We also allow for barcode sequences of 17—23 base pairs instead of exactly First, for all barcodes sequenced, the number of reads mapping to any genomic location and the number of reads mapping to concatemeric junctions are tabulated. Any barcodes for which the majority of TnSeq reads map ambiguously to the genome are removed and reported as ambiguous barcodes.
The remaining unfiltered barcodes are reported as the mutant pool. The script tracks number of insertions versus scaffold length for all scaffolds in the genome, GC content in the local regions of insertion, and insertion rates in promoter regions, 5-prime untranslated mRNA, exons, introns, 3-prime untranslated mRNA, and terminator regions. To assess fine-scale biases in insertion locations, all locations in the genome are apportioned to one of the above feature types, then for each feature type, the same number of insertions as were observed for that feature type in the mutant pool are sampled at random without replacement from all the genomic locations assigned to that feature type.
We used Q5 high-fidelity polymerase with GC-enhancer NEB, MS to amplify unique barcode sequences flanked by specific priming sites, yielding a bp Illumina-sequencing-ready product Figure 1—figure supplement 1. We used BarSeq primers from Wetmore et al. We quantified product yield with a Qubit 3. Coates Genomics Sequencing Laboratory. We sequenced each biological replicate to a depth of at least 20 million reads. Cultures were pelleted at 3, RCF for 5 min, washed twice in the appropriate media, and transferred to the condition of interest.
Samples were taken from the YPD starter cultures Time 0 and after 5—7 doublings in the experimental condition. Average fitness scores and T-like statistics as metrics for consistency between individual insertion mutants in each gene were calculated with the scripts combineBarSeq. R from Wetmore et al. These strain fitness scores are then normalized such that the median score is 0 to correct for coverage differences between the samples. The strain fitness scores are then assigned a weight proportional to the harmonic mean of counts at Time 0 and in the condition sample.
For any one barcode, the weighting mean is capped at 20 reads, which has the effect of limiting the influence of generally more abundant outlier strains Wetmore et al. T is calculated as the gene fitness divided by the square root of the variance in strain fitness scores. Because the list of genes satisfying this requirement can change from experiment to experiment, we established a list of genes that met this requirement in any of our experiments and used that list for our analysis. As a result, a minority of genes have fitness scores based on data from one or two barcodes. The number of barcodes used in fitness analysis of each gene is listed in all relevant tables in Supplementary file 2.
In general, genes with data from only one or two barcodes had smaller T-statistics and thus were filtered out in later analyses. Because Wetmore et al. We generated K-means clusters of fitness scores using Pearson correlation as the similarity metric using Cluster 3. We generated hierarchical clusters of enrichment scores using Pearson correlation as the similarity metric and average linkage as the clustering method.
We transferred the pellets to 1. We electroporated cells at 1. We then added 1 mL cold mixture of YPD and 0. We scored enrichment of gene ontology terms with a custom script that performs a hypergeometric test on the frequency of each term in the genome versus the frequency in given gene set script GOenrich.
We corrected for multiple hypothesis testing with the Benjamini-Hochberg correction Benjamini and Hochberg, We extended the GO terms associated with R. Cell lysis, extraction of total lipids, and conversion to fatty acid methyl esters FAMEs was based on a published protocol Browse et al. We then transferred the mass determination sample to a pre-tared 1.
Cell lysis and conversion to FAMEs occurs during this incubation. FAME concentrations were calculated by comparing the peak areas in the samples to the peak areas of ten commercially available high-purity standards C, C, C, C, C, C, C, C, C, C Sigma-Aldrich in known concentration relative to the internal standard, respectively. Due to logistical constraints, samples were processed in batches of at most 30 cultures at a time. Distribution of mutant strains into these batches was not explicitly randomized, but each batch included both strains expected to accumulate more lipid and strains expected to accumulate less lipid than the parent.
Each mutant was processed in at least two different batches.
We cultured the barcoded mutant pool in low nitrogen medium for 40 hr. We then diluted unfixed cells in 10 ml PBS with 0. We sorted a sample of 10 million cells with the scattering gate alone as a control population, to account for effects of growth, sorting, and collection that are independent of lipid accumulation.
We collected 10 million cells each for the high and low signal populations. We prepared linear sucrose gradients with the method of Luthe et al. Luthe, We selected appropriate gradients to maximize the physical separation of the cell population by running trial experiments with wild type IFO cultures on a number of sucrose gradients. The gradients used in each experiment are described in Table 3. Cover slips were submerged in 0. Cover slips were removed from polylysine and blotted dry from the bottom of vertically-held slips.
Slips were then washed several times with ddH 2 O and rapidly dried with compressed air. Directly prior to imaging, slips were visually inspected for streaks and dust and softly cleaned with lens paper. Cells were grown 40 hr in low nitrogen medium, 1 mL of culture was transferred to 2 mL microcentrifuge tubes with 1 mL of PBS, and tubes were mixed briefly by vortexing. Zvi files were converted to 16 bit TIFF images and representative fields of view were cropped and channels merged using FIJI image processing software Schindelin et al.
An effective functional genomics approach requires high quality genomic sequence and reliable gene models. To improve assembly, we added long-read sequencing from Pacific Biosciences to our previously published data from Illumina sequencing Zhang et al. Seven de novo scaffolds have telomeric repeats Ramirez et al.
- Lipid Metabolism.
- Services on Demand.
- LIPID METABOLISM.
For comparison, electrophoretic karyotyping of R. We also used bp paired-end Illumina sequencing of mRNA to improve gene model prediction. We have manually corrected several hundred fused models not supported by RNAseq data and encourage the R. We adapted a high-throughput phenotyping strategy previously demonstrated in bacteria Wetmore et al. Briefly, we created a large barcoded mutant pool in which A. From a mutant pool of approximately two million R. The remainder of sequenced barcodes could not be mapped for several reasons Figure 1—figure supplement 2A.
T-DNA is often inserted in concatemeric repeats Kunitake et al. T-DNA can integrate into multiple locations in the same genome, giving rise to confounding phenotypes between different mutations. To estimate the frequency of multiple insertion events from co-transformation, we isolated single colonies and then sequenced their barcodes using PCR amplification with Sanger sequencing. This estimate may be biased by sequence artifacts and should be taken as an upper bound.
These combined sources of confounding phenotypes highlight the importance of integrating data from multiple T-DNA insertions in any fitness analysis. As such, our main concern in constructing our mutant pool was to effectively probe the entire genome with multiple inserts per gene. On a genome level, there was no significant bias in rates of T-DNA insertion, with insertion number proportional to scaffold length Figure 1—figure supplement 3A and no apparent bias in insertion rates with respect to local GC content Figure 1—figure supplement 3B.
We did observe some bias in T-DNA insertion sites at the kilobase scale, however. T-DNAs were mapped within intergenic regions at a higher rate than expected given the composition of the genome Figure 1—figure supplement 3C. This bias towards promoter regions is consistent with observations in Cryptococcus neoformans Walton et al. We observed a wide range in relative abundance of individual mutant strains i. In a typical fitness experiment, we sequenced each sample to a depth of 20 million reads as opposed to million reads to map insertion locations by RB-TDNAseq.
Countable barcodes ranged from 1 to 1, counts per sample with a mode around 10 Figure 2—figure supplement 1A. This distribution of countable barcodes and sequencing depth translated to approximately — total usable reads per sample for estimating fitness of mutants in each gene Figure 2—figure supplement 1C. This depth of usable sequence on a per-gene basis is comparable to the point at which Wetmore et al.
Examples of how fitness scores from individual barcoded insertion strains are combined to calculate gene-level fitness scores are shown in Figure 2—figure supplement 2. Though raw counts varied over two orders of magnitude for these strains Figure 2—figure supplement 2A , strain fitness scores in non-supplemented media were similar across the entire length of the gene model, including introns, with a sharp change in fitness scores at the gene boundaries Figure 2—figure supplement 2B.
In supplemented media DOC , counts for all insertions are similar to the Time 0 samples Figure 2—figure supplement 2C. To calculate F and T for the gene, first fitness scores are averaged between insertions, with more abundant insertions weighted more than less abundant ones relative weights in this average are indicated by shading of the fitness scores in panel B. Intuitively, strains with more counts will tend to have scores less affected by sequencing noise or stochastic representation of cells in small samples from a given culture.
Because the weight is based on the harmonic mean of counts in Time 0 and the condition sample, strains with very low counts in one condition and thus more noisy log 2 ratios will be weighted lower than those with log 2 ratios with less noise in both the numerator and denominator. The effect of rare, high-abundance strains, which tend to have discordant fitness scores Wetmore et al. T is calculated as a moderated T-statistic: the gene fitness F is divided by the square root of a conservative estimate of variance in F amongst strains. These statistics are first computed for each biological replicate separately, then F is averaged across replicates and T is combined as a true T-statistic between independent experiments, by summing T and dividing by the square root of the number of replicates.
Visual inspection of the raw reads for these barcodes at Time 0 and after growth on non-supplemented media suggests that mutants in MET16 are deficient for growth in non-supplemented conditions, but two of the barcodes behave differently than the others; they show no significant reduction in counts in the non-supplemented conditions.
In particular, a single barcode that mapped near the three prime end of the gene has relatively high counts and fitness scores around zero, weakening the overall fitness score and yielding a marginal T-statistic. In this particular case, restricting the region of the gene used in the analysis would improve F and T and better capture the true phenotype of the mutants in this gene. However, any further global restriction on the barcodes analyzed per gene would compromise the data for other genes with fewer insertions.
In most cases, meaningful interpretation of fitness scores will require comparisons amongst several conditions. If the conditions of interest are tested in the same experiment with replicates inoculated from the same Time 0 samples, then direct comparison of the BarSeq counts between conditions would be the most straightforward and statistically powerful approach to make binary comparisons between conditions. That approach is difficult to implement across a large panel of conditions, however, and it is not amenable to an incremental inclusion of new experiments into an existing database.
Therefore to compare fitness between conditions we first calculate F and T versus Time 0 in all conditions, and then compare F and T directly between conditions. We calculate relative fitness by taking the difference in fitness scores in the two conditions. We then calculate T between conditions by estimating the variance we would observe in the direct comparison by adding the variance observed in both original comparisons to Time 0.
As each of those variances was a conservative estimation, this approach tends to inflate variance and conservatively bias the relative T-statistic. Thus working directly with F and T versus Time 0 to compare conditions is the most flexible, but often not the most sensitive approach.
Regardless of how a direct comparison is calculated, it is only interpretable if cultures are grown the same number of generations in both conditions. If samples have been grown a different number of generations, the best approach to synthesize data across multiple conditions is to subject fitness scores to clustering analysis as in Figure 2C. An important assumption in our analysis is that T, our metric of consistency, should assume the standard normal distribution when samples with no true abundance differences are compared. We analyzed the Time 0 replicates from our initial auxotrophy experiment treating two replicates as mock samples from an experimental condition and found that the resulting distribution of T-statistics was indeed very similar to the normal distribution, albeit with a small number of outliers at the tails Figure 2—figure supplement 3A.
This minor deviation was largely eliminated when we analyzed a mock experiment with biological replication by shuffling Time 0 samples between three independent experiments Figure 2—figure supplement 3B. This bias is reflective of the tendency of smaller genes to receive fewer insertions in a randomly mutagenized mutant pool. It is unclear if this bias towards higher GC content is the result of a technical artifact, a reflection of lower GC content of pseudogenes and other mis-annotated features, or evidence that genes with less impact on fitness in most conditions are under less stringent selection for high GC content in R.
Fitness scores are less sensitive to gene length than T-statistics Figure 2—figure supplement 3G. Our fitness data were consistent with established models of arginine biosynthesis in S. Discordant fitness scores between individual mutant strains drove the relatively weak T-statistics for ARG8 and MET16 see Figure 2—figure supplement 2 for a full breakdown of strain fitness scores contributing to the MET16 scores. Genes with very few insertions tend to have greater estimated variance in fitness scores and thus have T-statistics with smaller absolute values.
The fitness scores for these two insertions in MET8 were not inconsistent with its expected function in methionine synthesis, however. The mitochondrial ornithine transporter ORT1 was also required for arginine prototrophy, but AGC1 was not, suggesting alternative routes for glutamate transport. Cluster FA1 consists of 21 genes for which mutants had consistent growth defects across all three fatty acids. Clusters FA2 through FA7 were comprised of genes for which mutants had stronger fitness defects on one or two fatty acids, primarily genes with stronger defects on methylricinoleic acid and ricinoleic acids FA2 and FA3, 55 genes or on ricinoleic acid only FA4 and FA5, 30 genes.
These clusters were comprised of genes with predicted roles in various aspects of cellular homeostasis including amino acid metabolism, glycogen metabolism, phospholipid metabolism, protein glycosylation, the mitochondrial electron transport chain, and 17 genes with no well-characterized homologs. See Supplementary file 2 for a complete list. Clusters FA2 and FA7 also included 10 genes predicted to play direct roles in peroxisomal beta-oxidation, however.
Under carbon-replete growth conditions in which nitrogen, sulfur, or phosphorus are limiting, R. Because lipid droplets have lower density than most cell components, as cells accumulate large lipid droplets, they become more buoyant Figure 4—figure supplement 2. As we had no a priori estimate of how enrichment scores in FACS and buoyancy experiments would transfer to effect sizes on BODIPY signal in pure culture, we took a pragmatic approach to our validation experiments. We found that average BODIPY signal for a given strain could vary by as much as 2-fold between experiments on different days.
For each mutant we processed at least six biological replicates cultured in at least two different experiments cultured and processed on different days. A post-hoc power analysis of this strategy using published power analysis software Faul et al. Increasing sample size to 9 or 12 replicates would lower this detection threshold to 1. While the considerable variation we observed between cultures limits the sensitivity of our analysis, our candidate strains had sufficiently strong phenotypes such that 18 of our 21 strains from our high confidence clusters of enrichment scores nonetheless had significantly altered BODIPY signal.
The genes encoding calcineurin complex were also in this cluster as was another protein phosphatase and two protein kinases. Four genes with predicted roles in histone modification were included in cluster LA1 along with three transcription factors and the RNA splicing factor CBC2 , which is involved in mRNA processing and degradation Das et al.
Mutants in ten genes with likely roles in protein modification, protein trafficking or other processes in the ER and Golgi led to increased lipid accumulation Figure 6. These results show that protein trafficking plays an important role in lipid accumulation in R. Disruption of sulfur assimilation also increased lipid accumulation, with five genes involved in sulfate conversion to sulfide clustering in LA1.
In general, the sulfate assimilation mutants had reduced growth in low nitrogen conditions, as indicated by negative fitness scores for pre-enrichment control samples Supplementary file 2. As expected, the auxotrophic mutants identified in our supplementation experiments also had compromised growth in low nitrogen conditions, though the phenotype was generally less severe, likely reflective of slower growth of the population generally.
These data suggest that cysteine or intermediate sulfur compounds in the assimilation of sulfate to sulfide may be involved in regulation of lipid accumulation. We found evidence that tRNA thiolation plays a role in lipid accumulation in R. Enrichment scores for six genes known to be important in the thiolation of tRNA wobble residues Huang et al. Furthermore, we observed that for orthologs of S. Modification of tRNA wobble positions has been implicated in regulation of gene expression in response to heat shock Damon et al.
Our observations suggest that in R. The role that tRNA thiolation plays in this metabolic transition is unclear and deserves more detailed study. Efficient lipid accumulation also required the regulatory action of orthologs to the H. These genes are likely involved in signaling pathways mediating nutrient state. Mutants in nine core components of autophagy were deficient for lipid accumulation.
The vacuolar protease PRB1 and SIS1 chaperone mediating protein delivery to the proteasome were also required for efficient lipid accumulation, as were six genes implicated in protein ubiquitination Table 2. Ubiquitination can affect many aspects of gene function, but likely most of these genes participate in regulation of proteolysis. These results show that autophagy and recycling of cellular components are important for efficient lipid accumulation in R. While most genes encoding enzymatic steps in fatty acid and TAG biosynthesis had too few insertions to calculate reliable enrichment scores many are probable essential genes, see Supplementary file 1 , mutants in six genes with predicted function in TAG synthesis resulted in lower lipid accumulation see Figure 6—figure supplement 1.
The products of these genes were observed in proteomic analysis of R. Zhu et al. Neither gene had high confidence enrichment scores in our lipid accumulation assays. We mapped very low insertion density in the major enzymes of the pentose phosphate pathway the primary source for NADPH in Y. Our data are consistent with recent predictions from a simplified metabolic model for R. Long regarded as essentially inert spheres of lipid, eukaryotic lipid droplets have of late come to be recognized as complex, dynamic, organelles with unique proteomic content and regulated interaction with other organelles Walther and Farese, ; Farese and Walther, ; Gao and Goodman, In animal cells, seipin H.
We found evidence that the closest R. Perilipins H. Accordingly, we found that mutants for an R. Protein trafficking between the ER and Golgi has been implicated in lipid droplet accumulation in D. Increased lipid accumulation in mutants with defective COPII trafficking might also be a function of impaired protein quality control Fujita et al. These data are consistent with a hypothesis that interaction between protein sorting, quality control and the unfolded protein response play a role in regulating lipid accumulation through modulation of protein translation.
Alternatively, delivery of specific proteins to the lipid droplet via the vesicular trafficking system may be critical to lipid droplet growth and maintenance, or the effects of mutations in the endomembrane network on the lipid droplet may arise from redirection of carbon flux through membrane lipids. We identified 28 genes with high-confidence roles in lipid metabolism that are homologous to genes implicated in G protein—coupled kinase signaling cascades, including Rac, Ras and Rab family G proteins. Rab GTPases are implicated in several aspects of vesicular traffic Hutagalung and Novick, and are also thought to mediate droplet fusion and interaction with endosomes Gao and Goodman, Several Rab family members have been identified in lipid droplets in R.
Rac and Ras G proteins have diverse roles in regulating the actin cytoskeleton, cell proliferation, cell cycle progression and polarity Goitre et al. Likely both Rac1 and Ras1 interact directly with the lipid body, as Rac1 was detected in R. We were unable to quantify fitness scores for RHO1 , but that G protein was also found associated with lipid droplets in R. Undoubtedly these G proteins and downstream kinases function in a complex network of specific interactions, likely with considerable rearrangement of interactions from those observed in other species Choi et al.
Mapping these signaling networks in R. In mammalian and fungal cells, inhibition of autophagy has been reported to both decrease Rambold et al. These discrepancies may be reflective of competing roles in fatty acid mobilization from lipid droplets and lipid droplet biogenesis, with different processes dominating in different cell types and under different conditions.
Mechanisms of fatty acid mobilization have been proposed involving a macroautophagy-like process called lipophagy Singh et al. Why autophagy might be necessary for lipid droplet biogenesis is less clear, but autophagy-dependent recycling of membrane lipids to the lipid body has been demonstrated in mouse hepatocytes Rambold et al. In both Y. Further, in Y. Our findings demonstrated that autophagy was required for robust lipid accumulation in R. While we cannot rule out a more direct role in lipid droplet growth and maintenance, a simple theory for this requirement is that autophagy is required for extensive recycling of cellular resources during lipid accumulation.
Not only were several core components of autophagy necessary, but also the vacuolar proteases, and several proteins with predicted function in ubiquitination of proteins for proteosomal degradation. The methylcitrate cycle was required for robust lipid accumulation, which may be reflective of its proposed role in threonine recycling Luttik et al. How and why the role of autophagy in lipid droplet development varies by species and condition remains an open question, but R. We also noted that disruption of several amino acid biosynthesis genes, particularly genes involved in sulfate assimilation into cysteine led to increased lipid production.
These data are consistent with the repression of amino acid biosynthesis genes observed in R. Notably, mutants for genes involved in methionine biosynthesis but not required for sulfate assimilation did not have enrichment scores reflective of increased lipid accumulation, nor did several arginine biosynthesis genes, or other auxotrophic mutants such as insertions in PHA2 or ADE5.
- Data Networks!
- 2DArtist Issue 54 June 2010 issue 54.
- Lifestyle Modifications for Lipid Disorders | NYU Langone Health.
- ABC of occupational and environmental medicine.
- Physics of Precipitation: Proceedings of the Cloud Physics Conference, Woods Hole, Massachusetts, June 3-5, 1959.
These discrepancies suggest that the increased lipid accumulation observed for some mutants may not be simply attributable to redirection of carbon flux from amino acid biosynthesis, but might be the result of active regulation in response to specific amino acids or metabolic intermediates. The transcriptional and proteomic response during nitrogen limitation in these mutants warrants deeper study. Post-translational modification of tRNAs has long been known to be critical for efficient protein translation in general Agris et al. Both transcriptional Zinshteyn and Gilbert, and translational Laxman et al.
A commonality in these studies, however, is the altered expression of genes related to amino acid biosynthesis, protein expression and carbon metabolism. The dramatic metabolic changes entailed in lipid accumulation under nutrient limitation may make for an informative framework in which to explore the mechanisms by which tRNA thiolation interacts with cellular metabolism.
We noted extreme cell-to-cell variation in total lipid content in wild-type and mutant strains. This variation was evident in BODIPY fluorescence intensities that varied over at least an order of magnitude within any given sample Figure 4—figure supplement 2 and a wide range of lipid droplet sizes visible in microscopy images Figure 7. Extreme variation in lipid accumulation is typical across eukaryotes, and has emerged as a useful paradigm to explore phenotypic diversity within isogenic populations Gocze and Freeman, ; Herms et al. Our results indicate that R. In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses.
A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included. Thank you for submitting your article "Functional genomics of lipid metabolism in the oleaginous yeast Rhodosporidium toruloides " for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Naama Barkai as the Senior Editor.
The reviewers have opted to remain anonymous. Both reviewers were generally positive about the paper. They have also discussed the reviews with one another and agreed that the paper requires some substantial revisions as detailed in their reports. In this manuscript the authors adapted random barcoded random transposon insertion methods to the oleaginous yeast Rhodosporidium toruloides , enabling the functional inference into s - s of genes. Insert density was high 1 insert per bp and fairly even across the genome, though there were some biases regions of low insertion rates.
The authors were convincing in the power of this method, and inferred the function of putatively essential genes, including genes affecting lipid accumulation, of which 35 were validated through targeted deletion assays. The research appears to have been conducted with rigor and the manuscript is well written, figures are appropriate, etc.
I found this research quite exciting and do not see any flaws or major issues with this research. The only suggestions I have would be to shorten the text Discussion is six pages, for instance , as it is quite long and in some places repetitive. Also, there are many advantages to working with yeast.
The authors should discuss how well this approach would transfer to non-yeast fungi, which account for the vast diversity of Fungi and differ in biology, genome size and intron distribution. While the authors were able to dissect the lipid metabolism and importance of different processes e.
The paper entitled "Functional genomics lipid metabolism in the oleaginous yeast" represents a comprehensive random insertion mutagenesis effort in a relatively uncharacterized basidiomycete yeast. The successful construct of this library allows high resolution parallel genomic fitness analysis, whereby relative changes in mutant strain abundance in response to perturbation provide insight into gene function.
Oleaginous yeast are of high interest as they are an attractive host for the production of sustainable chemicals and diesel-like fuels. In addition, understanding the unique accumulation of triacylglycerides TAG observed in this organism may be of relevance to lipid storage disorders in human.
Here, re-annotation of the R. Before embarking on creating a random insertional mutation library, the authors first greatly improved the existing genome assembly using standard approaches including long read scaffolds and paired-end sequencing of messenger RNA to guide gene annotation. By modifying bar-coded transposon sequencing methodologies, a library of insertional mutations was constructed by taking advantage of established methods of Agrobacterium tumefasciens T-DNA mediated transformation to overcome the low transformation efficiency of R. All random mutagenesis techniques used for library construction suffer from insertional biases that can confound interpretation.
T-DNA insertions exhibit a preference for intergenic regions, can include multiple insertion events, and have been shown to cause local mutations and inversions among others. The authors then turned to applying their insertion library to identify genes required for fitness in various conditions. As a proof of principle, synthetic minimal media YNB was supplemented with arginine and, in a different experiment, methionine was used and at least in one case compared to YNB alone.
This is an ideal feasibility test, as any gene required for biosynthesis of these amino acids cannot grow in YNB alone, and therefore exhibit a maximum decrease in fitness. Though the authors did not address this issue, this experiment would be useful in defining the dynamic range and establishing the sensitivity of the assay.
Nearly all members required for arginine and methionine were identified, although there were a handful of exceptions. It seems that further study of the specific reason for these false negatives was a missed opportunity to learn about unforeseen issues inherent to the methodology. It is also noteworthy that the results were much less striking when the different fitness conditions were compared to the standard T 0 control, the metric of fitness that the authors argued as being the preferred method for accurate quantification of changes in fitness.
For this reason, why this analysis method was chosen over the other was not specifically addressed, a general finding echoed in other fitness experiments is throughout the manuscript. This concern is addressed more fully in the specific comments. Additional fitness tests included growth on three fatty acids as the sole carbon source; genes were identified with significant fitness defect scores that included genes involved in fatty acid oxidation, gluconeogenesis and mitochondrial amino acid metabolism.
Many of these genes were consistent with those known to be required for fatty acid oxidation in other species. To identify genes involved in lipid accumulation, the library was fractionated using two measures of lipid content — buoyancy separation and neutral-lipid staining. In this case, instead of using the initial library as the reference, genes involved in lipid accumulation were identified by comparing the abundance of each mutant in high and low lipid fraction and looking for strains that had the biggest differences measured in the two conditions compared to the control. Subsequent clustering of the data reveal enrichment in biological functions.
Select strains were again validated; in this case finding a greater number of inconsistencies between the two assays as well as a greater number of false positives. The authors therefore defined more stringent criteria to correct this; requiring gene mutations identified as exhibiting increased or decreased lipid content to be consistent between both the buoyancy and staining lipid content assays. The rationale behind using their genomic toolset to understand lipid metabolism and biogenesis in this particular fungus is clear, but the results need to be put in context of what has been observed in similar assays in the published literature i.
For example, the role of Sulphur metabolism in lipid biosynthesis and biogenesis would seem to be well served by a more extensive comparison to the data that has been collected and published with these other model systems. Indeed, it is unclear what, if anything, was specifically novel to R. Nonetheless, the morphological phenotypes observed were fascinating and would benefit from a more thorough treatment and a more in-depth comparison between these observations. Indeed, one of the key contributions represented by this thorough study is the power that it provides to compare and contrast these findings with findings observed in these other systems or by other methods.
In summary, the resource and library provided by this study will be extremely useful in further defining and characterizing the genetics and metabolic pathways unique to R. Importantly, the study supports other work suggesting that existence mitochondrial beta-oxidation is widespread in fungi. The involvement of both peroxisomal and mitochondrial beta-oxidation has been shown to be important in gluconeogenesis in mammalian cells and has also been linked to altered metabolism in cancer. Overall paper is well written and represents a significant contribution to the genomic analysis of another emerging model fungus.
However, there were several issues with the statistical analysis that must be addressed. First, the issue of sufficient coverage in counts per gene needs to be included and discussed as necessary. As mentioned, the author's state fitness was measured fore 68, representing genes. If the counts were evenly distributed, this should be sufficient for fitness measures.
We know that this is clearly not the case for sequencing reads as they are typically modeled using a noisy Poisson distribution or by a negative binomial. In the manuscript presented, neither the initial counts nor the final counts per gene are reported, not even including an example. This is a glaring omission: the authors state that the range of counts per gene varied broadly, with a mode of This seems too low for adequate gene coverage, particularly as the variance increases dramatically in this low count range, making significant fitness changes difficult to detect and accurately quantify.
Though the methodology used for the analysis was referenced PMID: , the assumptions made in this analysis were not discussed here.
Modification of Lipid Metabolism
Because the authors suggest that the use of the T 0 sample strengthens quantitative fitness measurements, yet the results in the supplementary files do not seem to reflect this, the analysis section needs to include this discussion. Overall, I found the statistical analysis difficult to follow and thinly presented — including details such as how the cells were grown.
As manuscripts become increasingly packed with massive amounts of data, these sections of the manuscript are critical in order for the reviewer to evaluate the data quality and robustness. F is the average of those ratios weighted by sequence depth for all the insertions disrupting a given gene. T is a modified student's T-statistic, a measure of statistical significance of F that incorporates consistency between individual insertions across biological replicate cultures. The supplementary data mentions several different metrics and it is difficult to know which is being used in the main text, or why they are all included to begin with.
For example in the auxotrophy experiments the results are presented in several different ways, the columns described by:. It is not clear which method of analysis was used on which dataset — every experiment should be annotated as such. This presentation of the data is especially confusing because fitness scores are weighted averages by sequencing depth across all insertions and then averaged to obtain a single score.
I can imagine all kinds of scenarios where this could be problematic. Problems when measuring the relative importance of different genes to each other may also arise for example, if shorter genes are penalized due to having fewer insertions, or may introduce bias due to unequal numbers of insertions associated with each gene and possibly influenced by variance as well. To avoid these issues it would seem necessary to use a metric that corrects or normalizes for these issues.
An additional concern as has already been mentioned is the higher confidence in measuring fitness relative to the T 0 condition; presumably due to the depth of sequencing and the ability to obtain associated 't-like' statistics. However, there seems to be some logic missing here. Although there is an accompanying website, it is clearly in its early stages and no key is provided for explanation of the metrics used in these files. In Supplementary file 2 multiple scores reported for each gene in each experiment.
However, each gene described as a weighted average of all of the insertions for a given gene. This raises the possibility that the assumption that genes with fewer than 2 inserts are essential — a key conclusion only briefly discussed — maybe misleading. This section needs to mention the logic and statistics behind this as well as to acknowledge previous work e. PMID: which used saturation transposition to identify essential genes. Another point that requires attention is why the variance and confidence in the counts is not addressed as it is in RNA seq prior to scoring log ratios by the average weighted counts for the insertions associated with each gene.
This paper relies on an advance more barcodes on the well-established barcoded Tn-Seq methodology pioneered by Adam Deutschbauer. The authors need to acknowledge the many Tn-Seq papers in many other model systems that have been successful in characterization of new genomes. Along these same lines, the authors mention that co-fitness analysis will accelerate annotation of new genomes.
This is likely correct, but the previously published concept of co-fitness should be elaborated upon and cited. Overall, the main difficulties in insertional mutagenesis in comparison to targeted gene deletions are that saturation is difficult as some regions are hotspots, while others are immune. This can be partially controlled by limiting insertion sites to those that interrupt coding regions.
If a gene-specific model is the intent for measuring fitness defects by relying on deep sequencing of the time zero sample, several specific issues need to be explained:. How many sequence reads are required for time zero? Shouldn't replicates of time zero be included? What is the supporting statistical test? What is the range of ratios and variance for F? To end on a more positive note, I do not doubt that the fundamental conclusions of the manuscript are correct.
Many if not most of the issues can be addressed by providing all of the data to the reader, preferably in a user-friendly format. However, the conclusions themselves did not obviously follow from presentation of the results, neither for the methodology or the biology. If the intent was instead to focus on novel biological findings using the technology, the biological findings seemed to rely heavily on homology and the presentation of complex model pathways was inappropriate without making clear in the model what is and is not known in other model organisms, or distinguishing what findings were novel and unique to R.
We have condensed the main text considerably in the revised draft. We have strong interest in leveraging these data to engineer greater titer, rate, and yield of lipidbased bioproducts. Such projects are beyond the scope of this work, which also has much broader applications and implications beyond metabolic engineering. Our purpose for these experiments was to establish the soundness of the mutant pool, sequencing protocols and analysis methods, and to establish useful statistical cutoffs to focus downstream analysis on high confidence sets of genes with minimal type I error.
We expect most readers with an interest in systems biology will be well aware of the inherent trade-off between sensitivity and specificity. To illustrate the various factors adding uncertainty to our fitness scores, we have added the following text to the Appendix detailing why three false negatives in this experiment had T-statistics that fell short of our consistency thresholds either too few barcodes, or barcodes with discordant data. We also added a clustering analysis of fitness data across supplementation conditions in Figure 2C, and moved figures summarizing fitness scores for genes in the predicted pathways from a supplemental figure to Figure 2D and E to more clearly show which genes in these pathways are identified with the two approaches.
Journal of Lipid Research :: About the JLR
The new figure is described in the Results:. It was not our intention to imply that comparing all samples to Time 0 first and then manipulating the resulting fitness scores and T-statistics is necessarily the most accurate and sensitive methodology. We, in fact, adopted this strategy out of a combination of practical considerations in experiment design and the intention to establish an analysis pipeline amenable to incrementally building large databases of fitness data. The flexibility that this approach affords comes with a price in the form of reduced sensitivity in some instances, as the reviewer correctly noted.
We have added some discussion of this tradeoff in the main text and the appendix. The appendix reports that the typical readsper barcodeused in fitness analysis was 10, not reads per gene. To help clarify this we have added a new figure Figure 2—figure supplement 1C showing the distribution of average counts per gene across our experiments and the following text in the:.
It should also be noted this is the number of reads per replicate.
Thus, with three biological replicates, approximately — reads are present in the Time 0 samples for comparison with the experimental conditions. At this read depth, variation between biological replicate cultures and between individual barcoded insertions in the same gene are a much greater source of uncertainty than sequencing noise.
We believe the experimental procedures were described adequately in the Materials and methods section, with several paragraphs devoted to culture conditions generally and procedures for the major experiments specifically. However, the above comments suggest more emphasis should be placed on an important detail of our experimental design: explicit pairing of all samples in a given condition with their own independent Time 0 starter cultures. To this end, we have amended our introduction of BarSeq analysis with auxotrophy experiments as follows:.
If the reviewers or editors are unclear about any other specific details of our experimental procedures we would be happy to address them. But we take the reviewers point that we may have given the details of the analysis somewhat short shrift. While we think it inappropriate to completely rehash all the details of the BarSeq analysis, we have added some additional detail in the Materials and methods section. We have also added a new section to the appendix contrasting two examples of how data from multiple insertions are combined for a gene fitness score: 1 A case of very coherent data from many insertions and 2 a case with fewer insertions with conflicting fitness scores.
This section is supported by a new supplemental figure Figure 2—figure supplement 2 , which displays the raw counts and fitness scores for the individual strains in each replicate. In addition to the individual examples described above we also note that Figure 3, Figure 3—figure supplement 1, and Figure 5 in the original submission all display gene fitness or enrichment scores for individual biological replicates demonstrating the consistency of these metrics.
Control Conditions averaged between replicates DOC vs. No Supplement Methionine vs. No Supplement Arginine vs. No Supplement YPD vs. No Supplement 3 T-like Statistics vs. Control Conditions DOC vs. For this reason we tried to be as transparent as possible and include all the BarSeq data we gathered in two accessible formats: as human readable, searchable, and programmatically accessible supplemental data tables; and as an interactive website, which has now been linked to the genome browser at the Joint Genome Institute. For readers simply trying to dive into the experiments as presented a bit more deeply, we can see that it might be a bit difficult to extract the relevant data from the comprehensive summary.
To make it easier to access the data for that purpose, we have included three new supplemental tables in Supplementary file 2, with the fitness scores that were clustered in Figure 2C, Figure 3—figure supplement 1, and Figure 5A. The supplementary tables also include the criteria by which these genes were selected e. We have also removed the Wilcoxon signed rank tests on the fitness scores from the manuscript. Indeed barcoded insertions in the same gene with differing phenotypes are common, as we would expect due to secondary mutations in the mutant pool, as discussed in the appendix.
The central problem of any such analysis is to aggregate the data from several mutant strains, while minimizing the error introduced by outlier strains to accurately infer gene function. As this is not a new problem, we adapted the proven methods of Wetmore et al. To more clearly frame this challenge and to explain the strengths of Wetmore et al. The T-like statistics are indeed somewhat sensitive to gene length.
Longer genes tend to have both more insertions and more total counts available for BarSeq analysis. Thus we have greater confidence in the metrics for longer genes. Note, however, that these biases result in a relatively small enrichment of longer genes amongst genes with consistent fitness scores see Figure 2—figure supplement 3C. In general, we use the T-statistics primarily to identify genes that have consistent, reliable fitness scores in a given condition and we subsequently focus on the fitness scores as the appropriate metric to compare effects between genes.
In our revised manuscript we have added more detail to the appendix discussing these biases in T-statistics and highlighting the preferred comparison of fitness scores as opposed to T-statistics to compare importance of different genes in a given condition:. The fitness browser will have increasing functionality as we add more data, particularly for analyzing co-fitness profiles between genes. The information on individual conditions is necessarily minimal, but of course the relevant conditions to this manuscript are more fully described here.
Once again we should reiterate that the fitness browser is cited as one of several ways to access the data, including the supplemental tables, the raw sequencing data at the NCBI short read archive and the analysis software on a third party software distribution service bitbucket. We have added information on the total number inserts mapped in each gene and the number of inserts with sufficient counts to contribute to fitness analysis at this sequencing depth to the relevant tables in Supplementary file 2.
We agree there is significant uncertainty in our provisional identification of essential genes. The list was generated from the straightforward observation that, on the gene level, insertion of TDNAs does appear to be close to truly random and the distribution of insertions per kilobase across genes reflects that, with the notable exception of a group of genes that also happen to be orthologous to known essential genes.
The presence of two distinct populations of genes in the histogram shown in Figure 2B is clear. We have edited the presentation of this data in the Results section to make this logic a bit more clear, and more clearly acknowledge some uncertainty in the designation of all these genes as essentials:. We have also edited the Discussion to more clearly highlight the recent use of transposon mutagenesis in a higher resolution application of the same principle in S.
While functionally equivalent to transposon mutagenesis in many respects, ATMT is more prone to introduction of more complex concatamers of the marker sequence and we find it useful to remind the audience of those differences. Also the use of TnSeq in describing an experiment that does not utilize transposons can cause some confusion. We have revised the Introduction to more clearly introduce the concept of TnSeq and distinguish the relative advantages of RB-TnSeq vs. The revised text includes a few more references to successful TnSeq studies and a comprehensive review.
We think it better frames the context in which RB-TDNAseq may enable more widespread application of functional genomics in fungi. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. We thank Morgan Price for his assistance and advice on TnSeq and BarSeq analysis, as well as for hosting our data on the fitness browser. This material is based upon work supported by the U. Preliminary work establishing genetic, culturing, and assay protocols with R. Department of Energy. The work conducted by the U.
The reaction begins when the two substrates, acetyl-CoA and malonyl-CoA, are transferred to the KS and ACP, respectively, which is catalyzed by acetyl and malonyl transacylases. The ACP is then free to accept another malonyl unit. The addition of two carbon units from malonyl-CoA to the growing acyl chain leads to the synthesis of palmitate which is released after being hydrolyzed by TE [ 3 ]. These promoters respond to glucose, insulin, thyroid hormone, catabolic hormones, and leptin [ 5 ].
Fasting inhibits ACC expression though re-feeding returns expression to normal levels. Activation and inhibition by insulin, catecholamines and glucagon, respectively, occur within minutes of exposure [ 5 ]. Acetyl-CoA carboxylase is also allosterically regulated, resulting in active and inactive protein conformations [ 1 , 5 ].
Phosphorylation of four or more serine residues on ACC results in inactivation [ 5 , 8 ]. Transcriptional regulation is the primary means of controlling FAS [ 1 , 7 ]. The FAS promoter has been studied in the rat, human, and chicken and the sequence is highly conserved among species [ 7 ]. Promoter activity and FAS expression have been shown to increase in transgenic mice when high carbohydrate diets are fed, after fasting, and with increased insulin and glucocorticoid levels. Palmitoyl-CoA and stearoyl-CoA are the primary substrates of the desaturation reaction and are converted to palmitoleoyl-CoA and oleoyl-CoA, respectively [ 10 , 11 ].
In liver, SCD is also required for synthesis of cholesteryl esters [ 9 ]. More double bonds can then be added by the elongation pathways discussed below [ 11 ]. Desaturation of 12 to 19 carbon fatty acyl-CoAs catalyzed by SCD-1, -2, -3, and -4 results in the addition of a cis -double bond between carbons nine and 10 and this reaction requires NADH, oxygen, NADH-cytochrome b 5 reductase and cytochrome b 5 [ 9 ]. Control of SCD-1, -2, -3, and -4 is mainly by transcriptional regulation [ 9 ].
Increased SCD-1 activity thus increases the conversion of saturated fatty acids to unsaturated fatty acids and changes the ratio of carcass fatty acids. These two carbon molecules can then enter the tricarboxylic acid cycle for energy production. The mechanism of the carnitine pathway is an ordered reaction where the binding of acyl-CoA begins the transport action [ 13 ].
Long chain fatty acids are converted to acyl-CoAs by acyl-CoA synthetase [ 14 ]. Acyl-CoAs are converted to acyl-carnitine molecules and transferred across the outer mitochondrial membrane by carnitine palmitoyltransferase-I CPT; EC 2. Carnitine palmitoyltransferase-II is located on the inner mitochondrial membrane and liberates the carnitine from the acylcarnitine after transfer across the inner mitochondrial membrane [ 15 ].
Two transmembrane domains anchor CPT-I to the outer mitochondrial membrane [ 13 ]. There are three isoforms of CPT-I [ 1 ]. The N-terminus of the enzyme, which is not required for catalytic activity, controls the response to malonyl-CoA [ 13 ]. The kinetics of inhibition by malonyl-CoA are responsive to temperature, pH, and lipids [ 13 , 14 ].
Fasting and glucagon increases CPT-I gene expression while hypothyroidism decreases expression by regulating the transcription level [ 13 ]. The insulin growth factor I receptor also controls CPT-I expression by mediating the inhibitory effects of insulin [ 13 , 14 ]. Sterol regulatory element binding proteins SREBP are helix loop helix proteins that are within the leucine zipper family of transcription factors [ 9 ].
Long-chain fatty acids are oxidized in the peroxisome by catalase, producing acetyl-CoA and hydrogen peroxide [ 15 ]. The catalase enzyme is induced by high-fat diets and proliferation of the peroxisomes is controlled by the peroxisome proliferator activated receptor PPAR , which is part of the nuclear receptor family [ 9 , 15 ].
Peroxisome proliferators also stimulate SCD-1 transcription levels [ 9 ]. Intracellular fatty acids contribute to the overall regulation of synthesis and oxidative pathways. Dietary triacylglycerol composition plays a major role in determining adipose tissue composition. Monogastric animals incorporate dietary fatty acids directly into tissue lipid deposits [ 21 , 22 ] and, therefore, to manipulate carcass lipid quality, it is important to understand the interactions of dietary lipids with carcass lipid.
Carcass fatty acid profiles closely mimic dietary fatty acid profile [ 21 , 23 ], and therefore, potential exists to modify carcass lipid properties i. One of the strongest determinants of carcass fat quality in pigs is the level and composition of lipids in the diet [ 24 ]. The impact of dietary lipids on carcass lipid may differ depending on the timing of feeding relative to growth and finishing, levels included in the diet, and interactions with other stressors.
Dietary triacylglycerols alter carcass lipid composition at the level of the fatty acid profile [ 21 ]. Mono-, di- and poly-unsaturated fatty acids have one, two, or many double bonds, respectively and as the level of unsaturation increases, the melting point decreases [ 21 ]. The ratio of saturated to unsaturated fatty acids is a way of describing the relative saturation of a fatty acid profile [ 21 ]. Iodine value, a measure of double bonds in a lipid, is a method used to composite characteristics of lipids in regard to fluidity [ 21 , 26 ]. Saturated to unsaturated ratios and iodine values can be utilized to describe the composition of lipids in both feedstuffs, total rations, and animal tissue.
Fat is commonly added in swine diets from 0. Because of the previously mentioned utilization efficiency and transfer coefficients, the level of saturation and iodine value of the feed lipid source will be strongly reflected in the carcass fatty acid profile and therefore, sources of dietary fats play a critical role in final carcass lipid quality.
Vegetable oils are typically high in linoleic acid, have an unsaturated to saturated fatty acid ratio of [ 22 ] and an iodine value greater than [ 21 ]. Diets high in these unsaturated vegetable oils will result in oily, soft carcass fat [ 21 ]. Conversely, tallow, which is high in palmitate and stearate, has a saturated to unsaturated fatty acid ratio of [ 22 ], an iodine value between 40 and 45 [ 21 ] and will result in firmer carcass fat when fed in the diet.
Due to differences in calculation of these indices, some variations in fatty acid profile are captured with one ration but not the other, as seen in Figure 1. For this reason, it is best to utilize both the IV and saturated:unsaturated indices when characterizing fat quality, in order to identify all variations in fatty acid profile.
Differences in carcass lipid quality alter final product characteristics. Higher iodine values IV are associated with fat that is softer, resulting in increased difficulty slicing and processing. Panel a is backfat with an IV of 69 which represents fat that is firm and maintains shape and structure, while panel b is backfat with an IV of 79 which represents fat that will lack the firmness required for processing.
Saturated to unsaturated fatty acid ratios are also used to characterize fatty acid profiles. While sausages bottom panels made from different animals have the same IV 59 , the differences in saturated:unsaturated fatty acids results in a higher quality, firmer product in panel c 0. Dried distillers grains with solubles DDGS is the by-product of yeast fermentation of grains such as corn for ethanol production [ 28 ]. During fermentation, corn starch is converted into alcohol and the remaining grain components, protein, fat, fiber, minerals, and vitamins are concentrated in the fermentative co-product approximately 3-times that of corn [ 28 ].
There are two processes by which ethanol can be extracted from corn, wet milling and dry grinding. Dry grinding yields the maximum ethanol from corn while wet milling yields other products including corn oil and corn gluten meal [ 31 , 32 ]. The dry grind process begins by grinding the corn and mixing it with water Figure 2. The resulting mash is then heated with enzymes to convert the starches to sugars which can be fermented by yeast.
Dry grind processing of corn to produce ethanol. Progression of processing steps are shown in ovals and gray arrows, with inputs and outputs indicated by black arrows. The major byproduct of ethanol production is dried distillers grains with solubles. The nutritional value of DDGS for pigs is influenced by the processing procedure and production plant equipment and techniques [ 33 , 34 ].
The nutrient profile of DDGS remains highly variable even within the same production site [ 30 , 35 ]. Two limiting factors for including DDGS in swine diets are the high level of unsaturation in the dietary fatty acid profile and the high fiber content [ 28 , 31 ]. As discussed above, the composition of these fat sources is important when considering the carcass fat firmness [ 21 ].
The high level of fat and fiber in DDGS have been shown to result in both decreased feed intake and increased unsaturated content of adipose tissue. The future direction of DDGS as a feed ingredient will likely be defined by the final use in global energy needs and not how it might be valued as a feed ingredient; that is, DDGS still contains a considerable amount of oil, a highly valued potential energy source. Today, DDGS is well suited for non-ruminants in terms of energy and protein content, price, and availability; however, the high linoleic acid content known to alter fat quality must be considered when determining dietary inclusions.
The levels of omega-3 and omega-6 fatty acids in the human diet are important for optimal health. Animals, including humans, lack the enzymes required to add double bonds between the methyl group and ninth carbon and therefore cannot synthesize omega-3 and -6 fatty acids, making these fatty acids essential in the diet [ 22 ].
The synthesis pathways of omega-3 and -6 fatty acids and the parallel omega-9 pathways are shown in Figure 3. Synthesis pathways for omega-3, -6, and -9 fatty acids in mammals. The ideal ratio of omega-6 to omega-3 fatty acids in human diets is between 4 to , although the average American diet is between 10 to [ 42 ].
The change in this ratio is due to the increase in omega-6 intake relative to the level of omega-3 fatty acids [ 41 ]. The need to increase dietary intake of omega-3 fatty acids, specifically EPA and DHA, has increased demand for products with a ratio of omega-6 to -3 fatty acids more closely related to American Heart Association AHA recommendations. The many health benefits of omega-3 fatty acids, such as lowering serum cholesterol and triacylglycerol concentrations, reduce platelet aggregation, reduction of blood pressure, and decreasing very-low-density and low-density lipoproteins, make dietary inclusion important [ 40 ].
The overall anti-inflammatory effects of omega-3 fatty acids have shown beneficial effects for arthritis and joint health in rats and humans [ 40 ]. Though it has not been directly studied in swine, omega-3 fatty acids could decrease the prevalence of lameness in sows if they result in the same joint and anti-inflammatory benefits noted in humans and rats. Lameness results in the removal of sows at a younger age than other culling reasons, thus decreasing breeding herd productivity [ 44 ]. Omega-6 fatty acids are the precursors of eicosanoids which include prostaglandins, thromboxanes and leukotrienes.
These metabolites of n-6 fatty acids exhibit inflammatory effects [ 45 ]. Omega-3 fatty acids inhibit eicosanoid synthesis by decreasing the available arachidonic acid available for eicosanoid production [ 18 , 45 ]. In addition to decreasing eicosanoid production, omega-3 fatty acids also decrease other inflammatory cytokines such as interleukin-1 and -6, and tumour necrosis factor [ 18 , 45 ]. Conjugated linoleic acids CLA are a group of polyunsaturated fatty acids that are positional and geometric isomers of linoleic acid C Because CLA and its precursor, trans vaccenic acid, are naturally produced during bacterial fermentation in the rumen of ruminant animals, the main sources of CLA in human nutrition are ruminant milk and meats [ 46 , 47 ].
The main isomers of CLA are cis -9, trans c9t11 and trans , cis t10c12; Figure 4. Though the main isomer produced by ruminants is c9t11, commercially available products commonly contain equal proportions of c9t11 and t10c12 [ 46 , 47 ]. Research in rodents, pigs, and humans has been conducted on the effects of CLA and has shown beneficial effects of CLA against obesity, cancer, atherosclerosis, and diabetes, some of which are isomer specific [ 46 , 47 , 48 ]. Structure of linoleic acid compared with cis-9, trans and trans, cis conjugated linoleic acid CLA.
In pigs, CLA inclusion in feed has resulted in decreased backfat thickness at finishing [ 50 , 51 ]. Overweight or obese humans supplemented with CLA for 12 weeks also demonstrated reduced body fat mass, although their body mass index remained unchanged [ 52 ]. Another noted effect of CLA is the inhibition of cancer, specifically, mammary, prostate, skin, colon, and stomach cancers [ 48 ].
The anti-carcinogenic effects of CLA have been mainly attributed to the c9t11 isomer [ 46 ]. Other studies of the same cell lines have not demonstrated these effects of CLA [ 48 ]. Atherosclerotic plaque formation is reduced by CLA [ 48 ]. Inclusion of 0. Effects of CLA on the onset of diabetes and insulin resistance are inconsistant. Rats fed CLA have shown significantly reduced fasting glucose, insulinemia, triglyceridemia, free fatty acids, and leptinemia [ 48 ].
Butter enriched with c9t11 CLA failed to reduce glucose tolerance, lower adipose tissue or enhance glucose uptake leading to the conclusion that perhaps it is the t10c12 isomer which is responsible for the antidiabetogenic responses [ 48 ]. Insulin tolerance testing on CLA-fed mice showed marked insulin resistance without changes to blood glucose concentrations after oral glucose tolerance testing [ 54 ]. Other studies have examined the reduction of plasma leptin by CLA and the concomitant changes in blood glucose level due to regulation by leptin [ 46 ]. The effects of feeding CLA to pigs have been evaluated in regard to fat quality [ 57 ].
Barrows fed CLA had improved feed efficiency, decreased backfat, and improved loin marbling and firmness when CLA was included at 0. When CLA was fed to genetically lean gilts for eight weeks, an increase in average daily gain and gain:feed was observed [ 59 ]. The same study also noted an increase in saturated fatty acids, decrease in unsaturated fatty acids, and an increased level of saturation of the belly tissue [ 59 ].
Several studies have shown that CLA feeding increases fatty acid saturation, and firmness in back fat and belly fat [ 60 , 61 , 62 ]. Additionally, use of CLA when feeding by-products may alleviate some or all of the negative impact on carcass quality. In barrows and gilts fed 0. Decreasing SCD-1 mRNA expression, and thereby decreasing the amount of saturated fatty acids being converted to unsaturated fatty acids, may be responsible for the increased levels of saturated fatty acids observed after feeding CLA [ 65 , 66 ].
Environmental stressors on pigs can impact lipid metabolism and overall carcass quality. Impacts of environmental stressors, including thermal stress and housing density, are through both direct effects of decreased growth efficiency and indirect effects of altered regulation of de novo lipogenesis. Managerial and nutritional strategies during critical growth periods may alleviate the impact of these environmental stressors.
Additionally, the regulation of de novo lipogenesis is influenced by the health status of the animal. Insults to health through disease or constant stress decrease feed intake and reduce de novo lipid synthesis. This decrease in de novo synthesis shifts the ratios of fatty acids in the adipose tissue to more unsaturated FA, further reducing lipid quality.
Decreasing space allocation reduces growth performance and the minimal spatial requirements for grow-finish pigs have been examined [ 36 , 71 ]. Housing densities between 0.
Related Modification of Lipid Metabolism
Copyright 2019 - All Right Reserved