Excess Intramyocellular Lipid Does Not Affect Muscle Fiber Biophysical Properties in Mice or People With Metabolically Abnormal Obesity

Abstract

Observational studies have shown correlations between intramyocellular lipid (IMCL) content and muscle strength and contractile function in people with metabolically abnormal obesity. However, a clear physiologic mechanism for this association is lacking, and causation is debated. We combined immunofluorescent confocal imaging with force measurements on permeabilized muscle fibers from metabolically normal and metabolically abnormal mice and people with metabolically normal (defined as normal fasting plasma glucose and glucose tolerance) and metabolically abnormal (defined as prediabetes and type 2 diabetes) overweight/obesity to evaluate relationships among myocellular lipid droplet characteristics (droplet size and density) and biophysical (active contractile and passive viscoelastic) properties. The fiber type specificity of lipid droplet parameters varied by metabolic status and by species. It was different between mice and people across the board and different between people of different metabolic status. However, despite considerable quantities of IMCL in the metabolically abnormal groups, there were no significant differences in peak active tension or passive viscoelasticity between the metabolically abnormal and control groups in mice or people. Additionally, there were no significant relationships among IMCL parameters and biophysical variables. Thus, we conclude that IMCL accumulation per se does not impact muscle fiber biophysical properties or physically impede contraction.

Article Highlights

• Excess intramyocellular lipid (IMCL) could physically disrupt sarcomere contraction by pushing myofibrils apart, increasing myofilament stiffness, or changing cytoplasmic viscoelasticity.

• We used isolated, permeabilized myofibers from metabolically normal and metabolically abnormal mice and people with metabolically normal and abnormal obesity to evaluate the relationships among IMCL lipid droplet characteristics and myofiber mechanics.

• We found fiber type–specific differences in IMCL accumulation among species and metabolic status but no relationships among any IMCL parameters and myofiber biophysical properties.

• If IMCL directly impairs muscle contraction, it is not through physical disruption of sarcomeres or cytoplasmic composition.

Graphical Abstract

Introduction

Excess accumulation of intramyocellular lipid (IMCL) is associated with broad skeletal muscle dysfunction in the context of obesity, including impaired metabolic and contractile functions (1). While substantial evidence supports a causative link between the accumulation of lipids and lipid intermediates and the development of insulin resistance (2), evidence supporting causal links between pathological IMCL accumulation and contractile deficits remains sparse. Muscle contraction is regulated by integrated systems, including neural innervation and vascular perfusion externally and excitation-contraction coupling, metabolic systems for ATP generation, and the biophysics of cross-bridge cycling internally. It has been hypothesized that IMCL interferes with force generation by chronically driving isoform shifts in proteins involved in excitation-contraction coupling, impairing mitochondrial ATP generation through the accumulation of toxic lipid intermediates or physically impeding cross-bridge cycling by affecting the myofilament lattice or cytoplasmic properties (3).

Here, we evaluate the relationship between IMCL and permeabilized muscle fiber biophysical properties in metabolically normal and abnormal mice and in people with metabolically normal and abnormal obesity, with varying amounts of IMCL, to test the hypothesis that increased IMCL content physically alters muscle fiber mechanics, including both active contractile and passive viscoelastic properties. The permeabilized fiber preparation bypasses neuromuscular transmission, membrane excitability, metabolic capacity, and calcium handling by directly supplying the calcium and ATP needed for force generation and, thus, isolates the biophysics of sarcomere contraction for examination. Any correlations between isolated fiber mechanics and IMCL parameters would therefore support an IMCL-force generation mechanism that is isolated to the myofilament lattice. Putative mechanisms include altered cross-bridge kinetics, myofilament stiffness, myofilament density, or cytoplasmic viscoelasticity. The last two mechanisms are rooted in biophysical interactions between IMCL and the myofilament lattice or cytoplasm. For myofilament density, if IMCL physically increases intermyofilament spacing, it would result in fewer myofibrils per cross-sectional area and, consequently, a lower specific force. For cytoplasmic viscoelasticity, finite element modeling predicts that increasing cytoplasmic stiffness or viscosity (which would increase apparent stiffness) could reduce active force generation (4). These two mechanisms should impact active and passive mechanics of muscle fibers, while cross-bridge–specific mechanisms should only affect active tension.

On the basis of experimental and modeling work of others, we hypothesized that IMCL would increase cytoplasmic viscosity in both mice and people, as predicted from oil-in-water emulsion measurements (5), but that this increase would not be associated with reduced active tension for conditions where IMCL is <10% of the volume fraction as predicted by finite element modeling (4). In opposition to this hypothesis, we found that even the highest IMCL concentrations in mice and people with diabetes had no impact on passive viscosity or elasticity. Furthermore, we found no difference in peak active tension between experimental groups, suggesting that normal contractile function is possible with IMCL volume fractions between 5 and 10%. Together this suggests that IMCL per se does not impact the biophysical properties of permeabilized fibers.

Research Design and Methods

Study Approval

All animal work was performed in accordance with the National Institutes of Health’s Guide for the Use and Care of Laboratory Animals and was approved by the institutional animal care and use committee of the Washington University School of Medicine. All work involving human biopsies was approved by the Human Research Protection Office at Washington University in St. Louis.

Animal Studies

All experiments were performed on male fat-free (FF) and wild-type (WT) littermate mice on a C57BL/6J and 129SV mixed background between 4 and 7 months of age. FF mice completely lack adipose tissue from birth (6) because of a breeding scheme that pairs adiponectin‐Cre mice (028020; The Jackson Laboratory) with lox‐stop‐lox‐Rosa diphtheria toxin mice (010527; The Jackson Laboratory) to ablate adipocytes. All mice were housed at thermoneutrality (30°C) with free cage activity. A subset of FF and WT mice were maintained on a high-fat diet (HFD) (60% fat, 20% protein, 20% carbohydrate, energy density 5.2 kcal/g) (D12492; Research Diets) ad libitum from weaning (4 weeks) through sacrifice (28 weeks). All other mice were maintained on standard chow (13% fat, 29% protein, 60% carbohydrate, energy density 3.0 kcal/g). At the experimental end point, mice were humanely euthanized by cervical dislocation under deep anesthesia.

Human Studies

Human biopsies were obtained from individuals participating in two studies at Washington University School of Medicine. The first study collected gastrocnemius biopsies of seven individuals with prediabetes and seven age-, sex-, and BMI-matched control individuals without prediabetes. Prediabetes was defined as a fasting plasma glucose value ≥100 mg/dL and/or a plasma glucose value ≥140 mg/dL at 2 h after ingesting 75 g of glucose. Participants in the control group were normoglycemic. The second study collected calf muscle biopsies from 15 individuals (11 with diabetes and 4 with prediabetes; 10 gastrocnemius, 2 peroneus longus, 3 unspecified calf) undergoing elective below-knee amputations typically for peripheral neuropathy–related morbidity of the foot. Diabetes was defined as a clinical diagnosis of type 2 diabetes and an HbA_1c ≥6.5%. Prediabetes in this study was defined as no diagnosis of type 2 diabetes and a fasting plasma glucose value ≥100 mg/dL or an HbA_1c ≥5.7%. Participants were selected a priori from the larger group of participants in each study to achieve matching on age and BMI. Participants in both studies were sedentary at the time of biopsy collection. Data were not significantly different when subdividing by study origin or muscle source, supporting our combining these studies (Supplementary Table 1). Demographic information, blood work, and body composition are provided in Table 1.

Table 1.

Study participant characteristics

Parameter	Control group	Prediabetes group	Diabetes group
Participants, n	7	11	11
Sex, n
Male	2	3	7
Female	5	8	4
Age (years)	64.9 ± 4.9	58.2 ± 14.8	57.0 ± 7.7
BMI (kg/m2)	31.0 ± 4.5	31.2 ± 3.8	34.6 ± 6.7
Fat mass (kg)	33.6 ± 9.4	34.8 ± 8.9	NA
Fat percentage	40 ± 9	42 ± 8	NA
Lean mass (kg)	47.3 ± 11.1	44.3 ± 5.8	NA
120-min OGTT (mg/dL)*	105.1 ± 20.13	157.1 ± 25.18a	NA
Fasting glucose (mg/dL)	94.45 ± 5.68	104.4 ± 16.66	157 ± 78.66a,b
Fasting triglycerides (mg/dL)*	57.1 ± 10.94	93.1 ± 13.92	165 ± 86.32a,b
HbA1c*	5.5 ± 0.4	5.4 ± 0.32	7.38 ± 0.82a,b
Duration of diabetes	NA	NA	6.7 ± 3.4
Glucose-lowering medications, n
Metformin	NA	NA	6
Empagliflozin	NA	NA	2
Comorbidities, n
Hypertension	NA	NA	7
Chronic kidney disease	NA	NA	4
Peripheral neuropathy	NA	NA	10

Data are mean ± SD unless otherwise indicated. NA, not applicable; OGTT, oral glucose tolerance test.

^*Only available in seven participants from the prediabetes group.

^aSignificantly different from control group, P < 0.05.

^bSignificantly different from prediabetes group, P < 0.05.

IMCL Quantification

The ankle plantarflexor muscles of the mouse (gastrocnemius, soleus, and plantaris) and the portion of the human muscle biopsies allocated to histology were affixed to cork with tragacanth gum and frozen in liquid nitrogen–cooled isopentane. Axial sections were cut at 10 μm on a cryostat (Leica Biosystems, Wetzlar, Germany) and stained with Oil Red O (ORO) as previously described (7) (Supplementary Methods) or fluorescent BODIPY 493/503 (Life Technologies, Carlsbad, CA). Mean ORO intensity was averaged over 25 fibers, or all fibers if <25 were present in a 10× magnification image, of each fiber type per plantarflexor muscle. BODIPY-positive particles between 0.1 and 5 μm² within a laminin boundary were identified as lipid droplets (LDs), and the average area of individual LDs, the density of LDs (number/fiber area), and the LD area fraction (sum of all LDs/fiber area) were averaged over 20 fibers, or all fibers if <20 were present in a biopsy, of each fiber type per plantarflexor muscle.

Measurement of Isolated Fiber Mechanics

Soleus muscles from mice and a portion of the human muscle biopsies were permeabilized and stored in a glycerinated storage solution at −20°C until the day of testing. Active and passive testing were performed as previously described (8,9) (Supplementary Methods). Peak active force was averaged from three trials recording tension development after rapidly transferring the fiber from a bath of low-Ca²⁺ relaxing solution to a bath of high-Ca²⁺ activating solution. At the end of active testing, fibers were returned to relaxing solution and then subjected to an incremental stress relaxation test. Stress relaxation data were fit to a three-element spring-dashpot Hill-type model with a time-dependent viscosity (9). K_p and K_s reflect the parallel and series components of elasticity, which are generally considered to reflect sarcomeric proteins and other cytoplasmic proteins, respectively. The initial viscosity (η₀) reflects the instantaneous viscosity during stretch, the rest viscosity (η_∞) reflects the final viscosity at the end of the hold, and α reflects the rate of change in viscosity over the course of stress and relaxation. All forces were normalized to fiber cross-sectional area using measured diameter and assuming a cylindrical shape. Following testing, fiber type was identified by immunostaining for myosin heavy chain isoforms. Ten to 20 fibers were tested per muscle or biopsy. The sample size for some of these assessments in human muscle is less than that for histology because of technical challenges with isolated fiber mechanics. Reasons for this include insufficient biopsy material (fibers too short or too little sample for both histology and mechanics) or fewer than five fibers with quality data of a given type.

Measurement of Muscle Contractility

Mouse soleus contractility was measured as previously described (7) (Supplementary Methods). Peak tetanic force was normalized to physiological cross-sectional area (peak tetanic tension), and twitch kinetics were determined as the time between the onset of stimulation and peak twitch tension (time to peak tension) and half the time between peak twitch tension and full relaxation (half-relaxation time).

Statistical Analyses and Reproducibility

Sample size was selected a priori using a power calculation (G*Power) with variance estimated from published data (7,10,11). Mouse and human biopsies were given an alphanumeric identifier, and all experimenters were blinded to genotype and demographics.

Between-group comparisons were made by one- or two-way ANOVA with repeated measures across fiber types where indicated, with a Tukey multiple comparisons test applied to comparisons with a significant interaction term. Actual numbers per group are indicated in each figure for each analysis. All results are presented as mean ± SD. All statistical analyses were performed using GraphPad Prism software.

Data and Resource Availability

The data generated from this study are available from the corresponding author upon reasonable request. No resources were generated.

Results

Chow-Fed FF Mice and HFD-Fed WT Mice Have Similar Quantities and Distribution of IMCL

We have recently used the FF mouse to disentangle adiposity from general metabolic dysfunction in the development of muscle pathology (7) since the FF mouse lacks adipose tissue but experiences other features of high-fat feeding, including hyperlipidemia (12). In line with our previous findings (7), we confirm here that FF mice on a chow diet have increased IMCL storage in the ankle plantarflexor muscle group compared with WT mice (Fig. 1A, top two rows) due to chronic hyperlipidemia. To determine the similarity of this pattern to diet-induced obesity, we maintained WT and FF mice on an HFD for 24 weeks. Detailed phenotypic characterization of these mice can be found in our previous study (12). HFD increased IMCL content in WT muscle but not in FF muscle (Fig. 1A, bottom two rows), suggesting that IMCL storage capacity is already maximized in FF mice on a chow diet. Quantification of ORO staining in the soleus muscle also revealed that only WT mice had an increase in IMCL content with HFD (Fig. 1B). There were no differences in fiber type distribution by genotype or diet (P > 0.1).

Figure 1.

IMCL accumulation and contractile dynamics in FF mouse muscle resembles HFD-induced obesity. A: Representative staining of histological sections comprising the ankle plantarflexors (soleus [SOL], plantaris [PLN], and gastrocnemius [GAS]). IMCL accumulation is apparent in both chow-fed FF and HFD-fed WT and FF mice as increased red intensity on ORO- stained sections (right). Sporadic high-intensity staining can be mapped back to type 2a and type 2x fibers in adjacent myosin heavy chain immunostained images (left). B: Quantification of ORO staining of the SOL across the most abundant fiber types revealed increased intensity primarily in type 2a fibers. C: Peak tetanic tension was significantly elevated in FF mouse SOL compared with WT mice, despite high quantities of IMCL. D: The rate of force development (time to peak) and relaxation (half-relaxation time) are prolonged in FF and HFD-fed WT mouse SOL. Statistics: two-way ANOVA, with Tukey multiple comparisons test (B–D). #P < 0.05 compared with the genotype-matched chow-fed condition. Percentages next to each fiber type in B indicate the average percentage of that type in the mouse SOL. AU, arbitrary unit.

Muscle Contractile Force But Not Kinetics Is Different Between Chow-Fed FF and HFD-Fed WT Mice

Also in alignment with our previous study (7), we found that isolated soleus muscles from FF mice have higher peak tetanic tension than WT despite the significant difference in IMCL content (Fig. 1C). However, chow-fed FF mice and both HFD-fed groups had similar increases in twitch kinetic times over chow-fed WT mice (Fig. 1D).

Muscles From Chow-Fed FF and HFD-Fed WT Mice Have Similar IMCL LD Features

One explanation for the differential effect of IMCL accumulation in the athlete’s paradox is the different size and density of LDs (11,13). To determine whether changes in LD size or number could explain the discrepant active forces following HFD in FF and WT mice, we stained soleus sections with fluorescent BODIPY and used confocal imaging to capture individual LD metrics. Qualitatively, BODIPY staining supported our ORO results in that LDs were more prevalent in HFD-fed WT mice and FF mice on both diets compared with chow-fed WT mice, with LDs predominantly concentrated in type 2 fibers (Fig. 2A). HFD increased the average area of individual LDs in type 1 fibers in WT mice but had no effect on FF mice whose average LD area was already elevated above chow-fed WT (Fig. 2B). Average LD area was also significantly increased with HFD in WT type 2 fibers but not in FF, which already had high LD areas (Fig. 2C). However, in type 2 fibers, the number of LDs was also significantly increased in HFD-fed WT mice and FF mice on both diets compared with chow-fed WT mice, leading to a larger difference among groups in LD area fraction. Taken together, these data indicate that IMCL accumulation in FF mice mimics IMCL accumulation in HFD-induced obesity.

Figure 2.

LD size and density are similar between FF and HFD WT conditions. A: Representative staining of histological sections of the soleus muscle imaged with confocal microscopy. Individual BODIPY-stained LDs can be discerned, accumulating primarily in fibers negative for type 1 myosin heavy chain. B: Quantification of LDs in type 1 fibers revealed that LD size (average area) was larger in HFD-fed WT and FF mice compared with chow-fed WT mice, but density (number per fiber area) and area fraction (the sum of all LDs divided by fiber area) were not different between groups. C: Quantification of LDs in type 2 fibers revealed that LD size, density, and area fraction were all increased in HFD-fed WT and FF mice compared with chow-fed WT mice. Statistics: two-way ANOVA, with Tukey multiple comparisons test (B and C). #P < 0.05 compared with the genotype-matched chow condition.

IMCL Accumulation in FF Mice Does Not Impact Active or Passive Mechanics of Permeabilized Fibers

Given the similarities in IMCL parameters between FF mouse soleus muscle and obesity models, we sought to determine whether IMCL would impact active or passive mechanics in the absence of obesity. Permeabilized fibers from chow-fed WT and FF mouse soleus muscles were subjected to active and passive mechanical testing (Fig. 3A), with passive data fit to a three-element Hill-type model with time-dependent viscosity (Fig. 3B). In line with our ex vivo data for the chow-fed mouse soleus muscles, there was no difference in peak active tension between genotypes in either fiber type (Fig. 3C). Parallel (K_p) and series (K_s) components of elasticity were both significantly lower in type 2 than type 1 fibers, as previously reported (14), but were not different between genotypes (Fig. 3D). Similarly, neither the initial (η₀) or rest (η_∞) viscosities or the rate of viscous change over relaxation time (α) were different by genotype (Fig. 3E). While all mechanics data were normalized to fiber cross-sectional area, it is important to note that fiber cross-sectional areas were not different between groups (type 1 fibers, P = 0.27; type 2 fibers, P = 0.80). Together, these findings suggest that IMCL accumulation at this level, on its own, does not impair active force generation or passive fiber mechanics.

Figure 3.

Permeabilized FF mouse fiber mechanics are unaffected by IMCL accumulation. A: Representative force traces from a test of active and passive mechanics of permeabilized fibers. B: Schematic of the three-element model used to calculate the parallel (K_p) and series (K_s) components of elasticity and viscosity as a function of time [η(t)]. C: Peak active tension was not different between genotypes. D: Passive elasticity was lower in type 2 than type 1 fibers but was not different between genotypes. E: Passive viscosity was not different between genotypes or fibers. Statistics: two-way ANOVA, repeated measures across subjects, with Tukey multiple comparisons test (C–E). #P < 0.05 compared with genotype-matched type 1 fibers.

Individuals With Prediabetes and Diabetes Exhibit Differing Patterns of IMCL Accumulation

To determine how representative the IMCL accumulation in FF mice is of human obesity and metabolic dysfunction, we also quantified IMCL LD metrics in biopsies from individuals across a range of glycemic status. In contrast to our data in mice, we found that IMCL was more evenly distributed between fiber types in the control group (Fig. 4A). LD density (number of LDs per fiber area) was higher in the prediabetic compared with the control group, and LD size was not different between the two groups (Fig. 4B and C, orange bars). The diabetes group had significantly increased LD size in both fiber types. Interestingly, the diabetes group had a lower LD density than the prediabetes group in type 1 fibers but a higher density in type 2 fibers. This ultimately resulted in a significantly higher LD area fraction in the diabetes compared with control group across both fiber types but only a higher LD area fraction compared with the prediabetes group in type 2 fibers. There was no difference in the distribution of fiber types between groups (P > 0.3), suggesting that only the overall IMCL quantity was adjusted by modulating LDs rather than fiber type proportions. Taken together these findings suggest that in the early phases of dyslipidemia and insulin resistance, IMCL is added primarily as new LDs of normal size in type 1 fibers, but at later stages, LD volumes increase across both fiber types in tandem with increased LD numbers in type 2 fibers.

Figure 4.

Human IMCL accumulation differs in prediabetes and diabetes. A: Representative images of IMCL accumulation in biopsies from the control, prediabetes, and diabetes groups. In the prediabetes images, IMCL can be seen primarily in type 1 myosin heavy chain–stained fibers (magenta far left; gray far right) in both ORO- and BODIPY-stained images. In the diabetes images, IMCL can be seen in type 1 and type 2 fibers with varying intensity. B: Quantification of LD metrics on BODIPY-stained type 1 fibers revealed an increase in average LD area in the diabetes group only (left), an increase in LD density in both prediabetes and diabetes groups (middle), and a resultant increase in LD area fraction only in the diabetes group (right). C: Quantification of LD metrics on BODIPY-stained type 2 fibers revealed an increase in average LD area, LD density, and LD area fraction in the diabetes group only. Statistics: one-way ANOVA, with Tukey multiple comparisons test (B and C). *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001.

Prediabetes and Diabetes Have No Effect on Permeabilized Fiber Mechanics

A subset of biopsies used for the histological analyses could be also used for permeabilized fiber mechanical testing. There were no significant differences between groups in peak active tension in either fiber type (Fig. 5A). Similarly, there were no significant differences between groups in parallel or series elasticity (Fig. 5B) or viscosity (Fig. 5C) or fiber cross-sectional area (type 1 fibers, P = 0.33; type 2 fibers, P = 0.13).

Figure 5.

Permeabilized fiber mechanics are not significantly affected by diabetes state. A: Peak active tension was not different between groups for either fiber type. B: Passive elasticity was not different between groups for either fiber type. C: Passive viscosity was not different between groups for either fiber type. Statistics: two-way ANOVA, repeated measures across participants (A–C).

Peak Active Tension Is Not Correlated With IMCL Parameters in Either Fiber Type

To determine whether any parameter of IMCL accumulation was predictive of peak active tension, we generated a Pearson correlation matrix for each fiber type. No IMCL parameter was significantly correlated with peak active tension or any measure of passive viscoelasticity (Fig. 6, black-outlined boxes). Some relationships between peak active tension and passive viscoelasticity were significant. Peak active tension was correlated with initial viscosity (η₀) in type 1 fibers and with the parallel component of elasticity in type 2 fibers (Fig. 6, top gray-outlined boxes). Additionally, some of the passive viscoelasticity parameters were intercorrelated (Fig. 6, bottom gray-outlined boxes).

Figure 6.

No IMCL LD parameter is significantly correlated with peak active tension or passive viscoelasticity in permeabilized fibers. A: For type 1 fibers, initial viscosity (η₀) was the only variable correlated with peak active tension. B: For type 2 fibers, the parallel component of elasticity (K_p) was the only variable correlated with peak active tension. Some LD variables were intercorrelated, and some passive viscoelastic variables were intercorrelated. Statistics: Pearson correlation. The Pearson r values are listed in each box; orange shading denotes significant correlations. Fr, fraction.

Discussion

In this work, we conclude that IMCL accumulation does not impact muscle fiber biophysical properties. In FF mice lacking adipose tissue, high IMCL levels comparable with those observed in obese animal models have no effect on permeabilized fiber mechanics. Similarly, in humans with metabolically abnormal obesity, metabolic abnormality was associated with increased IMCL with varied patterns of deposition but had no effect on fiber mechanics. These findings strongly support that increased levels of IMCL encountered in people with prediabetes and type 2 diabetes are insufficient to impact myofibril density or cytoplasmic viscosity.

Our data show that in FF mice, we can mimic the effects of HFD on IMCL accumulation without the confounding effects of obesity. We found that FF and HFD-fed WT mouse soleus muscles have a similar ∼2–2.5-fold higher ORO intensity compared with chow-fed WT mice. This aligns with other studies of long-term HFD, which found a 1.5–2-fold increase in ORO staining (15,16). In BODIPY-stained images, we found less difference in the LD area fraction in type 1 fibers (∼1.5-fold) and more difference in type 2 fibers (approximately threefold). This matches reports in the literature of higher IMCL in type 2a and 2x fibers than type 1 or 2b fibers both at baseline and following HFD (17,18). Interestingly, average LD areas and numbers per fiber area in FF mice are also comparable with what has been reported in humans with obesity (11,13,19,20). In our human biopsies, we found larger LD size in both type 1 and 2 fibers from individuals with diabetes, but other measures across groups were equal to or lower than values in type 2 fibers in the FF mouse.

Also in our human biopsies, we could characterize a wide range of IMCL deposition because we combined data from people across a range of the glycemic spectrum. The majority of studies investigating IMCL accumulation in people with diabetes enrolled participants with diabetes without comorbidities (renal failure, cardiomyopathy, hypertension, or peripheral neuropathy), which selects for a group more central on the spectrum. One study comparing individuals without diagnosed diabetes, stratified into insulin sensitive and insulin resistant groups (comparable with our control and prediabetes groups) also found differences in IMCL accumulation in type 1 but not type 2 fibers (10). Studies in individuals with diabetes (without comorbidities) also have found differences in IMCL accumulation in type 2 fibers compared with BMI-matched sedentary control individuals (11,21,22). Notably, our LD size in the diabetes group was two- to threefold higher than has been reported in individuals with diabetes without comorbidities with comparable values across control groups (11,20,21,23). It is tempting to speculate that in the early phases of hyperglycemia (prediabetes), IMCL accumulates in a physiologically accessible way, as increases in LD numbers in type 1 fibers are also found in elite athletes. But then in more significant hyperglycemia, the size of LDs increases, which is most pronounced in type 2 fibers. Our data suggest that LD size continues to increase through the course of diabetes in the context of multiple comorbidities, in particular diabetic peripheral neuropathy.

Our hypothesis that IMCL would impact fiber viscoelasticity was based on predictions in the literature (4,5). Rahemi et al. (4) generated a finite element model of IMCL accumulation that predicted reduced active tension at IMCL area fractions >10% because of changes in cytoplasmic material properties. On the other hand, direct rheology measurements on oil-in-water immersions predicted that oil volume fractions >25% are required to increase immersion viscosity (5). We found that LD area fractions were highest in type 2 fibers of FF mice (6–8%) and both fiber types in the diabetes group (3–7%), both of which are <10%. Recent work has demonstrated that the levels of IMCL in diabetes can impact cytoplasmic diffusivity as measured by magnetic resonance spectroscopy (24), suggesting that this quantity of IMCL can impact some cytoplasmic properties. However, data from this study do not support these changes manifesting as measurable differences in cytoplasmic viscosity. Our data also do not support IMCL changing myofilament density, as passive stiffness of permeabilized fibers was not different between groups and was not correlated with LD parameters. In type 1 fibers, initial viscosity (η₀) was significantly correlated with peak active tension, suggesting that higher viscosity at the initiation of contraction could lead to slower cross-bridge cycling. However, this relationship was independent of IMCL and not observed in type 2 fibers, which had similar values for η₀, suggesting that it is not a conserved mechanism related to IMCL accumulation. In type 2 fibers, the parallel component of stiffness (K_p) was significantly correlated with peak active tension, but the relationship was in the direction opposite of finite element modeling predictions for changes in cytoplasmic viscoelasticity (4). However, increased myofilament lattice stiffness is predicted to increase peak-specific force (25), and while this is typically thought to contribute to total tension during active cross-bridge cycling, weakly bound cross bridges may enable the myofilament lattice to contribute to passive stiffness as well (26).

Our finding that peak active tension is not different between groups in either the mouse or human data contrasts with a sizeable body of literature that has suggested that obesity and diabetes decrease normalized muscle contractile force. The majority of these data have come from rodent studies testing intact muscles where bioenergetic mechanisms contribute to force generation (27–31). As a metabolic fuel source, IMCL could impact the bioenergetics of contraction and impair intact muscle force generation. However, these data are not unequivocal, as some studies found an effect in only one muscle type (28–31) and some, no effect (32,33), which suggests a more nuanced relationship. In support of this, our data in FF mice revealed that although the FF soleus had as much IMCL as the HFD-fed WT mice, it had significantly higher contractile tension in the intact preparation. This is in contrast to the extensor digitorum longus, which had lower contractile tension in the FF mouse because of a fast-fiber–specific regulation by leptin (7). The soleus was chosen for this study because of its high IMCL content and lack of overt pathology.

Our data also contrast with work finding significant correlations between IMCL and permeabilized fiber active tension in humans (34,35). There are several factors that could contribute to this discrepancy. First, the participants in these studies were, on average, 10 years older than in our study (∼70 vs. ∼60 years of age), and the contribution of age-related changes could be confounding IMCL-related changes as suggested by rodent studies (27). Second, because of our study design matching participants on BMI, our control participants were all overweight or obese. It is possible that the prediabetes and diabetes groups would have significantly reduced active tension if compared with a healthy weight group. However, our groups had substantially different levels of IMCL accumulation, which strongly supports that such a reduction would not be driven by IMCL. Third, fiber mechanics could be affected by comorbidities (i.e., with or without peripheral neuropathy), which could increase the variability in our data. However, within-subject variability was not different between groups for any of the mechanics metrics (Supplementary Fig. 1). Finally, bias in fiber selection could have affected our results. Fiber morphology and quality varied considerably within a biopsy in the diabetes group (Fig. 4A), and since we selected fibers for testing that were cylindrical in shape without obvious kinks or bends, we may have biased our selection for fibers that were healthier. Future studies should leverage this variability to determine whether IMCL is correlated with fiber mechanics on a per-fiber basis.

In conclusion, we found that mouse skeletal muscle can accommodate IMCL levels comparable with those induced by HFD without a significant impact on permeabilized fiber active or passive mechanics. We observed that in the context of prediabetes and diabetes, human permeabilized fibers can also accommodate these levels of IMCL without compromising active or passive mechanics. Force generation in intact muscles could still be directly impacted by IMCL as some literature has suggested. However, our data collectively reject the hypothesis that changes in myofilament density or cytoplasmic biophysical properties represent the mechanism for these observations.

This article contains supplementary material online at https://doi.org/10.2337/figshare.25731582.

Funding Statement

This study was supported by National Institutes of Health grants AR075773, AG15768, AG46927, AR072999, AR073752, P30 AR074992 (Musculoskeletal Research Center) and P30 DK056341 (Nutrition Obesity Research Center) and Longer Life Foundation grant 2019-011.