne vulgaris followed in an academic dermatology clinic [University of California, Davis (UC Davis)].

DNL contribution to sebum palmitate and sapienate was 86 ± 9% (range, 57 to 98%) and 90 ± 12% (range, 61 to 100%), respectively, for the acne group (Fig. 5, A and B). These values were not significantly different from healthy controls (P ≥ 0.5; Fig. 5, A and B). Total production rate through DNL and non-DNL pathways was calculated for the two major fatty acids present in human sebum, palmitate (Fig. 5C) and sapienate (Fig. 5D), based on fractional contribution of DNL and rate of sebum production. Comparing patients with acne with the healthy individuals, the total rates of sebum production and of flux through the DNL pathways were >20% higher in those with acne (P < 0.05). These results demonstrate that increased flux through local DNL in the sebocyte is the pathway primarily responsible for increased sebum production rate in humans with a clinical diagnosis of acne vulgaris.

Fig. 5 Increased flux through DNL is the primary source of increased sebum lipids in human individuals with acne.

(A) The percent contribution of DNL to sebum lipid palmitate in healthy volunteers with normal skin (HV) or patients with acne. (B) The percent contribution of DNL to sapienate in healthy volunteers with normal skin or patients with acne. (C) Relative contributions of DNL-derived and non–DNL-derived palmitate to total sebum palmitate in healthy volunteers with normal skin or patients with acne. (D) Relative contributions of DNL-derived and non–DNL-derived sapienate to total sebum sapienate in healthy volunteers or patients with acne.

DISCUSSION

In the present study, we demonstrate that human sebum production in healthy individuals is highly dependent on local flux through the DNL pathway, that the two most widely used animal models of sebum production do not faithfully recapitulate the central role of sebocyte DNL in humans, and that human meibomian glands use circulating lipids rather than local DNL for meibum lipid biosynthesis. We also report that overproduction of sebum lipids in human patients with acne vulgaris is caused by increased flux through the local sebocyte DNL pathway and that an orally administered ACC inhibitor (ACCi-1), which inhibits hepatic DNL in healthy volunteers, suppresses sebum production in humans.

In general, humans on a western diet appear to rely on DNL as a source of fatty acids in adipose tissue and liver much less than rodents (17). DNL contributes a relatively minor fraction of nonessential fatty acids in human fat stores by comparison with the contribution from dietary sources (17, 18, 20, 22). Our observation that 24% of circulating triglyceride palmitate in VLDL was derived from DNL is consistent with this notion. However, this does not preclude certain lipid pools from depending heavily on DNL, perhaps to sustain critical biological processes, such as protection against infection, during starvation.

Similar to most cells, sebocytes and cultured human sebaceous glands (28) are capable of de novo synthesis of fatty acids. Studies using 14C-acetate administration have demonstrated that de novo synthesized fatty acids can be found in sebum collected from both humans (29, 30) and nonclinical species including Syrian hamsters (31, 32). Likewise, isolated sebaceous glands, similar to many organs including the liver/hepatocytes (33, 34), adipocytes (35), mammary gland (36), lung (37, 38), and meibomian glands (39, 40), are capable of DNL (28, 41, 42). These earlier studies, however, have two important limitations. First, the methods used in the in vivo setting could not distinguish whether these DNL-derived fatty acids were synthesized locally in the sebaceous gland or synthesized in other lipogenic organs and delivered via circulation and subsequently incorporated into sebum. Second, without the advent of modern combinatorial probability analytic methods [mass isotopomer distribution analysis (MIDA) (18, 43)], the quantitative importance of DNL-derived fatty acids relative to other sources of fatty acids including lipolysis and dietary fat could not be accurately determined.

The study reported here demonstrates that about 80 to 85% of the two major fatty acids in human sebum, palmitate and sapienate, is derived from DNL. The contribution of DNL to sebum was markedly higher than that to circulating palmitate (~20%), which can only be explained by local DNL within sebaceous glands playing a major role in sebum production. In contrast to human sebum, the relative contribution of DNL to palmitate in human meibum was similar to that in circulating lipids. These observations suggest that meibomian glands, a specialized form of sebaceous glands, use circulating lipids for meibum lipid biosynthesis.

The high fractional contribution of DNL to sebum lipids suggested that DNL inhibition might suppress sebum production. To test this hypothesis, we evaluated the impact of an ACC inhibitor (ACCi-1) on sebum production in both preclinical and clinical studies. ACCi-1 dose-dependently suppressed DNL in human sebocytes in vitro, blocking synthesis of fatty acids and fatty acid–containing lipid species (triglycerides and wax esters), whereas synthesis of cholesterol, which is not dependent on DNL or ACC, was unchanged. Moreover, human individuals dosed orally for 2 weeks with ACCi-1 demonstrated suppression of sebum lipid species containing fatty acids and a ~50% placebo-corrected suppression of facial sebum excretion. In these individuals, cholesterol and squalene were not suppressed, confirming specific modulation of DNL-dependent lipids.

Although we cannot rule out the possibility that DNL from other epidermal sources may also play a role, the following observations suggest that sebum production in humans is highly dependent on local DNL within sebocytes and susceptible to ACC inhibition. Lipids isolated from the forehead and face using Sebutape, reported to be mainly of sebaceous origin with a minor (3 to 6%) contribution of epidermal components (44), contained no phospholipids, which would have been indicative of contamination with epidermal lipids. Sapienate, a fatty acid unique to human sebum, was demonstrated to be highly dependent on DNL and was suppressed in sebum triglycerides and wax esters after treatment of human patients with ACCi-1. In addition, ACC inhibition also markedly suppressed the concentrations of wax esters, which are found in sebum but not in other dermal lipids.

Excess sebum production is an important component of the pathogenesis of acne vulgaris. Suppression of sebum secretion reduces acne incidence and severity (23, 4550). We show here that the local DNL pathway is primarily responsible for increased sebum lipid production in humans with acne. Meta-analysis across multiple therapeutic classes has demonstrated that sebum reduction on the order of 50%, as observed with ACCi-1, would be predicted to reduce acne lesions to a clinically meaningful extent (23). In contrast to oral acne medication isotretinoin, which is thought to lower sebum by causing sebocytes to undergo apoptosis (5153), the present studies open the possibility that substantial sebum lowering might be achieved by specific pharmacologic interventions, either systemically or locally, at a well-defined node in metabolism. ACCi-1 suppressed sebum production without lowering sebum squalene. Because squalene is an important emollient in skin (54), suppression of sebum production with a mechanism that spares squalene may minimize the drying of skin, an adverse effect observed with some sebum-lowering acne medications. Additional studies are required to test this hypothesis.

Limitations of the present work include (i) that studies conducted in human individuals were of relatively short duration, (ii) that ACCi-1 interventional studies were conducted in healthy individuals, and (iii) that the hepatic DNL inhibition study in humans was conducted with fructose loading. Fructose loading has been used in this study and others (9, 5557) to overcome the substantial diet-related variability of DNL measured in free-living human individuals (58) and ensure reproducible rates of hepatic DNL during repeat measurements. These conditions are artificial, and we cannot exclude the possibility that DNL dose response could be different in the absence of fructose administration. However, the EC50 for hepatic DNL measured in human individuals during fructose loading (321 nM) was in general agreement with the EC50 for inhibition of DNL in rats measured without fructose loading (326 nM) (9) and for inhibition of malonyl-CoA production (320 nM) and DNL (710 nM) in SZ95 sebocytes, also measured without fructose loading. Additional studies are also required to define the degree of sebum lowering that can be achieved through ACC inhibition in patients with acne vulgaris and to determine whether administration of a topical ACC inhibitor recapitulates the findings observed with oral ACC inhibitor administration. Further, studies of longer duration are required to determine whether sebum lowering with DNL inhibitors translates to improvements in acne lesion count and severity.

In summary, the mechanistic studies presented here identify sebocyte DNL as a pathway of importance in the biology of human skin and in the pathogenesis of acne vulgaris. Moreover, the observed dependence of human sebum production on local DNL flux and the effectiveness of DNL inhibition by an ACC inhibitor to suppress sebum production in humans indicate that clinical evaluation of this pathway for the treatment of acne may be warranted.

MATERIALS AND METHODS

Study design

The objective of this study was to investigate the importance of DNL for sebum biosynthesis based on the initial observation that an ACC inhibitor reduces sebaceous gland lipid content in rat toxicology studies. Assessment of sebaceous gland morphology was part of a routine rat toxicity study to support applications to regulatory authorities for conduct of human clinical trials. Animals were randomized into dose groups, and unblinded microscopic evaluation of the tissues was conducted and peer-reviewed by veterinary pathologists (K.E.B. and F.J.G.) board-certified by the American College of Veterinary Pathologists. In vitro studies in human SZ95 cells were sized to enable assessment of DNL and were not conducted blinded.

The study assessing fractional contribution of DNL to sebum, meibum, and VLDL-triglyceride in healthy human individuals was an open-label, randomized, parallel study where four cohorts, each consisting of five individuals, were randomized into four arms differing in the timing of procedures. The planned sample size of five individuals per cohort was empirically chosen on the basis of the need to understand the intersubject variability to quantify the fractional contribution of DNL sebum and meibum. This study was conducted unblinded.

The study assessing the effects of ACCi-1 on DNL in healthy human individuals was a randomized, investigator- and subject-blinded (sponsor-open), placebo-controlled four-armed (three doses of ACCi-1 and placebo), parallel-group study. During the study, a total of 30 individuals were randomized to receive one of the four doses (placebo, 100, 250, or 600 mg). Each individual had a baseline assessment of DNL in the absence of study treatment on day −6. Each individual had the second assessment of DNL while receiving either ACCi-1 or placebo on day 1. Eight individuals per dose with six completers provided 90% power to detect at least a 25% decrease in the percentage change in peak DNL between the baseline period (reference period) and treatment period at a significance level of 5%, assuming that the between-subject SD was 12%.

Assessment of effect of ACCi-1 on sebum lipidomics and sebum production was performed as an exploratory endpoint in a multiple-dose study designed to assess safety, tolerability, and pharmacokinetics of ACCi-1. The study was a single-center, randomized, double-blinded (investigator- and subject-blinded, sponsor-open), parallel-group, placebo-controlled, 14-day repeated-dose study of ACCi-1 in otherwise healthy overweight/obese individuals randomized in a 2:1 ratio to receive either ACCi-1 (n = 10) or placebo (n = 5). This sample size of 15 individuals was selected as a compromise between the need to minimize exposure to ACCi-1 and the need to have sufficient individuals to enable useful conclusions to be drawn on pharmacokinetic parameters.

The study comparing the contribution of DNL to sebum production patients with healthy skin (n = 10) and in patients with acne vulgaris (n = 9) was conducted unblinded. Sample analysis was conducted in a blinded manner. Sample size was determined on the basis of the SD for total sebum palmitate production rate through the DNL pathway in healthy volunteers (12% SD). Power calculations for sample size needed to achieve a 20% difference in patients with acne vulgaris with 80% power at P < 0.05 gave a sample size of six. We chose nine patients with acne vulgaris in case the SD was higher than in healthy individuals.

For the Syrian hamster studies, animals were randomized into treatment groups. Only daily drug administration was performed in an unblinded manner, and all other aspects of study conduct and analysis were blinded to treatment. Sample size was determined to enable 90% power to detect at least 30% inhibition of malonyl-CoA and at least 51% inhibition of DNL. The study in minipigs was sized on the basis of technical feasibility and was unblinded because there was a single study arm with no drug treatment.

Animal studies

All procedures performed on the study animals were in accordance with regulations and established guidelines by the Association for Assessment and Accreditation of Laboratory Animal Care International and Guide for the Care and Use of Laboratory Animals (National Research Council, eighth edition) and were reviewed and approved by the Institutional Animal Care and Use Committee.

Effect of ACC inhibition on sebaceous gland morphology in rats

Male and female Wistar Han International Genetic Standardization [CRL:WI(Han)] rats (10 per sex per dose) received once daily oral gavage doses of ACCi-1 at 200 mg/kg per day for 7 days (females) and 8 days (males). Control animals (10 per sex) received the vehicle (0.5% methylcellulose) at a dose volume of 10 ml/kg. Animals were humanely euthanized by exsanguination while under general anesthesia induced with isoflurane. A gross examination was performed, and a full set of body tissues was collected for microscopic examination and preserved in 10% neutral-buffered formalin. Tissues collected included the following: adrenal, aorta, bone marrow, brain, cecum, cervix, colon, duodenum, epididymis, esophagus, eye, gut-associated lymphoid tissue, Harderian gland, heart, ileum, inguinofemoral lymph node, jejunum, joint (stifle), kidney, larynx, liver, lung, mammary gland, mesenteric lymph node, optic nerve, ovary, oviduct, pancreas, parathyroid, peripheral nerve (sciatic), pituitary, prostate, salivary gland, seminal vesicle, skeletal muscle (gastrocnemius), skin (ventral abdomen, hind limb, forelimb, nose, lips, and eyelid), spinal cord, spleen, stomach, testis, thymus, thyroid, tongue, trachea, ureter, urinary bladder, uterus, vagina, and prepuce. Tissues were processed routinely for histologic examination [dehydrated through increasing concentrations of alcohol, perfused with paraffin, embedded in paraffin blocks, cut on a microtome (Microm HM355 or Leica RM2255) at 5-μm thickness, placed on glass slides, deparaffinized, and stained with hematoxylin and eosin]. Microscopic evaluation of the tissues was conducted by board-certified veterinary pathologists (K.E.B. and F.J.G.). Histologic images of representative sebaceous glands from a male animal receiving vehicle (Fig. 1, B and D) and a male animal receiving 200 mg/kg per day of ACCi-1 (Fig. 1, C and E) were captured on an Olympus BX51 light microscope using a SPOT RT color digital camera (Diagnostic Instruments Inc.).

Determination of contribution of DNL to sebum biosynthesis in Syrian hamster using stable isotopic labeling

Syrian golden hamsters weighing between 150 and 200 g were purchased from Harlan Laboratories and maintained on a PicoLab Rodent Diet 20 (LabDiet 5053). Animals were dosed with 2H2O to enrich body water with deuterium for assessment of the fractional contribution of DNL to plasma triglyceride palmitate and sebum total palmitate. At the start of 2H2O labeling, animals were injected intraperitoneally with a loading dose of 8% 2H2O at 3.5 ml/kg and maintained thereafter with 8% 2H2O for the remainder of the study via enrichment of the drinking water. At the end of the appropriate labeling period, hamsters were euthanized by CO2 asphyxiation, and tissue was removed and snap-frozen in liquid nitrogen. A portion of ear skin was removed by 8-mm punch biopsy. Ear skin was further separated into ventral and dorsal portions using tweezers to tease the ears apart. The ventral portion of ear skin was used for all analyses. Blood was collected using cardiac puncture after CO2 asphyxiation, and plasma was frozen on dry ice. Fractional contribution of DNL to plasma triglyceride palmitate and skin total palmitate was determined using MIDA as described in the Supplementary Materials.

Effect of oral and topical administration of ACC inhibitor on malonyl-CoA production and DNL in Syrian hamsters

Syrian golden hamsters weighing between 150 and 200 g were purchased from Harlan Laboratories and maintained on a PicoLab Rodent Diet 20 (LabDiet 5053). For assessment of malonyl-CoA concentrations, hamsters were treated with a single dose of ACC inhibitor administered orally (ACCi-1, 100 mg/kg) or topically (ACCi-2, 10% solution, 5 μl/cm2) 2 hours into the light phase. One hour after administration of the ACC inhibitor, hamsters were euthanized via CO2 asphyxiation. Tissue was rapidly removed and snap-frozen in liquid nitrogen. A portion of pinna ear skin was removed by 8-mm punch biopsy. Ear skin punch was further separated into ventral (front) and dorsal (back) skin layers using tweezers to tease the ears apart. The ventral portion (separated from the dorsal skin and intervening cartilage) of ear skin was snap-frozen in liquid nitrogen and subsequently used to quantify malonyl-CoA concentrations. Malonyl-CoA content was analyzed by liquid chromatography–tandem mass spectrometry in the Metabolomics Core Facility at the Sanford Burnham Prebys Medical Discovery Institute for Biomedical Research (Orlando, FL, USA).

For assessment of DNL inhibition, hamsters were treated with a single dose of ACC inhibitor administered orally (ACCi-1, 100 mg/kg) or topically (ACCi-2, 10% solution, 5 μl/cm2) 2 hours into the light phase. One hour after the dose, 14C-acetate (0.1 μCi/g) was administered by intraperitoneal injection to each animal. One hour after administration of 14C-acetate, the animals were euthanized via CO2 asphyxiation, and tissues were collected. A portion of the liver (~400 mg) was removed using an 8-mm punch biopsy, and ear skin was collected as described above. The liver tissue was weighed and immediately placed in capped tubes containing NaOH (1.5 ml of 2.5 mM) and heated (~60°C) until the tissue was fully degraded. Neat ethanol (2.5 ml) was added to each tube, and the samples were vigorously mixed and allowed to sit overnight at room temperature. Petroleum ether (4.8 ml) was added, and tubes were vortexed vigorously. The samples were centrifuged (1500g for 5 min), and the organic layer (top) was removed and discarded. Concentrated HCl (0.6 ml of 12 M) was added to the remaining aqueous phase of each sample (including the interface material), which was then capped and vortexed vigorously. The acidified aqueous phase was extracted with petroleum ether (4.8 ml) and then centrifuged (1500g for 5 min). The upper organic phase from each sample was removed and placed into 20-ml scintillation vials, capped, and set aside. The remaining aqueous phase (including the interface material) was again extracted with petroleum ether (4.8 ml) and vortexed vigorously. The samples were centrifuged (1500g for 5 min). The upper organic phase was again removed and then pooled with the previous extract in the scintillation vial. The pooled organic extractions were evaporated to dryness under nitrogen at room temperature, Ultima Gold scintillation fluid (10 ml; PerkinElmer) was added to each vial, and 14C content was quantified by scintillation counting.

Determination of contribution of DNL to sebum biosynthesis in Göttingen minipigs using isotopic labeling

Naïve, male Göttingen minipigs (16 weeks old; 7 to 8 kg) were purchased from Marshall BioResources. To assess fractional contribution of DNL to plasma triglyceride palmitate and sebum total palmitate, deuterated water was administered to the animals to enrich body water with deuterium. To commence labeling of body water, animals were given a loading deuterated water dose equivalent to 2.5% of their total water content. The loading dose was divided into two equal volumes given by oral gavage 60 min apart. After this and throughout the study, animals were provided drinking water ad libitum containing a final deuterated water concentration of 5%. Blood was collected via ear vein on days 1, 3, 4, 7, 10, 14, 21, 24, 28, and 32 of the study, processed for plasma, and frozen on dry ice. Sebutape was used for collection of sebum from the lateral surface of the Göttingen minipig on days 2, 3, 4, 7, 10, 14, 17, 21, 24, 28, and 32. The fractional contribution of DNL to plasma triglyceride palmitate and sebum palmitate was determined using MIDA.

Human studies

All studies involving human individuals (ClinicalTrial identifiers: NCT01537497 and NCT01807377) were conducted in compliance with the ethical principles originating in or derived from the Declaration of Helsinki and in compliance with all International Conference on Harmonisation Good Clinical Practice Guidelines. In addition, all local regulatory requirements of the investigational centers participating in the studies (University of Cincinnati Children’s Hospital Medical Center, Miami Research Associates, Profil Research Institute, and UC Davis) were followed, in particular, those affording greater protection to the safety of trial participants. Final study protocols and informed consent documents were reviewed and approved by the investigational centers participating in the studies and by an independent institutional review board (IRB). The investigator was required to inform the IRB of the study’s progress and occurrence of any serious and/or unexpected adverse events. A signed and dated informed consent was required before any screening procedures were initiated. The investigator or appropriate delegate explained the nature, purpose, and risks of the study to each individual. Each individual was informed that he/she could withdraw from the study at any time and for any reason. Each individual was given sufficient time to consider the implications of the study before deciding whether to participate. Individuals who chose to participate signed an informed consent document.

Determination of contribution of DNL to biosynthesis of sebum and meibum in human individuals using stable isotopic labeling

The study included male (n = 7) or female (n = 15) healthy individuals between the ages of 18 and 49 years. Healthy was defined as no clinically relevant abnormalities identified by a detailed medical history, full physical examination (including blood pressure and pulse rate measurement), 12-lead electrocardiogram, and clinical laboratory tests. Individuals were randomized into four arms differing in the timing of procedures. Eligible individuals who met the entry criteria were admitted to the Clinical Translational Research Center (CTRC) at Cincinnati Children’s Hospital Medical Center for about 24 hours to receive multiple oral loading doses of 70% deuterated water (60 ml every 3 hours for a total of 480 ml). All cohorts continued to consume daily oral 2H2O doses (60 ml of 70% 2H2O per day) as outpatients until day 14. Individuals returned to the CTRC for five outpatient visits (including the follow-up visit) and underwent sebum and blood collection. Sebum was collected using Sebutape (CuDerm Corp.). Before application of the Sebutape strips, the skin was cleansed of debris by washing with soap and water and defatted by wiping with a gauze pad saturated in hexane. Once the skin was dry, the Sebutape was peeled from its backing paper using defatted forceps and affixed to the cleansed surface with gentle pressure for adequate adhesion. Three patches were placed in the following areas: one patch on each cheek (caudal of the middle line of the eye) and one on the forehead (cranial of the glabella). After 3 hours, the Sebutape was removed and placed in acid-washed Teflon screw-cap vials. For the meibum collection, tetracaine local anesthetic was applied by eye drop to reduce pain and discomfort during collection of meibum. After the tetracaine was applied, the upper and lower lids of the eye were carefully cleaned of oil and debris using cotton swabs moistened with sterile water. The meibum was expressed from the meibomian glands of the eyelids with gentle digital pressure, using the cotton swabs. The secreted material was harvested using a Kimura platinum spatula. During collection, the lid margin was rolled away from the eye globe to minimize contamination of the sample with tears and conjunctival cells. The secreted material did not come into contact with the swab. At all outpatient visits, blood samples were taken for determination of deuterated water enrichment in body water (from plasma) and for determination of VLDL lipids. Timing of sebum and meibum collection was staggered across cohorts, allowing incorporation of deuterium into de novo synthesized fatty acid to be measured at different time points over the 14-day period, whereas sebum was collected on day 14 from all individuals. Incorporation of deuterium into plasma VLDL-triglyceride palmitate and sebum total palmitate was assessed at each time point to establish steady-state fractional contribution of DNL to these lipid pools. Fractional contribution of DNL was determined using MIDA.

Assessment of DNL inhibition in human individuals using stable isotopic labeling

Inhibition of DNL by ACCi-1 was assessed in a randomized, double-blinded, placebo-controlled four-armed (three doses of ACCi-1 and placebo), parallel-group study in healthy volunteers. Healthy was defined as above.

During the study, a total of 32 individuals (ages 22 to 52) were enrolled and randomized to receive 100, 250, or 600 mg of ACCi-1 or placebo. To complete the experimental procedures under standardized conditions (DNL measurements, observation of overall safety and tolerability, and acquisition of pharmacokinetic samples), individuals were confined to the Clinical Research Unit (CRU) for about 12 overnight stays. Each individual had a baseline assessment of DNL in the absence of study treatment on day −6. Each individual had the second assessment of DNL after receiving either a single oral dose of ACCi-1 or placebo on day 1. Assessment of DNL required 13C-acetate infusion to be initiated the night before the assessment day. At about 10:00 p.m. on day −7, a continuous infusion of 13C-acetate (9 to 9.5 mg 1-13C-acetic acid sodium salt/min via an infusion pump) was started and continued until about 6:30 p.m. on day −6 (until the last sample had been drawn for DNL measurement). Blood samples for assessment of the fractional contribution of DNL to VLDL-triglyceride palmitate were collected hourly for 10 hours. Similarly, at about 10:00 p.m. on day 0, a comparable continuous infusion of 13C-acetate was started and continued until about 6:30 p.m. on day 1 (until the last sample for DNL measurement had been drawn). On day 1, after an 8-hour fast, individuals received study medication (ACCi-1 or matching placebo) at about 8:00 a.m. (±2 hours), followed by the second DNL assessment. In addition to the blood samples collected for the DNL measurements, samples were also collected at times ranging from 0.5 to 10 hours after a dose for pharmacokinetic analysis.

Oral fructose loading was used during the DNL assessments to ensure reproducible fractional contribution of DNL to VLDL-triglyceride palmitate from one assessment period to the other. On days −6 and 1, each individual received a bolus of 0.25 g fructose/kg body weight every 30 min starting at about 8:30 a.m. for about 9.5 hours (total of 20 fructose administrations). All individuals were required to lie in a semisupine position (except when required for blood pressure, pulse rate, and electrocardiogram measurements). Individuals refrained from eating and drinking beverages other than water and the fructose drinks administered as part of study procedures. DNL was measured by 13C incorporation into VLDL-triglyceride palmitate using MIDA.

Assessing the effect of ACC inhibition on sebum production in human individuals

The effect of ACCi-1 on sebum production was assessed in a randomized, double-blinded, parallel-group, placebo-controlled, 14-day repeated-dose study in 15 healthy individuals randomized in a 2:1 ratio to receive either ACCi-1 (n = 10) or placebo (n = 5). Eligible individuals were admitted to the CRU and remained confined to the CRU for about 20 days. While confined, individuals were on a standardized diet (nutritional composition about 50% carbohydrates, 35% fat, and 15% protein). Individuals were dosed daily in the morning with 200 mg BID of ACCi-1 or placebo. Safety and tolerability were assessed by adverse event monitoring, clinical laboratory safety tests, and cardiovascular parameters (blood pressure, heart rate, and electrocardiogram). The amount of sebum was assessed before (days −3 and −2) and at the end of the study (days 13 and 14) using a Sebumeter SM 815 (Courage + Khazaka electronic GmbH). Sebumeter measurements were performed in the morning (about 9:00 a.m.). Just before Sebumeter measurements, the forehead of each individual was cleansed by study site personnel to remove existing sebum. Measurements were made on each individual’s forehead at 5 min and at 3 hours after cleansing. Samples of sebum were also collected on these same days using Sebutape as described above. Lipid content of Sebutape was measured using MS at Metabolon by applying their sebum lipidomics panel as described in the Supplementary Materials.

Assessment of DNL in patients with acne vulgaris using stable isotopic labeling

Nineteen patients with acne and 10 patients with normal skin (age, 25.8 ± 6.7 years old) were recruited through the Department of Dermatology, UC Davis. Acne severity was graded on the basis of the global acne grading score. All individuals provided informed written consent before participation, and the protocol was approved by the UC Davis IRB. Those who had initiated a systemic retinoid in the previous 6 months or had another facial inflammatory condition were excluded from enrollment. Individuals ingested deuterated water for 1 week, and facial sebum was collected through the use of Sebutapes and cigarette paper 1 week after completion of deuterium ingestion. The cigarette paper (rizla red medium weight) was cut into 2 cm by 2 cm pieces, washed in diethyl ether, and dried before use. The individual’s forehead was washed with soap and water and defatted using alcohol wipes. The washed cigarette paper was placed against the clean forehead skin, covered with a 4 cm by 4 cm piece of aluminum foil, and held in place with an elastic bandage. After a 3-hour collection period, the cigarette paper was removed, placed in acid-washed vials, and stored at −20°C for future lipid analyses. The sebum excretion rate was assessed at each visit with the use of a Sebumeter.

Statistical analysis

Statistical analyses were performed at two-sided α level of 5%. When two groups were compared and after assessment of data normality and equality of the variances between groups, the statistical analyses were performed using either a parametric t test or a nonparametric Wilcoxon signed-rank test. When more than two groups were compared and after assessment of data normality and homoscedasticity of the different variances, the statistical analyses were performed using either an analysis of variance (ANOVA) test, followed by a Tukey post hoc test or a nonparametric Kruskal-Wallis ANOVA.

SUPPLEMENTARY MATERIALS

stm.sciencemag.org/cgi/content/full/11/492/eaau8465/DC1

Materials and Methods

Fig. S1. ACC inhibition does not alter cell viability or lower phospholipid concentrations in SZ95-cultured human sebocytes.

Fig. S2. The chemical structure of ACCi-2 is shown.

Fig. S3. The effect of ACCi-2 on conversion of 14C-acetate to individual lipid classes was assessed by thin-layer chromatography.

Fig. S4. Fractional contribution of DNL to VLDL-triglyceride in healthy human volunteers is variable.

Fig. S5. Fractional contribution of DNL as measured in plasma free fatty acid is much lower than DNL measured in VLDL-triglyceride.

Fig. S6. Plasma drug concentration response relationship is shown for hepatic DNL inhibition in human individuals.

Fig. S7. ACC inhibitor treatment in healthy volunteers for 2 weeks does not alter circulating triglyceride concentrations.

Fig. S8. Sebum biosynthesis in Syrian hamster and Göttingen minipig is not dependent on sebocyte DNL.

Table S1. Incidence of histologic findings in the ventral abdominal skin of male rats treated orally with ACCi-1 for 14 days.

Table S2. Incidence of histologic findings in the ventral abdominal skin of rats treated orally with ACCi-2 for 1 month.

Table S3. Incidence of histologic findings in the ventral abdominal skin of male rats treated orally with PF-05221304 for 16 weeks.

Table S4. Individual triglyceride species quantified from SZ95 cells using liquid chromatography–tandem mass spectrometry and associated multiple reaction monitoring transitions.

Table S5. Individual phosphatidylcholine species quantified from SZ95 cells using liquid chromatography–tandem mass spectrometry and associated multiple reaction monitoring transitions.

Table S6. Individual sphingomyelin species quantified from SZ95 cells using liquid chromatography–tandem mass spectrometry and associated multiple reaction monitoring transitions.

Reference (59)

REFERENCES AND NOTES

Acknowledgments: We thank C. C. Zouboulis for supplying the SZ95 human sebocyte cell line and J. Whitsett and B. Trapnell for help in the design and the conduct of the healthy individual isotopic labeling study. We also acknowledge C. Pertucci and S. Gardell from the Sanford Burnham Medical Research Institute for malonyl-CoA measurements. We would like to thank A. Bergman and J. Rusnak for input on the clinical studies and M. Birnbaum, S. O’Rahilly, and M. Wan for helpful discussions. We thank S. Chari for assistance with DNL analysis. Funding: These studies were funded by Pfizer Inc. Support for the studies in patients with acne vulgaris was through unrestricted funds to M.K.H. at UC Berkeley. Author contributions: W.P.E. contributed to the design and interpretation of nonclinical and clinical studies, led the research project, and wrote the paper. G.J.T., P.A.A., and J.J.P. contributed to the design and interpretation and conduct of in vitro and/or in vivo studies and reviewed the paper. S.M.T., C. Beysen, M.K.H., and M.F. contributed to the design and interpretation of stable isotope studies, performed analysis for these studies, and helped in writing and reviewing the paper. S.M.W. developed analytical methodologies for analysis of sebum lipids, analyzed the human sebum lipid samples, contributed to data interpretation, and helped in writing and reviewing the paper. S.C.-G. contributed to the clinical and nonclinical study design, performed statistical analysis, and reviewed the paper. K.E.B. and F.J.G. performed the microscopic evaluation of rat sebaceous glands, captured images, and reviewed the paper. C. Buckeridge and A.M.S. contributed to the design of the human sebum reduction study and reviewed the paper. D.A.G. designed ACCi-1, contributed to the design and interpretation of nonclinical studies, and edited and reviewed the paper. M.G. performed cell-based studies in cultured sebocytes and reviewed the paper. N.B.V. developed analytical methodologies for analysis of in vitro sebocyte triglycerides, analyzed the in vitro sebocyte lipid samples, and reviewed the paper. R.S., L.H., and S.S.B. conducted the clinical study comparing the fractional contribution of DNL to sebum in human patients with healthy skin or acne and reviewed the paper. T.P.R. and J.A.P. contributed to hypothesis generation, study design, and data interpretation and contributed to writing the paper. G.E.S. oversaw the design and execution of clinical studies and helped in writing the paper. Competing interests: W.P.E., G.J.T., P.A.A., S.C.-G., F.J.G., K.E.B., C. Buckeridge, A.M.S., D.A.G., M.G., N.B.V., J.A.P., and G.E.S. are employees and shareholders of Pfizer Inc. J.J.P. and T.P.R. were employed by Pfizer Inc. at the time these studies were conducted. C. Beysen, M.K.H., and S.M.W. are or were previously scientific consultants to Pfizer Inc. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials. The compounds used in this study are patented by Pfizer Inc. (see, for example, US8288405, US8802688, US8859577. and US9145416. Requests for these reagents for academic research purposes can be made to the Pfizer Chemical Library and Transfer Program. SZ95 cells were licensed under a biological materials licensing agreement from C. C. Zouboulis.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article