Hydrogel delivery of lentiviral particles

An H2B-GFP lentivirus was generated to assess the efficiency of lentiviral delivery to 3D-tissue-engineered-skeletal muscles (3D-TESMs). We first tested whether lentiviral transduction and H2B-GFP overexpression might affect myogenic differentiation. To this end, myogenic progenitor cells (MPCs) grown in 2D were transduced by the addition of virus to the medium [13]. Analysis was performed 48 h after transduction and showed that 88% of nuclei were GFP positive (Fig. S1). Next, myogenic differentiation was induced through serum deprivation, and immunofluorescent staining with a myosin heavy chain (MYH) antibody was used to detect terminally differentiated MYH-positive multinucleated myofibers. We observed similar fusion indexes between the transduced and untransduced MPCs (79% and 84%, respectively, Fig. S1). GFP-positive nuclei were found to be present throughout the terminally differentiated myotubes after lentiviral targeting with a similar nuclear distribution as untransduced myotubes. This indicates that the MPCs retained their myogenic potential upon lentiviral transduction with H2B-GFP.

Next, we evaluated different lentiviral delivery methods for transducing 3D-TESMs: via the medium of MPCs grown in 2D, followed by 3D-TESM generation (Method 1) and via the medium of preformed 3D-TESMs (Method 2): or by mixing lentiviruses with MPCs and the hydrogel during 3D-TESM formation (Method 3) (Fig. 1A). Lentiviral transduction was tested at 1:4 serial dilutions, and after 48 h, the percentage of GFP-positive cells was determined using flow cytometry. A targeting efficiency of ± 70% was obtained when using method 1 with the undiluted viral titer (Fig. 1B–C). Transduction via the culture medium in 3D-TESMs (Fig. 1, Method 2) resulted in a very low efficiency of < 1% at the highest titer. In contrast, transduction via the hydrogel (Fig. 1, Method 3) resulted in enhanced efficiency compared to Method 1 of up to 88% at the highest titer (Fig. 1B–C). Serial dilutions of the lentivirus resulted in concomitant lower transduction efficiencies. Confocal microscopy confirmed these results (Fig. 1D). Only a minor fraction of GFP-positive nuclei were detected in the 3D-TESMs that were transduced via Method 2, in which GFP-positive nuclei were confined to the outer layer of the 3D-TESMs, indicating that lentiviral particles were unable to penetrate into the interior of the tissue using this delivery method. Analysis of the 3D-TESMs that were transduced via Method 3 showed that almost all nuclei were GFP positive (Fig. 1D). GFP-positive nuclei were found to be equally distributed throughout the 3D-TESMs, indicating that in this delivery method, lentiviral particles were able to target cells of both the inner and outer layers of the 3D-TESMs. As a final confirmation, myogenic differentiation was induced in the 3D-TESMs that were transduced with Method 3 for a 7-day period. Analysis of these 3D-TESMs, using titin (TTN) as a marker to identify terminally differentiated myofibers, showed an abundance of GFP-positive nuclei in between and within differentiated myofibers (Fig. 1E), indicating that transduced MPCs were able to undergo myogenic differentiation in 3D-TESMs.

Fig. 1

Efficient lentiviral transduction of 3D-tissue-engineered-skeletal muscle. A Experimental approach. Method 1: cells were transduced in 2D, before the generation of 3D-TESMs. Method 2: 3D-TESMs were generated from untransduced cells and were then transduced with lentiviral particles delivered through the cell culture medium. Method 3: 3D-TESMs were generated from untransduced cells and were then transduced with lentiviral particles delivered through the hydrogel during 3D-TESM formation. An H2B-GFP Lentivirus was used to assess the efficacy of the delivery method. B FACS analysis of the number of GFP-positive cells as a percentage of the whole population for the three delivery methods. Four H2B-GFP lentiviral concentrations were tested per method. Data are derived from three independent 3D-TESMs and expressed as mean ± SD. C Representative FACS plots from B. D Whole-mount immunofluorescent analysis of 3D-TESMs showing GFP 48 h after transduction with undiluted H2B-GFP virus using Method 2. Zoomed-in region of a single Z-stack is shown on the right. Nuclei were stained with Hoechst (in blue). E as D but for Method 3. F Sagittal Z-stack of 3D-TESMs transduced with Method 3, after 7 days of myogenic induction, stained with an anti-titin antibody (red)

Generation of 3D-TESM disease models through shRNA-mediated knockdown

To apply lentiviral transduction for disease modelling using 3D-TESMs, lentiviruses expressing 2–3 shRNA targeting sequences per gene for dystrophin (DMD), calpain-3 (CAPN3), or myostatin (MSTN) were tested individually (Table 1). Optimal viral titers required for knock down were determined by transducing MPCs in 2D using serial dilutions for each target sequence, followed by RT-qPCR analysis. 3D-TESMs were then generated and transduced using hydrogel delivery of shRNA expressing lentiviruses (Method 3). Myogenic differentiation was induced for a period of 7 days, and 3D-TESM tissues were analyzed for force-generating capacity in response to electrical stimulation, expression of the targeted gene, and tissue morphology using whole-mount immunofluorescence of TTN (Figs. S2–S4). The 3D-TESMs that were found to have a > 50% reduction of CAPN3 (shRNA #2 and #3) or DMD (shRNA #1 and #3) expression after shRNA transduction showed a reduced or a complete lack of contractile force and a reduction of aligned TTN-positive myofibers (Figs. S2 and S3). No significant differences in morphology or contractile force were found between the 2 MSTN knock down shRNAs and non-targeting control 3D-TESMs, while both targeting shRNAs caused a significant reduction in the expression of MSTN (Fig. S4).

Table 1 Viral constructs used in this study

Next, we performed a time course experiment to characterize the effects of DMD, CAPN3, and MSTN knock downs in 3D-TESMs in more detail. Using the best-performing shRNA construct from Figs. S2–S4 for each target gene (Table 1), we analyzed myofiber diameter (Fig. S6A), myofiber alignment (Fig. S6B), the force generating capacity of 3D-TESMs in response to twitch stimulation (Fig. 2A; “force-generating capacity”), and tetanic stimulation (Fig. 2B; “maximum force-generating capacity”) after 3, 5, 7, and 9 days of 3D-TESM formation. 3D-TESMs that were transduced with the non-targeting shRNA generated contractile forces of ~0.4 mN already at 3 days of differentiation and showed a gradual increase in force-generating capacity to a maximum of ~1.2 mN at day 9. Parallel-oriented TTN-positive myofibers were observed from day 3 of differentiation onwards (Fig. 2C). A strong phenotype was observed for the knock down of DMD in 3D-TESMs. Contractile forces were close to zero starting from day 3 of differentiation onwards (Fig. 2A–B). This was paralleled by morphology: from day 3 of differentiation onwards, TTN-positive myofibers were shortened, they lacked cross-striation, and the 3D-TESMs appeared disorganized, lacking fiber alignment (Figs. 2C and S6B).

Fig. 2

shRNA-mediated knock down of DMD, CAPN3, and MSTN in 3D-tissue-engineered-skeletal muscle. A Effect of shRNA-mediated knock down on force-generating capacity (twitch force) in 3D-TESMs. 3D-TESMs were transduced (using Method 3) with shRNAs targeting DMD, CAPN3, or MSTN. Force-generating capacity was measured in response to stimulation at 1 Hz. Data are represented as mean ± SD derived from three independent 3D-TESMs at each timepoint (12 3D-TESMs total) for all conditions. Statistical significance is indicated for all conditions, compared to the corresponding non-targeted control. *p < 0.05, **p < 0.01, ***p < 0.001. B as A but now showing the maximum force-generating capacity in response to tetanic stimulation (20 Hz). C Whole-mount immunofluorescent stainings of all experimental conditions from A and B. Green, anti-titin antibody. Blue, nuclei stained with Hoechst

Knock down of CAPN3 in 3D-TESMs resulted in contractile forces that were similar to those generated in the non-targeting controls at day 3 of differentiation (Fig. 2A–B). From day 5 of differentiation onwards, CAPN3 knock down caused reduced contractility, ultimately resulting in a complete absence of contractile response on day 9 of differentiation. Also in this case did the morphology parallel the force-generating capacity: aligned and cross-striated myofibers were present at day 3 of differentiation, but from day 5 onwards, fiber organization appeared irregular and disrupted, which was accompanied by a progressive increase in the presence of TTN-positive spherical structures (Figs. 2C and S6). We hypothesize that these structures represented partially detached myofibers, since their increase appeared to correspond with the loss of myofibers. At day 9 of differentiation, striated myofibers were no longer present.

Knock down of MSTN failed to cause significant effects on force-generating capacity, although a nonsignificant increase in twitch and tetanic force was observed at day 9 of differentiation (Fig. 2A–B). The morphology and organization of the myofibers appeared unchanged (Fig. 2C). Lentiviral transduction resulted in a vector copy number (VCN) of 2–6 (Fig. S5A, B), and did not affect fusion index in myotubes grown in 2D (Fig. S5C, D). This suggest that the knock downs did not interfere with formation of myotubes, and that the observed phenotypes resulted from pathology-induced downstream of myogenic fusion. For all knock downs, tissue diameter appeared similar (Fig. S7). We conclude that knock down of DMD and CAPN3, but not MSTN, caused a severe reduction of contractile force in 3D-TESMs from days 3 to 5 until day 9, the last day tested, which was accompanied by disruption of myofiber morphology and organization.

Proteomic analysis of shRNA-mediated knock downs in 3D-TESMs

We performed mass spectrometry of 3D-TESMs to validate knock down of targeted genes at the protein level and to test whether disease-specific proteomic profiles were obtained. Relative levels of Calpain-3 and dystrophin were reduced at all timepoints of knock downs of CAPN3 and DMD, respectively, compared to non-targeting control or MSTN knock downs (Fig. 3A). This effect was most pronounced at the latest timepoint, resulting in an 18-fold reduction of Calpain-3 following CAPN3 knock down and a 5-fold reduction of dystrophin in the DMD knock down. Levels of control proteins vinculin and GAPDH were similar for all knock downs at all time points. Myostatin levels were undetectable, precluding their analysis.

Fig. 3

Proteomic analysis of DMD and CAPN3 knock downs in 3D-tissue-engineered-skeletal muscle. A Effect of knock down on Calpain-3 (CAPN3) and dystrophin protein levels as a function of time in 3D-TESMs. For reference, the abundances of vinculin and GAPDH are shown. Data are expressed as log2 LFQ values and represented as means ± SD derived from three independent 3D-TESMs at each timepoint. B Volcano plot of the CAPN3 knock down compared to the non-targeting control at day 7. Proteins that were significantly up and downregulated are indicated in red and blue, respectively. Average protein LFQ values were derived from three independent 3D-TESMs from each timepoint. Proteins were considered significant when FDR was < 0.05 in ≥ 2 timepoints. C Supervised clustering of the CAPN3 knock down. Clusters with similar expression changes over time are indicated. Clustering was based on the Z-scores of the LFQ values of significant proteins in the CAPN3 knock down relative to the non-targeting control. n = total number of proteins per cluster. D Z-scores of each protein cluster from C. E Top 5 enriched GO pathways in the upregulated protein clusters of the CAPN3 knock down. F Top 5 enriched GO pathways in the downregulated protein clusters of the CAPN3 knock down. G Heatmap of proteins found significantly altered in the CAPN3 knock down relative to the non-targeting control. Colors and numerical values represent the mean LFQ as a percentage of the most enriched condition. H–M as B–G but for the knock down of DMD

To analyze disease-specific protein signatures, we performed unsupervised clustering of all significant proteins, based on the Z-scores for each protein normalized for the non-targeting control for each respective timepoint. Expression of a total of 596 proteins was significantly changed (FDR < 0.05 in ≥ 2 timepoints) for the knock down of CAPN3 (Fig. 3B). Four distinct protein clusters were identified based on their expression from d3 to d9 of culture, resulting in two clusters with upregulated (cluster 1 and 2) and one cluster with downregulated proteins (cluster 4) compared to the untargeted control (Fig. 3C–D). Cluster 3 contained only 32 proteins and could not be classified as up or downregulated and was therefore omitted from GO enrichment analysis. GO enrichment analysis of clusters 1, 2, and 4 showed enrichment of pathways involved in cell adhesion and cytoskeletal binding in both up- and downregulated protein clusters. In the downregulated protein cluster 4, enrichment of proteins associated with the proteasomal complex (PSMA3, PSMA5, PSMB3, PSMB4, PSB2, PSB3) and several pathways involved in skeletal muscle contractility was observed (Fig. 3D–F). These results parallel studies performed using muscle biopsies or primary cell lines from LGMD2A patients describing downregulation of the 26S proteasome and dysregulation of cytoskeletal proteins due to defects in the proteolytic function of Calpain-3 in these patients [22, 23].

Knock down of DMD in 3D-TESMs resulted in 711 significantly altered proteins that could be divided in three protein clusters: clusters 1 and 2 with upregulated proteins and cluster 3 with downregulated proteins compared to the non-targeting control (Fig. 3H–J). GO enrichment analysis showed that the downregulated cluster was enriched for proteins involved in the development and function of skeletal muscle tissues (MYHs, NEB, OBSCN, ATP1A2, ATP2A1) (Fig. 3I). The upregulated protein clusters were found to be enriched for proteins involved in cell-matrix attachment and ECM proteins, including ANXA1, COL4A1, COL4A2, COL8A1, LAMA1, ITGA3, and VTN (Fig. 3J–K). These results are in agreement with results obtained in the mdx mouse model and in primary muscle biopsies from human patients, whereby fibrosis is one of the main pathological features of Duchenne muscular dystrophy [24].

The proteomic analysis also offered an opportunity to analyze proteomic changes that occur during human muscle development in vitro. To this end, we analyzed the non-targeting control during 3D-TESM development and found 1282 proteins that were significantly altered between at least 2 timepoints of development. These proteins could be classified within one of five distinct protein clusters based on their expression pattern (Fig. S8A–B). Quality controls are shown in Fig. S9. Gene Ontology (GO) enrichment analysis was performed for the two main protein clusters, cluster 1 (n = 408; upregulated during development) and cluster 5 (n = 519; downregulated during development) (Fig. S8A–B). Cluster 1 was predominantly associated with cellular respiration, muscle contraction, and glucose metabolism, while cluster 5 contained proteins involved in the regulation of mRNA processing and stability (Fig. S8D–E). The developmental stage of skeletal muscle is characterized by the expression of distinct myosin heavy chain (MYH) isoforms. The embryonic isoform (MYH3) was the most abundant isoform at all timepoints. During the development of 3D-TESMs, there was a gradual increase of MYH1 expression (associated with both fetal and adult fibers), MYH4 (postnatal and adult fibers), and MYH8 (embryonic and fetal fibers) from day 3 to day 9 (Fig. S8E). This reflects a gradual maturation from the embryonic into the neonatal stage that occurs during the normal development of skeletal muscle [25]. In addition, at day 9, the highest abundance of sarcomeric proteins including TTN, NEB, ACTN3, and OBSCN were found, while proteins associated with the initial stages of myogenesis (e.g., MYOG, MYOF, ACTN4) showed reduced expression (Fig. S8F). This indicates that at 9 days, the hiPSC-derived 3D-TESMs had completed the initial stages of skeletal muscle development (e.g., myoblast fusion and sarcomere formation) and had entered the process of skeletal muscle maturation to a neonatal developmental stage, similar to other 3D-TESM models published to date [26,27,28,29].

Micro-dystrophin rescues the DMD phenotype

We utilized the 3D-TESM system to evaluate the efficacy of micro-dystrophin, a truncated version of DMD that is currently under evaluation as gene therapy in clinical trials, to rescue the Duchenne phenotype in vitro. Since the micro-dystrophin and DMD cDNAs share most of their coding regions, rendering micro-dystrophin susceptible to the DMD-targeting shRNAs, we generated DMD-targeting shRNA#4 to specifically target a region that was only present in the endogenous DMD gene but not in micro-dystrophin (Fig. 4A). Using RT-qPCR with primerset 1 that was designed to only amplify endogenous DMD transcripts, a ~90% reduction of DMD expression was obtained upon knock down using DMD-targeting shRNA #4 (Fig. 4B). Combining DMD-targeting shRNA #4 with a second lentiviral expression construct, expressing either micro-dystrophin or GFP, did not affect expression of endogenous DMD. RT-qPCR with primerset 2 was designed to amplify both micro-dystrophin and endogenous DMD transcripts showed that micro-dystrophin was expressed upon inclusion of the lentiviral expression construct. Similar to the results obtained with the DMD-targeting shRNAs #1–3 (Figs. 2 and 3), DMD-targeting shRNA #4 caused a strong reduction of contractile force in 3D-TESMs (Fig. 4C). Co-expression of micro-dystrophin in DMD shRNA #4 targeted 3D-TESMs resulted in a significant increase in both the force-generating capacity (twitch) and the maximum force-generating capacity (tetanus). However, the micro-dystrophin-induced rescue of contractile force in DMD 3D-TESMs was only partial, reaching 20% (twitch) and 30% (tetanus) of forces reached by the non-targeting control. As a control, co-expression of GFP in DMD shRNA #4-targeted 3D-TESMs had no effect. Whole-mount immunofluorescence corroborated these findings and showed TTN-positive and cross-striated myofibers exclusively in control and in micro-dystrophin-treated DMD knock down 3D-TESMs, while none was present in the DMD knock downs and the knock down in the presence of GFP (Fig. 4D–F). Interestingly, when analyzing the 3D-TESMs with DMD knock downs in the presence of GFP, we observed elongated multinucleated GFP-positive cells that resembled short myofibers that lacked striation, suggesting that DMD knock down allowed myogenic differentiation but resulted in a lack of maturation or a loss of matured myofibers (Fig. 4D).

Fig. 4

Rescue of DMD knock down by micro-dystrophin in 3D-tissue-engineered-skeletal muscle. A Cartoon of endogenous DMD and micro-dystrophin mRNA. The subset of DMD exons used to generate micro-dystrophin is indicated in orange. RT-qPCR primers that are specific for endogenous DMD (primer set 1) or that recognize both endogenous DMD and micro-dystrophin (primer set 2) are indicated. B RT-qPCR analysis of endogenous DMD and micro-dystrophin mRNAs. C Force-generating capacity (twitch stimulation) and maximum force-generating capacity (tetanic stimulation) of 3D-TESMs after 7 days of myogenesis. Data are represented as mean ± SD derived from six independent 3D-TESMs. D Representative images of whole-mount immunofluorescent stainings of experimental conditions from B and C. Red, anti-titin antibody. Green, GFP. Blue, nuclei stained with Hoechst. E Myofiber diameter of 3D-TESMs (n = 40–60 myofibers for each condition). F Myofiber alignment of 3D-TESMs. Myofiber alignment was measured using Δ of the angle of each myofiber relative to the perpendicular axis of the 3D-TESM (n = 40–60 fibers per condition). Statistical significance is indicated relative to the DMD knock down (B and C) or non-targeting control (E and F). For F, significance was calculated using the distribution of the variance. ns, not significant, *p < 0.05, **p < 0.01. #Indicates the conditions where the analyses were prevented by the poor quality of the 3D-TESM tissues

Taken together, these results confirm the specificity of the knock down approach for the modelling of muscular dystrophies in 3D-TESMs, and they demonstrate the utility of the shRNA 3D-TESM system to predict the therapeutic value of novel treatment options in human muscle in vitro.