Design considerations for an integrated microphysiological muscle tissue for drug and tissue toxicity testing

Microphysiological systems provide a tool to simulate normal and pathological function of organs for prolonged periods. These systems must incorporate the key functions of the individual organs and enable interactions among the corresponding microphysiological units. The relative size of different microphysiological organs and their flow rates are scaled in proportion to in vivo values. We have developed a microphysiological three-dimensional engineered human skeletal muscle system connected to a circulatory system that consists of a tissue-engineered blood vessel as part of a high-pressure arterial system. The engineered human skeletal muscle tissue reproduces key mechanical behaviors of skeletal muscle in vivo. Pulsatile flow is produced using a novel computer-controlled magnetically activated ferrogel. The system is versatile and the muscle unit can be integrated with other organ systems. Periodic monitoring of biomechanical function provides a non-invasive assessment of the health of the tissue and a way to measure the response to drugs and toxins.

human tissue-engineered blood vessel (TEBV) (Figure 1). Myoblasts were obtained from biopsies of the vastus lateralis of healthy middle-aged volunteers, endothelial cells were cultured from blood-derived late outgrowth endothelial progenitor cells in human umbilical cord blood [10], and vessel wall medial cells were either human dermal fi broblasts or mesenchymal stem cells. Th ese individual units are cultured separately for about 2 weeks to enable the skeletal myoblasts to fuse and diff erentiate and to allow the blood vessels to develop suffi cient mechanical strength. Th e two engineered tissues are inte grated into a perfusion system that enables monitoring of function.
Major challenges for in vitro culture of human skeletal muscle are the lack of effi cient methods to diff erentiate large numbers of human muscle cells from induced pluripotent stem cells and the slower doubling time and rate of fusion for primary human skeletal muscle cells compared with mouse or rat myoblasts. Further, rat and mouse myotubes undergo spontaneous contractions in two-dimensional and three-dimensional cultures, whereas contractile human muscle fi bers in vitro have only been obtained following co-culture with rat or human motoneurons [11].
We utilized our prior fi ndings with engineered rodent muscles [12][13][14][15][16][17][18] to fabricate three-dimensional primary human skeletal muscle tissue bundles. Optimization of the dimensions of the muscle bundles, the media and hydrogel composition, the cell density, and the diff erentiation procedure ( Figure 1) resulted in three-dimensional human muscle bundles with highly aligned, cross-striated myofi bers containing myogenin-positive nuclei and acetylcholine receptor clusters (Figure 2b1). Functional properties of such engineered muscle were tested using standard force test protocols [17]. In response to single electrical stimuli, the bundles produced twitch contractions; at higher stimu lation frequencies, the twitch responses fused into a more forceful tetanic contraction. Amplitudes of twitch and tetanus-specifi c force were signifi cantly lower than those of native muscle, probably due to a lower muscle fi ber density and a smaller fi ber diameter [19], as well as the presence of immature forms of muscle proteins. Similar to native muscle, engineered human muscle bundles exhibited a positive Starling-like force-length relation ship. After the necessary optimization, the method for engineering functional human muscle tissues is robust and reproducible and tissue bundles remain functional for 4 weeks.
To promote diff erentiation and the production of adult forms of muscle contractile proteins, and thereby increase contractile force [20], we have transfected myoblasts with miRNAs prior to formation of the engineered human muscle bundles. miR-133a and miR-696 are of particular interest because they aff ect important develop mental processes: miR-133 a blocks myoblast prolifera tion by inhibiting serum response factor [20], and miR-696 inhibits mitochondrial biogenesis and oxidative metabolism by blocking the metabolic transcriptional co-activator PGC-1α [21]. Engineered human skeletal muscle myoblasts were transfected with anti-miR-133a, with anti-miR-696, or with both anti-miRNAs. Th e miRNA transfection produced longer and more aligned and cross-striated myofi bers when compared with myoblasts that received a scrambled RNA sequence. Myoblasts transfected with both anti-miR-133a and anti-miR-696 exhibited considerable slow myosin heavy chain, indicative of type I muscle fi bers, and generated higher specifi c contractile forces compared with bundles pre pared from myoblasts transfected with only a single anti-miRNA.
To further enhance diff erentiation, normal contractile function of skeletal muscle under various physical demands can be simulated by applying cyclic stretch or electrical stimulation (Figure 1, inset). Electrical stimulation is applied via two platinum electrodes and can be synchronized with the mechanical stimulation [14]. Th e electrical stimulation waveform is precisely bipolar and the amplitude of electrical stimulation is adjusted to accommodate potential changes in the excitation threshold expected to occur with the muscle maturation.

Designing a microperfusion system for engineered skeletal muscle
Engineered human muscle systems are used either as a dynamically conditioned standalone culture system to enable muscle maturation or as an integrated module with other microphysiological units to examine metabolic and drug interactions ( Figure 2). Th e microphysiological system includes a tissue-engineered blood vessel, 750 to 1,000 μm in diameter, in parallel with the muscle tissue. Th e blood vessel consists of a confl uent layer of human endothelial cells and a contractile medial layer of human dermal fi broblasts. Th e medial layer of the TEBV was fabricated either: from aligned human mesenchymal stem cell sheets rolled into a tubular structure, cultured in a rotating wall bioreactor, followed by maturation in a perfusion bioreactor; or by preparing a tubular structure of dense collagen containing human neonatal dermal fi broblasts that is then cultured in a perfusion chamber for 1 week [22]. TEBV mechanical strength is enhanced using oscillating pressure and fl ow, similar to approaches used for larger diameter vessels [23][24][25]. Physiological pulsatile fl ow is generated using a novel magnetoactive porous ferrogel [26,27] that acts as a valve, changing its hydraulic conductivity under applied magnetic fi elds to tune the fl ow rate in the microphysiological system on demand. Other microphysiological organs (for example, myo cardium, liver) can be added in parallel to the muscle (Figure 2a).
Functional evaluation in the system involves monitoring the mechanical behavior of engineered muscle and blood vessels. Th e diameter of the engineered blood vessel is measured in response to changes in pressure, from which the incremental elastic modulus and the ultimate tensile strength prior to failure are determined [24]. Vessel dilation in response to changes in fl ow and nitric oxide release are used to monitor the function of the endothelium. For the muscle tissue, oxygen uptake and the contractile force provide an assessment of muscle function.
Th e design of the system with human cells requires that the shear stress on endothelium in the blood vessel ranges from 0.4 to 2.0 Pa [28]; that the rate of oxygen delivery to the muscle equals or exceeds the rate of oxygen uptake by muscle; and that materials used should not bind drugs. Additional requirements for multiorgan systems are that the relative size of diff erent micro physiological organs and their fl ow rates should be scaled in proportion to in vivo values [1]; that a common media is used for all of the diff erent engineered tissues; and that the system must operate for at least 4 weeks.
Under resting conditions, oxygen uptake in humans in vivo is approximately 1 × 10 -8 moles O 2 /second/cm 3 muscle tissue and can increase 50-fold during exercise [29], similar to our estimate for the oxygen uptake levels of murine skeletal muscle fi bers in three-dimensional cultures [13]. Consequently, we assume that the in vivo uptake rates are representative of muscle fi bers in vitro. For a muscle cell density of 1 × 10 8 cells/ml and oxygen dissolved in culture medium, the fl ow rate to muscle tissue under resting conditions needs to be at least 2.54 μl/second. Since 21% of the cardiac output fl ows through skeletal muscle, the total fl ow rate in the system is 12 μl/second. Adjusting the culture medium viscosity to the blood viscosity, the time-averaged shear stress acting on the endothelium in the TEBV is 0.43 Pa for a 1,000 μm diameter vessel and is 0.84 Pa for an 800 μm diameter vessel, well within the range of values reported in vivo. Th e fl ow is laminar, with Reynolds number around 5 for the TEBV, and is quasi-static, similar to conditions in arterioles [30].
Since skeletal muscle is more abundant than other tissues included in an integrated microphysiological system, this fl ow rate should meet metabolic demands of other tissues -including the liver, which has a high metabolic rate but has 20 times less mass than muscle. Control of fl ow distribution with magnetoactive valves allows simple interfacing with lower-fl ow microfl uidic platforms and permits adjustment of the fl ow rate distribution in response to diff erent physiological stimuli. Continuous monitoring of the oxygen concentration in the perfusate combined with a feedback loop can serve to tune fl ow rate in a physiological-like fashion.

Validation and testing of microphysiological systems
Validation of the microphysiological system involves measuring vessel relaxation and constriction, skeletal muscle contractile force and metabolism over a 4-week ECs muscle Electromagnet period. Th e robustness of the system can be assessed by evaluating perturbations from normal physiology, including acute simulated exercise and acute exposure to TNFα.
Once the function of the tissues in a microphysiological system is validated, a set of candidate drugs known to aff ect skeletal muscle and blood vessels, as well as other tissues, can be tested. Compounds to be examined include those that test normal function (phenylephrine), drugs that are safe and eff ective (lovastatin and metformin [31]), drugs that are eff ective and unsafe (cerivastatin [32,33]), and compounds with known toxicity to skeletal muscle or blood vessels (for example, antimycin [34], rotenone [34] and doxorubicin [35,36]). Initial results indicate that, based upon contractile force generation, engineered human skeletal muscle bundles are more sensitive to doxorubicin than rodent skeletal muscle fi bers [35].
In summary, we have developed the components of a microphysiological system that uses functional measures of blood vessel and skeletal muscle to assess the eff ect of drugs and toxins. By adjusting cell numbers and fl ow rates, the system is fl exible enough to integrate with other microfl uidic and perfusion systems to examine the response of a number of organs and tissues to drugs. Abbreviations miRNA, microRNA; TEBV, tissue-engineered blood vessel; TNF, tumor necrosis factor.

Competing Interests
The authors declare that they have no competing interests.

Authors' contributions
GAT designed experiments, discussed interpretation of the results, and wrote the paper. HEA analyzed data, suggested experiments and edited the paper. NB designed and analyzed experiments, wrote portions of the paper, and edited the paper. HFC performed experiments and analyzed data. CSC designed and performed experiments and analyzed data. CF performed experiments and analyzed data. SH designed and performed experiments and analyzed data. YJ designed and performed experiments and analyzed data. TK analyzed results and edited paper. WEK analyzed results and edited paper. KL designed and analyzed experiments, wrote portions of the paper, and edited the paper. LM designed and analyzed experiments. WMR designed and analyzed experiments, wrote portions of the paper, and edited the paper. XZ designed and analyzed experiments, wrote portions of the paper, and edited the paper. All authors read and approved the fi nal manuscript.