Functional Genomics in the Study of Mind-Body Therapies

NEW TOOLS FOR ANCIENT PRACTICES

To uncover genomic determinants and molecular pathways involved in the changes brought about by MBT, researchers are turning to functional genomics, a developing field that encompasses genomics, epigenomics, transcriptomics, proteomics, and metabolomics. Since the completion of the first draft of the human genome in 2003, there has been exponential growth in understanding the genomic determinants of disease and identifying biomarkers for disease management. Through the measurement of thousands of genomic determinants, functional genomics technologies are able to generate hypotheses about disease processes and intervention effects at molecular levels. More recently, functional genomics has been applied to further our understanding of MBT. A 2013 review described 3 early trials,²⁴ but no review has tackled the full breadth of functional genomics approaches to MBT research.

Genomics targets an individual's entire complement of genetic material. This DNA blueprint is used to build proteins through the processes of transcription and translation (Figure 1). While an organism's DNA sequence remains essentially unchanged throughout its lifespan (with some exceptions—for example, the mutations that can cause a cell line to become cancerous), diverse processes control how the genes encoded by a cell's DNA sequence are expressed. The story of how genes switch on and off throughout time, from fractions of seconds to generations, is complicated. Work from 2012 and 2013 reveals this story to be even more complicated than once thought.^25,26

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Functional genomics consists of several subfields, defined by their examination of different phases of the cellular cycle. Double helix DNA carries the genetic code, coiling around histone proteins within chromosomes. Epigenomics analyzes modification of histone proteins as well as methylation of regions rich in cytosine-guanine pairings on the DNA strand itself. Transcriptomics involves assessing gene expression by sequencing free-floating messenger RNA strands. Proteomics explores expression levels by identifying and quantifying amounts of circulating proteins. Metabolomics assesses small molecule end-products of cellular processes to quantify the rate at which various processes occur. (Source: Darryl Leja, of the National Human Genome Research Institute. www.genome.gov. Accessed September 23, 2014.)

A broad array of processes modulates gene expression; a subset of these processes is the group of molecular mechanisms known collectively as epigenetics. Epigenetics includes stable, long-term changes in gene expression through DNA methylation, histone modification, and chromatin remodeling, some of which offspring can inherit.^27,28 By definition, epigenetic mechanisms do not involve alteration of the DNA sequence (ie, do not involve DNA mutation). The sum of active epigenetic mechanisms in a given cell comprises the epigenome; these mechanisms effectively activate and deactivate sections of the genome, enabling or preventing their transcription into RNA and subsequent translation into proteins. Histone protein modifications change the way the DNA strand is wrapped around the nucleosome, thus altering which regions are accessible for transcription. While many histone modifications have been reported in the literature (eg, lysine acetylation, lysine methylation, and serine phosphorylation), the functional meaning of these changes is not well understood.²⁹ DNA methylation involves the addition of a methyl group to the cytosine in a cytosine-guanine sequence (CpG). High-precision technologies are now able to analyze the methylation status per base level. While the majority of these pairings are scattered throughout the genome, approximately 10% of human CpGs cluster in distinct groupings known as CpG islands. Methylation of these CpG islands is believed to be responsible for long-term heritable changes as well as the process of cell lineage differentiation.³⁰ However, epigenomic comparison across tissue types is challenging because methylation patterns vary among different cell types.

Transcriptomics, or gene expression profiling, is the quantification of messenger RNA (mRNA) in specific cell or tissue types. While the genome itself is essentially stable for a given cell line, the transcriptome is constantly in flux, responding to external and internal conditions, and is thus a good indicator of gene activity and regulation. Several commonly used techniques for measuring gene expression activity are DNA microarrays, multigene transcriptional profiling, and next-generation sequencing technologies.³¹ DNA microarrays allow for measurement of tens of thousands of genes in a single assay and are useful for generating hypotheses about intervention mechanisms and for understanding the pathophysiology of diseases, including the gene expression signature of treatments or disease states. Microarrays provide limited quantitative data on gene expression, and on a single gene level they are limited in accuracy and reproducibility. Multigene transcriptional profiling, usually by real-time polymerase chain reaction (RT-PCR), lacks the breadth of microarrays but allows for the quantitative measurement of expression activity in a limited set of a priori identified genes. Next-generation sequencing technologies, such as RNA-Seq, provide analysis of the complete transcriptome, including noncoding microRNAs that may regulate expression. Unlike microarray-based technology, RNA-Seq is not limited to detecting transcripts corresponding to existing known genomic sequences, making RNA-Seq particularly adapted to the study of complex transcriptomics. However, because of the challenging and highly computationally intensive data analysis associated with RNA-Seq, microarrays remain the analysis method of choice in transcriptomics.³²

Transcriptional profiling measures snapshots of gene activity at a particular point in time and therefore has the potential to identify rapid gene expression changes resulting from mind-body practices. However, this technique often lacks consistent findings, in part because of the diversity of expression across tissue types and the technical noise of the assay. For peripheral-blood studies, in which changes in the differential composition of leukocyte types may drive gene expression changes more than gene regulation within specific cell types, there is often insufficient discussion of the cell types involved in transcriptomic changes.³³ To handle these biases, some researchers recommend analysis on the level of gene ontology characteristics, as general trends in clusters of genes should provide a stable marker of expression that is robust to variation between trials.³³

Proteomics is the quantification of protein in a given tissue sample, combining a technique to separate proteins (eg, 2-dimensional gel electrophoresis) with techniques to identify and quantify those proteins (eg, matrix-assisted laser desorption/ionization mass spectrometry). Proteomics is the next step in the study of biological systems after epigenomics and transcriptomics. As in transcriptomics, the proteome varies from cell to cell and from time to time because distinct genes are expressed in distinct cell types. Functional proteomics is a subfield that seeks to elucidate the specific cellular function of unidentified or novel proteins. Current proteomic methods are limited in scope; by some estimates, current methods are able to sample only a small portion of the available core proteome, consisting only of the most highly expressed proteins.³²

Metabolomics is the study of the terminal downstream products of the genome, mainly small molecules and other low-molecular-weight end products of cellular processes. While the scope of metabolomic methods such as mass spectrometry and high-resolution nuclear magnetic resonance spectrometry is limited to a subset of the available small molecules in the cellular milieu, these analyses can be paired with quantitative measurements of small-molecule metabolites to provide a fuller picture of cellular activity.

In general, functional genomics approaches can be used in 2 ways. Omics approaches, because of their ability to look at the entire epigenome, transcriptome, proteome, or metabolome, are well suited for asking questions that are hypothesis generating, making them useful for exploring new areas. To make sense of the large quantities of data generated by omics methods, bioinformatics and systems biology approaches are frequently used to identify genes clustered into specific pathways, often centered around focus genes or hubs.³⁴ Using gene ontology groupings, it is also possible to ask targeted questions with omics data by analyzing data based on clusters of functionally related genes. Targeted studies, on the other hand, focus on the expression of specific genes or downstream markers. Because of the relative statistical power granted by the limited number of target genes or biomarkers, these targeted methods can provide more reliable quantitative data on specific differential expression than omics approaches. Targeted methods have a limited ability to detect patterns not contained within a priori hypotheses and can risk confirming false hypotheses by missing unmeasured mediators of effects.

MBT STUDIES IN THE LITERATURE

To assess the research applying functional genomics to MBTs, we searched MEDLINE for terms describing MBT (yoga, meditation, tai chi, stress management, stress reduction, and mind-body) and functional genomics (epigenetic, transcriptomic, gene expression, proteomic, and metabolomic). This search yielded 105 papers published between April 1999 and May 2014. Of these, 14 studies reported results of functional genomics analysis of humans practicing MBTs. One trial³⁵ did not appear in the initial search results but was included because of repeated mention in multiple reviews appearing in the results, bringing the total number of studies to 15. The studies included for analysis are described in the Table.

Study Designs

In the application of functional genomics to MBTs, the studies employed diverse designs and methods. Of the 15 trials, 4 were exclusively cross-sectional, comparing practitioners experienced in a given MBT to matched control subjects. Five trials used standard interventional designs with individuals or groups, assessing genomics outcomes before and after training in a specific MBT. Three studies solely employed a rapid-response method to assess changes in response to MBT by comparing samples from experienced practitioners before and after practice. Last, 3 studies used multiple methods: 1 was cross-sectional and interventional, 1 was cross-sectional and rapid-response, and 1 used all 3 methods. In the 2 studies using both cross-sectional and interventional methods, researchers compared naive practitioners to themselves and experienced practitioners. Of the 5 interventional-only studies, 3 used active control interventions, 1 used a waitlist control, and 1 had no controls, employing a quasiexperimental pre/post design. Figure 2 shows a detailed flow of study design types.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Trials included in this review by study design type, use of controls, and time course of effects measured.

COMMON FINDINGS

The majority of published research applying functional genomics methods to MBT is in the field of transcriptomics, supporting the correlation of transcriptomic, and thus epigenetic, changes with MBT. Despite the heterogeneity of the reviewed studies, the study results link MBTs to changes in cellular processes of inflammation, apoptosis, ubiquitin-dependent protein catabolism, and telomere regulation.

PARALLELS AND POTENTIAL TARGETS

To better understand genomic commonalities among these diverse practices, it is important to draw comparisons of functional genomics changes brought about by MBTs to other practices—for example, exercise and massage—that share some characteristics and produce similar changes in gene expression. On the transcriptome level, exercise interventions show some similarities with MBTs. A 2011 trial of resistance exercise indicated differential expression of the NF-κB signaling pathway,⁶⁹ and a 2-month aerobic exercise intervention reduced age-related upregulation of NF-κB and the related cytokine TNFα.⁷⁰ However, rapid-response studies of aerobic and anaerobic exercise reveal upregulation of inflammation-related gene pathways,^71-75 indicating a paradoxical response with short-term upregulation of inflammatory pathways leading to long-term downregulation. This response stands in contrast to the progressive pattern of rapid-response signaling seen in MBTs⁴⁸ that correlates positively to long-term effects. Denham et al found that multimonth exercise interventions significantly altered epigenetic mechanisms in peripheral blood leukocytes,⁷⁶ although whether parallel changes occur with MBTs remains to be determined, as the only MBT study to analyze methylation featured tai chi, a form of light exercise.³⁹

Massage therapy has been demonstrated to reduce psychological stress⁷⁷ and to influence immune function^78,79—both effects commonly seen with MBTs—and may bear genomic similarities to MBTs. A pilot trial found that a 40-minute session of traditional Japanese massage therapy effected PBMC upregulation of genes related to immune response compared to a 40-minute resting control period in 2 healthy women.⁸⁰ Bittman et al found that participation in a creative music-making group significantly reversed proinflammatory cytokine expression activity.⁸¹ These studies suggest shared gene expression changes that may be related to a common signature around perceived stress reduction.

Finally, there is a growing body of research regarding gene expression and other functional genomics technologies in acupuncture. While commonly cited findings overlap with some potential targets of MBT, such as regulation of apoptosis, direct comparisons are difficult, as acupuncture trials often target pain-related genes because of their clinical relevance for acupuncture.⁸²

DISEASE AND TREATMENT CONDITIONS

MBTs have been used to treat a wide variety of diseases, among them irritable bowel syndrome,¹⁵ hypertension,^6,14 insomnia,^6,16 and depression.^19,20,21 By comparing the gene expression changes associated with disease states to the gene expression changes elicited by MBTs, it is possible to identify diseases in which MBTs counterregulate gene pathways central to the illness (ie, downregulate or upregulate genes typically upregulated or downregulated, respectively, by the disease process). Such elucidation of counterregulated gene expression signatures may improve knowledge of the mechanisms of MBTs and suggest the use of MBTs in novel treatment contexts. Two studies suggest that the genomic signature of chronic stress in humans is driven primarily by upregulation of the NF-κB pathway, coupled with blunted glucocorticoid signaling.^52,83 This stress signature shares common domains with, but runs opposite to, the signature of MBT. This stress signature shows strong overlap with the gene expression profile of traumatic stress seen in PTSD that involves significant upregulation of the NF-κB pathway⁸⁴ and apoptotic modulation,⁸⁵ thus overlapping significantly with domains of the MBT signature.

Given the link between MBT, inflammation, and immune response, cancer is another potential epigenetic target of MBT. Our group is investigating hypotheses stating that the MBT-associated gene signature runs counter to gene expression changes seen in the progression of hematological cancers, such as multiple myeloma.⁸⁶ Studies have demonstrated MBT-related gene expression changes in patients with breast cancer⁴⁰ and leukemia,³⁸ although no connection to disease status or progression has been observed. While the impact of MBT on disease progression is modest in aggregate, functional genomics is useful in identifying individual propensities to improve through treatment, as well as propensities to develop diseases, and may prove beneficial in understanding treatment response to MBT.

BEYOND THE TRANSCRIPTOME

LIMITATIONS, CHALLENGES, AND MOVING FORWARD

While several general trends can be deduced from functional genomics outcomes with MBT, the results reported thus far must be interpreted with caution. For example, while downregulation of the NF-κB pathway is the dominant genomics trend in the mind-body literature, several studies did not report NF-κB–related downregulation.^36,37,45 Such variation in results may be because of the diversity of study designs and research questions chosen by individual research groups in this developing field. In a 2013 case, Qu et al used a within-subjects model examining experienced yoga practitioners on a retreat.⁴⁵ While this design reduced potential confounds, the short time window did not allow for analysis of long-term gene expression changes that may have resulted from subjects' years of practice, highlighting the difficulty in comparing these results with other trials.

In general, greater coordination between groups in assembling similar study designs and questions will be critical for moving the field of mind-body genomics forward. Uniformity in sample populations, intervention lengths, and genomics methodologies will ensure easier comparisons among trials. Another key development that we recommend is supplementation of microarrays with more advanced techniques. As mentioned above, several genomics techniques, such as RNA-Seq and DNA methylation sequencing, have yet to be fully utilized in mind-body research. Greater use of these techniques will accelerate our understanding of MBT-related genomics changes.

In addition to these broad recommendations, the field may now be developed enough to address several significant limitations in the current body of literature. For example, most of the studies included in this review are small trials, limiting the strength of conclusions. Our review of the literature found only 1 study with more than 100 subjects, the findings of which are difficult to compare to those of other studies as it is the only trial to target DNA methylation.

Several of the studies covered in this review used active control conditions^40,41,44 but not all adequately controlled for attention. In 1 study, the active group met weekly for 10 weeks; the control group received a 1-day intensive intervention.⁴⁰ Within the field of mind-body medicine, there have been recent advancements in using health or stress education programs as active controls for meditation interventions.⁹⁶ Providing controls with a daily activity, such as reading or listening to music or audiobooks, can control for MBT practice time and the presence of a daily routine. With the exception of a 2014 trial,⁴¹ none of the studies included in this review used this design for mind-body genomics analyses. We recommend that, when applicable, future studies feature adequately sized control groups receiving non-MBT interventions, such as stress education.

As mentioned earlier, the diversity in specific cell types used in transcriptomic analysis deserves careful consideration. Individual white blood cell types vary in their expression of mRNA, and thus observed changes in expression for whole blood may be driven in part by alterations in leukocyte composition or by 1 cell population rather than indicating changes in expression across multiple cell types.³³ We recommend that future trials include an examination of specific peripheral blood cell lines; in addition, the investigation of non-PBMC tissues (such as the prostate biopsies in men with prostate cancer noted above^91,92) would greatly expand our understanding of the effect of MBT on different tissue types and in different disease states.

Heterogeneity of subject populations creates further contextual challenges. Some trials reviewed above investigated patients with specific disease states, while others opted for healthy subjects. Comparisons between these populations should be made cautiously, as certain diseases may predispose subjects to show blunted or heightened genomic responses to MBTs, or disease-related genomic shifts may hide subtle MBT-related responses. While we advocate for trials researching the effect of these practices in clinical populations, well-powered studies with healthy subjects are also needed to determine a baseline genomic signature with greater confidence.

Additional research to discern the length of practice that is needed for long-term benefit from MBT would be helpful. Cross-sectional studies of long-term practitioners have clear methodological limitations, and long-term longitudinal studies are practically difficult and often prohibitively expensive. A research design that has been suggested to address these issues is temporary practice pause then resumption (TPPR).⁹⁷ TPPR pairs experienced practitioners against each other as controls, as some practitioners are randomized to halt their mind-body practice for an observed interval. The TPPR model might be a novel and cost-effective method to address several issues at once: the possibility of a confounding characteristic (personality or otherwise) that influences someone to choose to practice a mind-body technique long term, the potential for enduring epigenetic changes to be overlooked in rapid-response study designs, and the difficulty in determining how much practice experience is necessary to evoke the long-term response signature.

Another point of consideration is the heterogeneity of MBT practices themselves. The literature contains trials of both sedentary practices (such as meditation or yogic breath regulation) and dynamic practices (such as yoga asanas and qigong) that may differ in objects of mental attention, levels of physiological arousal, and general goals of practice. Even within meditation, the most common practice encountered in this review, there may be considerable variation. A 2013 trial of MBSR that included training in multiple mind-body techniques, such as yoga, body scans, and sitting mediation, found differential levels of circulating inflammatory markers depending upon which technique was used most frequently for daily practice.⁶² Furthermore, while we have used the broad term MBT to describe these practices, many studies did not involve a specific therapeutic context. Further studies may need to account for social effects within these various techniques and determine specific benefits that may be associated with individual or group mind-body practice. In addition, comparisons between mind-body group interventions and group-based active controls may be able to differentiate effects that are specific to MBTs from those common to group therapies.

The frequency and time of MBT practice in research interventions also deserve consideration. Current protocols for short-term interventions favor 8- to 12-week models with weekly visits and daily practice; these have demonstrated clear clinical benefits. However, the specific daily prescription varied among studies covered here; for example, 1 study involved practice only twice weekly.⁴¹ For the trials included in this review, practice fidelity of study participants was either not recorded or not reported, so conclusions cannot be drawn about the importance of intervention adherence for genomic determinants. MBT-related gene expression may be a biomarker suited to a dose-response relationship, and future studies should strive to accurately record subjects' frequency and duration of practice.

Despite the heterogeneity of mind-body practices in the studies reviewed, we hypothesize that these practices share a common underlying physiology—just as different forms of exercise share a common set of underlying physiological changes whether one is running, kayaking, biking, or swimming. Needless to say, different MBTs cannot be identical in their effects on the brain and the body. Still, the ability of varied mind-body practices to demonstrate similar physiological effects⁹⁸ and to effect correspondingly similar clinical changes^6,22,99 is evidence that diverse MBTs may take effect through shared molecular mechanisms and may share a core genomic signature with reliably regulated components.