Abstract

Background

   Long non-coding RNAs (lncRNAs) can perform tasks of key relevance in
   fat cells, contributing, when defective, to the burden of obesity and
   its sequelae. Here, scrutiny of adipose tissue transcriptomes before
   and after bariatric surgery ([71]GSE53378) granted identification of
   496 lncRNAs linked to the obese phenotype. Only expression of
   linc-GALNTL6-4 displayed an average recovery over 2-fold and
   FDR-adjusted p-value <0.0001 after weight loss. The aim of the present
   study was to investigate the impact on adipocyte function and potential
   clinical value of impaired adipose linc-GALNTL6-4 in obese subjects.

Methods

   We employed transcriptomic analysis of public dataset [72]GSE199063,
   and cross validations in two large transversal cohorts to report
   evidence of a previously unknown association of adipose linc-GALNTL6-4
   with obesity. We then performed functional analyses in human adipocyte
   cultures, genome-wide transcriptomics, and untargeted lipidomics in
   cell models of loss and gain of function to explore the molecular
   implications of its associations with obesity and weight loss.

Results

   The expression of linc-GALNTL6-4 in human adipose tissue is
   adipocyte-specific and co-segregates with obesity, being normalized
   upon weight loss. This co-segregation is demonstrated in two
   longitudinal weight loss studies and two cross-sectional samples. While
   compromised expression of linc-GALNTL6-4 in obese subjects is primarily
   due to the inflammatory component in the context of obesity,
   adipogenesis requires the transcriptional upregulation of
   linc-GALNTL6-4, the expression of which reaches an apex in terminally
   differentiated adipocytes. Functionally, we demonstrated that the
   knockdown of linc-GALNTL6-4 impairs adipogenesis, induces alterations
   in the lipidome, and leads to the downregulation of genes related to
   cell cycle, while propelling in adipocytes inflammation, impaired fatty
   acid metabolism, and altered gene expression patterns, including that
   of apolipoprotein C1 (APOC1). Conversely, the genetic gain of
   linc-GALNTL6-4 ameliorated differentiation and adipocyte phenotype,
   putatively by constraining APOC1, also contributing to the metabolism
   of triglycerides in adipose.

Conclusions

   Current data unveil the unforeseen connection of adipocyte-specific
   linc-GALNTL6-4 as a modulator of lipid homeostasis challenged by
   excessive body weight and meta-inflammation.

   Keywords: linc-GALNTL6-4, Adipose tissue, Adipocytes, Triglycerides,
   Obesity

Graphical abstract

   [73]Image 1
   [74]Open in a new tab

Highlights

     * •
       The non-conserved expression of linc-GALNTL6-4 in adipose tissue is
       adipocyte-specific and opposite to obesity.
     * •
       Adipogenesis requires the transcriptional regulation of
       linc-GALNTL6-4, which reaches an apex in differentiated fat cells.
     * •
       Impaired adipocyte-specific linc-GALNTL6-4 levels are mainly due to
       the inflammatory state in the context of obesity.
     * •
       The loss of linc-GALNTL6-4 in adipocytes displays altered lipidomes
       and the downregulation of genes related to cell cycle.
     * •
       While propelling inflammation, defective linc-GALNTL6-4 may impair
       FA metabolism and the expression of APOC1.
     * •
       In engineered adipocytes, increased linc-GALNTL6-4 compels changes
       related to FA sensing and an improved lipid landscape.

1. Introduction

   During the last few decades, second-generation sequencing has revealed
   that the main body of the human genome is transcribed into molecules of
   ribonucleic acid (RNA) that do not code for proteins [[75]1,[76]2].
   Among these, ∼27% are long non-coding (lnc)RNAs with sequences scarcely
   conserved during evolution [[77]3] and mostly expressed in a
   tissue-specific manner [[78][4], [79][5], [80][6]]. Accounting for over
   68% of bulk human transcriptomes [[81]7], the widespread activity of
   lncRNAs hinges on their interaction with DNA, RNA and/or proteins
   [[82]8]. Through these interactions, lncRNAs have emerged as scaffold
   components of nuclear architecture [[83]9], regulators of chromosome
   conformation and transcription factor recruitment [[84]10], and
   modifiers of messenger (m)RNA synthesis and stability, as well as its
   final translation into functional proteins [[85]11,[86]12]. Located
   within the nucleus, or contained in the cytoplasm of a large variety of
   mammalian cells, lncRNAs perform a broad range of functions in
   differentiation and development [[87]13], also contributing to the
   burden of disease [[88]14,[89]15]. On this subject, the most recent
   research has shown how important the epigenetic regulation exercised by
   lncRNAs can be for adipocyte biology and function [[90]16], also
   participating in the pathophysiology of obesity [[91]17,[92]18]. As a
   result, it is now recognized that deranged lncRNA expression patterns
   in adipose tissue may engage a range of metabolic disturbances
   associated with an obese phenotype [[93]19]. However, identification
   and functional evaluation of specific adipose-derived lncRNAs related
   to obesity and its clinical sequelae are still far from complete. The
   current study stems from the analysis of bulk transcriptomes of human
   adipose tissue before and after weight loss, which allowed the
   identification of linc-GALNTL6-4, a novel adipocyte-specific lncRNA
   only found in humans and showing expression levels steeply increased
   upon the loss of fat. Scrutiny of additional fat samples further
   confirmed that this unique lncRNA opposes body weight, is solely
   present in the adipocyte cell fraction of adipose tissue, and is
   increased during the course of adipocyte differentiation. Observations
   made in human patient cohorts and functional experiments carried out on
   cultured adipocytes suggest that linc-GALNTL6-4 regulates the
   commitment of adipocytes towards the control of APOC1 levels and
   appropriate lipid patterns, potentially entangling adipose tissue
   function in metabolic homeostasis and dyslipidaemia.

2. Results

2.1. Expression of linc-GALNTL6-4 in adipose tissue is inverse to obesity

   To identify transcripts correlating with adipose tissue function,
   expression profiling of 16 severe obese women undergoing bariatric
   surgery was performed. Anthropometric and biochemical phenotyping of
   these participants is shown in [94]Table S1 and has been reported
   previously [[95]20]. Microarray analysis depicted a total of 5,018
   transcripts differentially expressed (DE) when comparing subcutaneous
   (SC) adipose tissue at the baseline and ∼2 years after surgery-induced
   weight loss (adjusted false discovery rate (FDR) p-value<0.05)
   [[96]20]. Among these, 496 transcripts were defined as putative
   lncRNAs, after excluding pseudogenes as well as spurious and
   uncharacterized ncRNAs, based on manual annotation [[97]21]. Two
   hundred and eighty-two (56.9%) of the lncRNAs with dynamic changes were
   upregulated, while expression of 214 candidates decreased upon the loss
   of weight ([98]Figure 1A). Only two lncRNAs displayed an average
   fold-change > [[99]2] and adjusted FDR p-value<0.0001: linc-GALNTL6-4
   (also known as LINC01612, RP11-789C1.1, XLOC_003775, and
   TCONS_00008319), and linc-NUDT10 (lnc-BMP15-1:1, XLOC_007982,
   TCONS_00017171, CATG00000111229.1, and ASMER-2 [[100]22]). Expression
   of linc-NUDT10 was diminished, whereas linc-GALNTL6-4 values in paired
   pre-post surgery samples displayed a notable increase ([101]Figure 1B).
   While the decay of adipose-derived linc-NUDT10 in obese patients upon
   weight loss was documented in 2018 by Zhang and co-workers [[102]23],
   we were struck by the unprecedented regulation of linc-GALNTL6-4, and
   hence we focused our research on this lncRNA. First, we confirmed the
   hitherto overlooked expression of linc-GALNTL6-4 in human adipose
   tissue. Results collected from the Genotype-Tissue Expression (GTEx)
   Analysis Release V8 (dbGaP Accession phs000424.v8.p2) showed consistent
   expression of linc-GALNTL6-4 in SC (n = 663) and visceral (great
   omentum; n = 541) adipose tissue, ranking up to 9 and 20 transcripts
   per million (TPM), respectively ([103]Fig. S1a). Additionally,
   co-expression network analysis of linc-GALNTL6-4 and other RNAs of
   reference [[104]20] pointed that inter-individual variations affecting
   this lncRNA were associated with concomitant fluctuations in 181 mRNAs
   ([105]Table S2). Interpretation of the available annotations by
   Ingenuity Pathways Analysis (IPA) and g:Profiler highlighted the
   enrichment of genes related to energy handling, including fatty acid
   and carbohydrate metabolism ([106]Table S3). Then, we did real time PCR
   assays in an extended sample of 23 obese patients following
   surgery-induced weight loss, including those of the discovery cohort
   ([107]Table S1). In agreement with our microarray results, expression
   of SC linc-GALNTL6-4 rose (3.8-fold change) in fat samples obtained ∼2
   years after gastric bypass ([108]Figure 1C). To better substantiate the
   apparent regulation of adipose linc-GALNTL6-4, Robust Multichip Average
   (RMA) expression measures were retrieved from reference [[109]24].
   There, biopsies of SC adipose tissue were obtained by needle aspiration
   from 50 obese women at the baseline and 2 (n = 49) and 5 (n = 38) years
   after gastric bypass, as well as in a nonoperated group of 28 healthy
   weight women. Also in this independent dataset ([110]Figure 1D), the
   utility of linc-GALNTL6-4 as a biomarker of dynamic adaptations
   affecting adipose tissue function was outlined by the increased
   expression found in age-matched non-obese subjects (1.24-fold) and
   obese patients following weight loss (1.16- and 1.28-fold change at 2
   and 5 years after surgery, respectively). We queried next the abundance
   of linc-GALNTL6-4 in SC and omental (OM) fat samples of an independent
   cohort of individuals with and without obesity (body mass index (BMI)
   threshold of 30 kg/m^2) ([111]Table S4). In concordance with our
   longitudinal findings, cross-sectional comparisons confirmed the
   partial loss of SC and OM adipose linc-GALNTL6-4 in obese subjects
   ([112]Figure 1E). Notably, linc-GALNTL6-4 levels, which were
   consistently higher in OM than in SC fat depots ([113]Figure 1E), were
   inversely correlated with BMI ([114]Figure 1F) and fasting glycaemia,
   and also opposite to the expression of leptin (LEP) and tumor necrosis
   factor alpha (TNFα), while running together with biomarkers of
   adipocyte function such as FASN, ACACA, IRS1, and GLUT4
   ([115]Table S5). Additionally, multiple linear regression analyses
   highlighted that BMI greatly contributed to the variance of SC and OM
   linc-GALNTL6-4, after controlling for gender, age and fasting glucose
   ([116]Table S5). For a replication of these cross-sectional
   observations, we employed RNA sequencing (RNA-seq) data from the
   subcutaneous fat of 335 male subjects ([117]Table S6) in the Finnish
   METSIM cohort [[118][25], [119][26], [120][27]]. Here, the expression
   of linc-GALNTL6-4 showed a significant inverse trend with BMI
   ([121]Figure 1G) and % fat mass ([122]Figure 1H), lending further,
   independent validation to the detected association of impaired
   linc-GALNTL6-4 expression levels with obesity and biomarkers of
   adipocyte function ([123]Table S7). On the basis of these lines of
   evidence, we hypothesized that, in human adipose tissue, the loss of
   linc-GALNTL6-4 might be relevant for energy handling and
   obesity-associated adipocyte dysfunction, and thus play a role in the
   set of metabolic disturbances that string along with obesity.

Figure 1.

   [124]Figure 1
   [125]Open in a new tab

   linc-GALNTL6-4 in adipose tissue is opposite to obesity. (A) Volcano
   plot of upregulated (red) and downregulated (green) lncRNAs in SC
   adipose tissue after weight loss (n = 16), as per inclusion criteria
   noted in Methods, and listed in (B). Dynamic adaptations affecting
   adipose linc-GALNTL6-4 after weight loss were confirmed by means of (C)
   real time PCR (n = 23), and in (D) an independent dataset of 50
   participants (microarray) [[126]24]. linc-GALNTL6-4 levels are provided
   as box plots of 75th to 25th percentiles with the median, and whiskers
   at maximum and minimum values. (E) Mean and S.D. for linc-GALNTL6-4
   values in SC (for obese and non-obese (BMI<30 kg/m^2) subjects, n = 73
   and 136, respectively) and OM (n = 59 and 111) fat samples from an
   independent cohort of 212 participants, and in (F) association with
   BMI. ∗∗p < 0.001 for comparisons between groups of subjects segregated
   according to their BMI, and ^$p < 0.05 and ^$$p < 0.001 for comparisons
   of OM vs SC. Coloured numbers depict Spearman's rank (r) correlations
   for men (green and blue) and women (red and purple), and ∗∗p < 0.001.
   Continuous and staggered black lines show regression and 95% confidence
   interval. Plots (G) and (H) depict linc-GALNTL6-4 values in the METSIM
   study, a population-based study including bulk RNA-seq in the SC fat of
   335 Finnish men, in association with BMI and fat mass, respectively.
   (For interpretation of the references to color in this figure legend,