Reply from Elizabeth Schroder, Brian Hodge, Lance Riley, Xiping Zhang and Karyn Esser

We would like to thank Drs Dyar, Schiaffino and Blaauw for their letter in response to our article ‘Intrinsic muscle clock is necessary for musculoskeletal health’ (Schroder et al. 2015), and for highlighting some of the differences between our mouse models of skeletal muscle knock out of the core molecular clock gene Bmal1. We think the differences between the phenotypes of our mice raise important biological questions, most of which require further research. We would like to point out at the start, however, that we never concluded in our paper that inactivation of the intrinsic muscle clock causes sarcopenia. While we think there could be links between the molecular clock in muscle and ageing it is still too early to make that conclusion. Rather we argue that our results indicate that the endogenous molecular clock in skeletal muscle is necessary for maintenance of skeletal muscle phenotype and, additionally, it contributes to tendon/bone homeostasis. We address several of the specific issues they raised in the paragraphs below.

We are in full agreement with Drs Dyar, Schiaffino and Blaauw that locomotor activity can serve as a non‐photic time cue/Zeitgeber with significant impacts on both the systemic and muscle circadian transcriptome (Zambon et al. 2003; Schroeder et al. 2012; Wolff et al. 2013; Dyar et al. 2015). However, based on our previous work (Hodge et al. 2015), and results presented in Schroder et al., we demonstrate that there is a significant loss of circadian expression in 629 non‐redundant mRNAs in muscle from mice that maintain normal feeding and cage activity. We believe that this number underestimates the effect due to the fact that our collections occurred at a lower temporal resolution, every 4 h, compared to the sensitivity when collections occur every 2 h (Hughes et al. 2007). Thus, we stand by our conclusion that loss of Bmal1 and molecular clock function in adult skeletal muscle is sufficient to induce significant alterations in the circadian transcriptome in a mouse that exhibits normal activity behaviour.

The primary concern raised by Drs Dyar, Schiaffino and Blaauw was that the ex vivo methods we used for functional assessment of the mouse extensor digitorum longus (EDL) muscles were not appropriate so the muscle weakness we reported from our inducible skeletal muscle specific Bmal1 KO mouse (iMSBmal1 KO) may not be valid. In particular they raise concerns with our measurements at 37°C. We do recognize that ex vivo muscle function can be affected by incubation temperature but measurement of specific force of small mouse muscles like the EDL at 35–37°C have been performed and published in peer reviewed journals; some examples include Luff (1981), Hornberger et al. (2005) and Gilliam et al. (2009). As reference, the maximum specific force numbers we obtained for our wildtype/vehicle treated EDL muscles were right in line (see Fig. 3B, C and E) with well‐established values for mouse EDL muscles reported by Brooks & Faulkner (1988) and those provided by the working group of TREAT NMD (Drs Barton, Lynch, Khurana, Grange, Raymackers, Dorchies, Carlson, Brinkmeier & Wells: SOP (ID) no. DMD_M.1.2.00: version 2: updated June 2015). The TREAT NMD document also discusses issues with temperature and notes that caution is required for muscles over 20 mg but all the EDL muscles of wildtype, vehicle treated, Bmal1 KO and the iMSBmal1 KO mice were approximately 10 mg, well below 20 mg. Lastly, we would like to point out that the functional deficit reported in Schroder et al. for the Bmal1 KO EDL muscle at 20–22 weeks is very similar to the weakness we reported for EDL muscles from 14‐week‐old Bmal1 KO EDL muscles in the Andrews et al. paper (Andrews et al. 2010). The reproducibility of EDL specific tension of both wildtype and Bmal1 KO muscles with independent groups of mice performed in two different locations (University of Illinois, Chicago and University of Kentucky) and spanning 5 years also provides strong evidence for the validity of our methods.

The second issue they raised is with our use of adjusting muscle length during twitch contractions for determination of optimal length. While we recognize the work of Close (1972) we are also influenced by the large number of investigators that use the same approach we did for determination of optimal length. More specifically, we point to page 9 of the ‘Measuring isometric force of isolated mouse muscles in vitro’ standard operating procedures of the working group for TREAT NMD (2015). Thus, we are confident that the differences we detected in maximum specific force at 37°C between the EDL muscles of vehicle control vs. iMSBmal1 KO mice and wildtype control vs. Bmal1 KO mice are valid. We suggest that much more work is required to determine the mechanism(s) underlying the decreased specific force in our model.

In regards to our conclusion that skeletal muscle specific loss of Bmal1 induces a similar phenotype to the germline Bmal1 knock out, we want to highlight that we relied on multiple measures to arrive at this point in the discussion. These measures included the analysis of walking gait mechanics at 58 weeks, alizarin red/alcian blue staining of bone and tendon (58 weeks), muscle fibrosis staining (58 weeks) as well as the ex vivo muscle mechanics at both 20 weeks and 58 weeks post recombination. Not only are we confident in the observation of diminished specific force in the iMSBmal1 and Bmal1 KO mice but we put forward that the combination of muscle weakness, muscle fibrosis, altered tendon/bone calcification and gait mechanics provides sufficient data to argue for similarities between the two mouse models.

Drs Dyar, Schiaffino and Blaauw also note that the use of the Mlc1f‐Cre × Bmal1^fl/fl mice is the best model to determine the contribution of muscle to the Bmal1 KO phenotype. The initial goal of our study was to determine the effect of loss of Bmal1 in adult skeletal muscle so our research design and choice of mouse model was selected to address our goal. The fact that we saw similar phenotypes in our iMSBmal1 KO mouse when compared to the Bmal1 KO was an additional observation that we interpret to mean that loss of Bmal1 in adult muscle is sufficient to induce some of the systemic phenotypes seen in the Bmal1 KO mouse.

The primary differences we note between the inducible mouse model used in our study vs. the Dyar study were: (a) the Cre driver mouse model, and (b) the age at time of tamoxifen treatment. Dyar et al. used the Cre transgenic mouse described by Schuler et al. (2005) in which the Cre recombinase is fused with one mutated oestrogen receptor domain (Cre‐ER^t) and inserted in frame into a 90 kb P1‐derived artificial chromosome containing the human skeletal actin gene. We used the Cre recombinase mouse we described in McCarthy et al. (2012) in which we used a 2 kb upstream region of the human skeletal actin gene to direct expression of a modified Cre recombinase that included two mutated murine oestrogen‐binding domains, which were located at the amino and carboxyl termini of Cre (e.g. ER^t‐Cre‐ER^t). It was demonstrated previously that control of the recombination process is more stringent with diminished background when utilizing the ER^t‐Cre‐ER^t system compared with other inducible Cre‐ER^t recombinase systems (Verrou et al. 1999). We have provided extensive PCR results to demonstrate the efficacy of recombination in response to vehicle or tamoxifen treatment across many different skeletal muscles of different types (Hodge et al. 2015). In addition we performed time course collections of muscle for real time PCR analysis to functionally test loss of clock function through statistical analysis on established clock controlled genes (Hodge et al. 2015; Schroder et al. 2015). The combined results demonstrate (a) that we observed very good skeletal muscle specific disruption of Bmal1 only in our tamoxifen treated mice, and (b) that the functional evidence confirms we did indeed significantly disrupt core molecular clock components (Per1, Rora, Rev‐erb) and Dbp, the most well characterized clock controlled gene. Thus, we are confident that our mouse model successfully disrupted the molecular clock mechanism only in skeletal muscle and we stand by our methods and results obtained.

The other option we raise in the discussion of our paper, and it was also addressed by Drs Dyar, Schiaffino and Blaauw, is regarding the age of the mice when treated with tamoxifen to minimize the influence of satellite cell fusion in young mice. On this point, we are in agreement that our current understanding regarding satellite fusion into muscle during mouse maturation is not clear with arguments for myonuclei number, which can be different from satellite cell fusion, being complete by 3 weeks of age (White et al. 2010), in contrast to satellite cell fusion reaching a plateau between 8 and 12 weeks of age (Pawlikowski et al. 2015). All we can highlight is that we waited until after 12 weeks of age to induce recombination to limit the potential for satellite cell dilution of recombined nuclei in the adult muscle fibres. Without a side by side comparison with data demonstrating the extent of recombination, loss of BMAL1 protein and a more thorough presentation of downstream clock genes (e.g. Rev‐erb, Rora and Dbp) from muscle of the imKO mice generated by Dyar et al. it is difficult to completely resolve the underlying differences through which the phenotypes of the iMSBmal1 KO mouse we present in our paper compare to the imKO mouse described by Dyar et al. (2014).