The experiments described in this paper were motivated by the need to improve the efficacy of enzyme replacement therapy in a metabolic myopathy, Pompe disease. Multiple factors contribute to the difficulties in treating skeletal muscle: the sheer mass of muscle tissue; the low density of the receptor responsible for the uptake and delivery of the recombinant enzyme to the lysosomes; and the diversion of the enzyme to the liver. We have previously reported that dysfunctional autophagy and accumulation of autophagic debris in fast muscles of the Pompe model add considerably to these difficulties 
. A profound abnormality in the autophagic pathway also occurs in skeletal muscle in humans with the disease 
. In late-onset patients, as in the mouse model, the enormous autophagic buildup causes greater skeletal muscle damage than the enlarged lysosomes outside the autophagic regions. Therefore, we looked for ways to eliminate the autophagic buildup in the hope of improving the effect of therapy. Indeed, our previous data showed that genetic suppression of autophagy combined with ERT resulted in complete removal of muscle glycogen 
Converting fast to slow fibers, which in the KO do not have autophagic buildup and respond well to therapy, looked like an attractive and more physiological approach. This approach would avoid the need for the genetic suppression of autophagy in skeletal muscle, a condition that has been found to be associated with some abnormalities 
. Both developing and adult skeletal muscle have considerable plasticity with respect to fiber-type switching 
. For example, endurance training stimulates mitochondrial biogenesis and a switch from fast to slow fibers. We have attempted the fiber type conversion in Pompe muscles by transgenic expression of PGC-1α, a factor which drives the slow muscle metabolic program 
As expected, transgenic PGC-1α KO mice, like transgenic PGC-1α WT mice 
, showed conversion of glycolytic fibers into mitochondria-rich oxidative fibers. The converted fibers were atrophic with the degree of atrophy similar to that seen in the KO muscle. In the literature, the data on the effect of PGC-1α on muscle fiber size do not paint a clear picture. Overexpression of PGC-1α was shown to inhibit denervation atrophy and protect skeletal muscle against atrophy induced by expression of FOXO3 
. An anti-atrophic effect of PGC-1α in skeletal muscle was also observed during ageing 
. On the other hand, Miura et al. 
reported that overexpression of PGC-1α in skeletal muscle resulted in a marked depletion of ATP leading to atrophy, especially in type IIB-rich muscles. By contrast, we found a negligible effect of PGC-1α on muscle atrophy in the tgKO muscle.
The fiber type conversion resulted in a striking disappearance of the autophagic buildup. How and why the autophagic buildup is formed in type II fibers of the KO is puzzling. The buildup is very particular in its content and position; it is commonly located in the core of the fibers, and it contains a subset of clustered lysosomes with compromised membranes that appear different from those in the rest of the fiber. It was not clear whether it is the intrinsic properties of this subset of lysosomes that make them incapable of resolving the incoming autophagosomes or whether it is the metabolic/contractile properties of the fiber itself which create this unusual pathology. Since the conversion of fast muscle into muscle with slow profile in the KO mice completely eliminated the autophagic inclusions, the latter scenario appears much more plausible.
Despite the absence of autophagic buildup, the therapy was not successful in tgKO, most likely because of the unexpectedly high glycogen burden leading to lysosomal rupture and release of glycogen into the cytoplasm where its fate and possible effects on muscle function remain to be determined. The reported data on the role of PGC-1α in muscle glycogen and glucose metabolism is somewhat controversial. In transgenic mice, PGC-1α was shown to suppress glucose transport 
, and to decrease insulin-stimulated muscle glucose uptake in mice on a high-fat diet 
. In contrast, expression of PGC-1α resulted in an induction of glucose transport in muscle cells in vitro 
and in vivo
leading to an increase in cytoplasmic glycogen 
. This increase in muscle glycogen stores in skeletal muscle of PGC-1α transgenic mice was also due to the inhibition of glycolysis and down-regulation of glycogen phosphorylase, the enzyme responsible for the degradation of glycogen in the cytoplasm 
We, too, found an increase in glycogen in control PGC-1α transgenic mice. Although this increase is statistically significant compared to the controls, the absolute levels of the accumulated cytoplasmic glycogen are still barely above the detectable threshold. This slight increase, however, leads to an early, massive accumulation of lysosomal glycogen when the transgene is placed on the GAA KO background suggesting that a significant portion of cytoplasmic glycogen rapidly ends up in the lysosomes, where it cannot be digested. The muscle pathology in tgKO is comparable to that in skeletal muscle of infantile Pompe patients who have much greater glycogen load than the knockout mice 
Glycogen is thought to reach the lysosomes, at least in part, via the autophagic pathway. Glycogen autophagy and lysosomal degradation of glycogen to glucose have been shown in the liver 
and in skeletal muscle 
during the early postnatal period as a response to a high demand for this sugar. We have previously demonstrated that genetic suppression of autophagy in skeletal muscle significantly reduced the glycogen burden in KO mice 
. Consistent with these data, we now show that the increase in lysosomal glycogen load in tgKO is associated with an up-regulation of autophagy even greater than that seen in KO muscles.
There are only a limited number of reports on the role of PGC-1α in the regulation of autophagy. A positive correlation between an increase in PGC-1α and autophagy has been shown in lipopolysaccharide-treated neonatal rat cardiomyocytes 
. Similarly, activation of PPAR-γ induced autophagy in breast cancer cells through upregulation of the HIF-1α protein and BNIP3 
. On the other hand, PGC-1α was shown to inhibit autophagic/lysosomal protein degradation in myotubes 
and to suppress autophagy in muscles in aged PGC-1α transgenic mice 
. In our system PGC-1α clearly induced autophagy in both control and KO mice.
Another finding related to the function of PGC-1α itself is a significant increase in the number of lysosomes which became obvious because of the KO background. Thus, our data suggest that PGC-1α is a regulator not only of mitochondrial but also of lysosomal and autophagosomal biogenesis. A combination of enhanced lysosomal capacity and increased production of autophagosomes without autophagic buildup resulted in more efficient disposal of autophagic cargo, in particular Ub-proteins, in tgKO than in KO mice.
Clearance of potentially toxic Ub-substrates is a major problem in neurodegenerative diseases, and the induction of autophagy has emerged as a therapeutic approach designed to rid the cells of these abnormal protein aggregates [reviewed in 
]. Our data on the upregulation of autophagy by PGC-1α suggest that pharmacological activation of this molecule might have a therapeutic benefit for a range of neurodegenerative diseases caused by the accumulation of such aggregates.
Finally, the failure of ERT in tgKO mice may be due to very high, non-physiologic levels of PGC-1α transgene expression, which even exceed the endogenous levels of PGC-1α in type I soleus muscle (). (The levels of PGC-1α expression in tgKO mice are higher compared to those in the “original” PGC-1α strain 
, which was used for crosses to the KO; these high levels could be due to the differences in the background). It has been shown that modest PGC-1α expression in skeletal muscle increased insulin sensitivity 
whereas excessive PGC-1α expression in transgenic mice rendered skeletal muscle resistant to insulin 
. A moderate increase in PGC-1α can be achieved by exercise (which can influence fiber type distribution) as demonstrated in humans 
and in rats 
. Other routes of fiber type conversion may be more successful. It has recently been shown that the expression of the myosin intronic microRNA (miR-499) in skeletal muscle powerfully induced the conversion from fast to a slower myofiber type 
. Thus, the disappointing outcome of therapy in tgKO cannot be viewed as a final verdict on the merits of fiber type conversion.