The calpain-calpastatin (calp/cast) system is constituted by a family of intracellular Ca²⁺-dependent cysteine proteases and a single natural inhibitor called calpastatin (Goll et al., 2003). Among the multiple calpain forms present in mammals, six are tissue-specific, and the others are expressed across all tissues, including CNS (Ono and Sorimachi, 2012). The most studied are the ubiquitous calpain-1 (Calp-1) and calpain-2 (Calp-2), which were previously indicated as μ- and m-calpain based on their [Ca²⁺] requirement for activation in vitro. Both Calp-1 and Calp-2 are heterodimers containing a common 28-kDa regulatory subunit (calpain small subunit 1, CSS1, also named CAPN4) and a large 80-kDa catalytic subunit (CAPN1 and CAPN2, respectively) which shares 55%–65% sequence homology between the two enzymes. An increase in [Ca²⁺]_i promotes the dissociation of the heterodimer and activation by autoproteolysis of the protease. This same autolysis reduces the mass of both Calp-1 and Calp-2 catalytic subunits to 76-kDa and 78-kDa, respectively, and CSS1 mass to 18-kDa (Cong et al., 1989; Ono and Sorimachi, 2012). Calpastatin (Cast) is the specific inhibitor of the ubiquitous calpains and, although it exists as a single gene [in humans ref. Inazawa et al. (1990)], multiple protein forms resulting from both the use of different promoters (Lee et al., 1992; Takano et al., 2000; Raynaud et al., 2005) and alternative splicing mechanisms are produced (Lin et al., 2004; De Tullio et al., 2007). Full-length Cast polypeptide consists of four repetitive inhibitory domains and an NH₂-region containing XL- and/or L-domain, which has no inhibitory activity and undergoes alternative splicing (Figure 1). Cast can also be selectively cleaved by calpain in fragments retaining inhibitory activity. Such a process leads to the amplification of the inhibitory effect of Cast, a condition required to block and regulate the activated calpain molecules generated following [Ca²⁺]_i increase (De Tullio et al., 2000; Stifanese et al., 2010). In normal conditions the efficient control of calpain activation/activity is also achieved through cytosolic diffusion of calpastatin in colocalization with calpain (De Tullio et al., 2009) and an increase in the inhibitor biosynthesis (Stifanese et al., 2010).

Ubiquitous calpains in addition to being involved in normal processes of neurogenesis, cell proliferation, differentiation, long-term potentiation, and memory (Amini et al., 2013; Machado et al., 2015; Wang et al., 2020; Baudry et al., 2021), play a role in various CNS pathologies generally associated to altered Ca²⁺-homeostasis (Stifanese et al., 2010; Rao et al., 2016; Mahaman et al., 2019; Muruzheva et al., 2021). In this context, the role of calpains has been studied through inhibition of calpain activity either by overexpressing calpastatin (Schoch et al., 2012; Yu et al., 2020) or using synthetic calpain inhibitors (Kerstein et al., 2017; Hu et al., 2019; Ivleva et al., 2022). Calp-1 or Calp-2 knockout models were used to explore the role of a specific calpain isoform in the CNS (Wang et al., 2020; Baudry et al., 2021; Wang et al., 2021). Calpains are also involved in the development and progression of different cancer types including glioblastoma (GBM) (Storr et al., 2011a; Moretti et al., 2014; Nian and Ma, 2021; Shapovalov et al., 2022), the most common and malignant adult brain tumor, a deadly disease without effective treatment (Ostrom et al., 2019; Tan et al., 2020). GBM is characterized by high cellular heterogeneity consisting of various cell populations including glioblastoma stem cells (GSCs), intrinsically resistant to chemo- and radio-therapy, sustaining cancer initiation, maintenance and recurrence (Bao et al., 2006). GSCs are able to self-renew and differentiate into a heterogeneous progeny of differentiated non-stem GBM cells, constituting the tumor bulk. GSCs and more differentiated GBM cells are two phenotypic states functionally defined, with distinct transcriptional, epigenetic, and metabolic profiles leading to different proliferation rates, migratory ability and chemo-sensitivity (Gimple et al., 2019). However, since plasticity of GBM cell populations is bidirectional and mainly modulated by environmental factors, successful GSC eradication should be paralleled by non-GSCs impairment. Deciphering novel potential molecular pathways, such as those involving the calp/cast system sustaining GBM cell stemness, could prospectively serve as prognostic and/or therapeutic factors for GBM. For example, recent findings suggest that calpain-2 contributes to temozolomide (the first-line chemotherapy drug for GBM patients) resistance in GBM cells, opening future implications for GBM patient management (Stillger et al., 2023).

Upregulation of Calp and Cast mRNAs was first reported in a single GBM compared to normal brain tissue (Ray et al., 2002). In a recent paper, Guo et al. (2022) demonstrated that in human glioblastoma cells, Calp-1, through the degradation of T291-phosphorylated IkBα, leads to NF-kB activation, which, in turn, by upregulating PD-L1 expression, results in increased tumor immune evasion (Guo et al., 2022). In this context, the phosphorylation of IkBα is catalyzed by hexokinase-2 (HK2), a glycolytic enzyme that was found overexpressed in numerous types of cancer (Garcia et al., 2019). Overexpression of CSS1, described in glioma tissues, was associated with low miR-124 expression and proposed as a potential adverse prognostic marker (Cai et al., 2014). CSS1 downregulation or miR-124 mimics suppressed migration and invasion of GBM cell lines (Cai et al., 2016). In zebrafish, Calp-2 expression is involved in GBM cell brain invasion (Lal et al., 2012). Vo et al. (2019a) demonstrated that a member of the nuclear factor I family, NFI, involved in GBM cell migration and infiltration, regulates the expression of the various Cast transcripts (i.e., Type III) by targeting the usage of alternative promoters in the CAST gene. The phosphorylation of NFI is regulated by the phosphatase calcineurin which is activated by calpain-mediated digestion (Vo et al., 2019b). Calpain exerts antiproliferative and antimigratory effects in established GBM cell lines by inducing the cleavage of filamin A (Cai et al., 2020), while Cast phosphorylation promotes radiation resistance in GSCs by increasing the activity of calpains (Bassett et al., 2018).

We have previously identified a new Cast variant belonging to Type III isoform (GenBank MK585065) but lacking the inhibitory units in a glioblastoma, a meningioma and four breast cancer biopsies as well as in PBMC from healthy donors (Sparatore et al., 2019). This “anomalous” Cast was named hcast 3-25 because the sequence of its transcript includes exon 2 and 3 of the L-domain and then “skips” into exon 25 immediately after the fourth inhibitory domain without changing the reading frame (Figure 1). Our in vitro experiments using the recombinant protein evidenced that hcast 3-25 associates with native Calp-1 in the absence of Ca²⁺ and protects the activated form from degradation when Ca²⁺ is present (Sparatore et al., 2019).

All these observations prompted us to explore if hcast 3-25 could be included as a new member in the calpain/calpastatin system acting as a positive modulator of calpains in GBM exploiting the availability of several fully characterized patient-derived GBM stem cultures.

In the present study we investigated a possible correlation between the expression of hcast 3-25 and the other components of the calp/cast system in a set of matched stem and differentiated GBM cultures. We determined the mRNA and protein levels of Calp-1, Calp-2, CSS1 and Cast and the expression of hcast 3-25 has been evaluated in comparison with the other full-length Casts using primers spanning the specific XL-L region of each Cast isoform. We assessed the activation of the calp/cast system in cell crude lysates by detecting both Calp-1 and Calp-2 autoproteolyzed forms and Cast as well as CSS1 species resulting from calpain-mediated cleavage. The activity of calpains has been evaluated as the production of the specific calpain-derived α-II spectrin fragments (Czogalla and Sikorski, 2005). On this basis, new insights into the calp/cast system in GBM stemness or the phenotypic shift to a more differentiated state may identify novel molecular vulnerabilities of GBM cells.

The modulation of the calp/cast system correlates with multiple physiological and pathological processes, including cancer. While the role of calpains is mainly ascribed to the proteolytic cleavage of selected targets, calpastatin biological function is intrinsically related to calpain inhibition. Indeed, very little is known about the calpastatin regulatory/non-inhibitory domain localized upstream of the inhibitory units and undergoing alternative splicing. Besides full-length calpastatins responsible for calpain inhibition, we identified a new calpastatin named hcast 3-25 in human CNS, devoid of the four repetitive inhibitory units and containing the N- (exon 1u-3) and C-terminal (exons 25-29) of calpastatin Type III (Sparatore et al., 2019). The challenge of studying hcast 3-25 ex vivo is the impossibility of measuring its activity on calpain, and the lack of a specific antibody against the protein. Furthermore, since calpastatin undergoes calpain-dependent digestion (De Tullio et al., 2000; Stifanese et al., 2010), it is impossible to distinguish among native low Mr Casts (i.e., hcast 3-25) and the various fragments containing the identical C-terminal end, common to all calpastatins.

In human GBM, the transcriptomic data elaborated from the TCGA and GTEx gene expression datasets show that calpain and calpastatin are upregulated in GBM tissues compared to normal brain tissue, likely suggesting a role of the calp/cast system in tumor development and aggressiveness. Recent proteomic data (Stillger et al., 2023), although obtained from a limited series of GBMs, indicate upregulation of Calp-2 and, to a lesser extent, of Calp-1, compared to adjacent brain tissue.

Interestingly, Calp-2 gene expression was also significantly higher in LGGs than in normal tissue, while Calp-1, CSS1, and Cast levels were comparable to normal samples and lower than in GBM. LGGs include a heterogeneous group of tumors generally characterized by well-differentiated cells, low growth rate and invasiveness, less aggressive behavior, and better prognosis than GBM. Therefore, we first analyzed the mRNA amounts of the different Calp and Cast isoforms in classical preclinical models of LGG (Res186, Res259) and GBM (U87) cell lines. Overall, calpains and calpastatins were upregulated in LGG cell lines, poorly mirroring the tumor sample expression profile. Glioma cell lines, derived from patients’ samples but grown indefinitely in serum-enriched media, scarcely recapitulate the original tumors’ in vivo genotypic and phenotypic features, and lose the tumorigenic cancer stem cell population (Lee et al., 2006). Thus, we used low-passage serum-free stem cell cultures derived from primary GBMs as a more reliable model for preclinical studies. We analyzed 14 patient-derived GSC cultures by RNA-seq, confirming high transcription levels of the calp/cast system components, and then we focused on hcast 3-25 mRNA in 7 of the 14 cultures.

While Calp-1, Calp-2 and CSS1 mRNAs were higher in cell lines than in GSC cultures, the Cast isoforms were variably expressed among the individual GSC cultures, indicating that GSCs retain patient-specific expression. These observations are consistent with a differential usage of the three promoters upstream of exons 1xa, 1xb, and 1u, each one containing transcription factor-binding motifs sensible to a variety of stimuli (Sensky et al., 2006; Masilamani et al., 2022). Therefore, transcription of the different Cast variants changes according to stimuli received during GBM growth and development, producing molecules with inhibitory activity that differ in the non-inhibitory domain. Calpastatin can also undergo internal splicing in the L-domain by removing single exons. Such a process generates other full-length isoforms starting with the same Met but having a peculiar set of exons in the L-domain (Sensky et al., 2006; De Tullio et al., 2007; Sparatore et al., 2019).

In the case of hcast 3-25, the internal splicing occurs to Type III and generates a variant lacking the inhibitory units but containing exon 3. Exon 3 is prevalently skipped in porcine and human full-length Casts (Parr et al., 2004; Sparatore et al., 2019), whereas it is differentially transcribed in beef muscle (Motter et al., 2021). Storr et al. found an association between Cast-containing exon 3 and lymphovascular invasion in breast cancer (Storr et al., 2011b). Interestingly, hcast 3-25 and Type III show the most significant fluctuations among the seven GSCs, and their reciprocal ratio within the same sample was unrelated. Hcast 3-25 is the least represented variant among the different Casts isoforms, suggesting that, due to the lack of inhibitory units, it may play a decisive role in supporting the activity of these proteases. Indeed, calpains exert their function despite cells containing calpastatin activity exceeding the activity of the two calpains together (Goll et al., 2003). In GSCs, hcast 3-25 expression correlated neither with the mRNA levels of the other Cast isoforms nor with the transcription of calpains.

Based on previous experiments using recombinant hcast 3-25 (Sparatore et al., 2019), we hypothesized that the GSCs showing top levels of hcast 3-25 mRNA (GBM19 and GBM23) could be characterized by increased activation/activity of the calp/cast system. Due to colocalization of calpains and the inhibitory effect of calpastatins, the activity of the two proteases cannot be directly measured in crude lysates, thus we used the autoproteolyzed forms of the catalytic subunits together with CSS1 digestion and Cast low Mr species, as markers of calpain activation (Stifanese et al., 2010), and α-II spectrin, a specific calpain target, as a marker of calpain activity (Czogalla and Sikorski, 2005; Higuchi et al., 2012).

Several Authors demonstrated that CSS1 modulates the proliferation and metastasization of different types of human cancer cells (Li et al., 2014; Cai et al., 2016; Cheng et al., 2018), including gliomas (Cai et al., 2014; Cai et al., 2016). A recent meta-analysis reports that overexpression of CSS1 protein is closely related to worse outcomes and more aggressive tumor features in cancer patients (Tang S. et al., 2019). We observed that all the GSC cultures show similar levels of both CSS1 mRNA and protein (except GBM39 that contained very low amounts of protein). Notably, all the cultures were obtained from high grade gliomas.

It has been widely demonstrated that overexpression of the calpastatin inhibitory unit helps to relieve the effects of excessive calpain activity (Aluja et al., 2022), also in the CNS (Schoch et al., 2012; Diepenbroek et al., 2014; Yu et al., 2020). Since calpastatin behaves simultaneously as a calpain inhibitor and substrate, the presence of discrete Cast fragments indicates that the calp/cast system is activated. Hcast 3-25, by protecting the activated protease from degradation, might support the limited digestion of calpastatin (producing free inhibitory domains) and other calpain targets.

In this context, evaluation of calpain-specific α-II spectrin 150 and 145 kDa breakdown products in the CNS (Czogalla and Sikorski, 2005; Higuchi et al., 2012) allowed us to measure the actual calpain activity in each GSC culture at the moment of processing. Indeed, the presence of Cast fragments indicates the activation of the calp/cast system. However, α-II spectrin digestion might occur only if calpastatin inhibitory units do not occupy the calpain catalytic site.

Thereby, GBM4 and GMB39, in which we observed the highest cleavage of α-II spectrin, contained the lowest CSS1 and/or Cast protein amounts (GBM4 contain high levels of native Cast but the digested forms are also present, see Figure 3G). As expected, due to the different genotype of the patient-derived cultures, each GSC displays a different activation pattern of the calp/cast system evaluated as the content of native and activated Calp-1 and Calp-2 forms. Detection of similar amounts of CSS1 and the prevalence of native Cast revealed that in these cultures, the calp/cast system seems to act in basal conditions (Stifanese et al., 2010).

Overall, no correlation between the extent of calpain activation/activity and the transcription of hcast 3-25 was detected, but data confirm that the effective cleavage of calpain targets results from a balanced combination of the various calpain and calpastatin isoforms and species (including “the invisible” hcast 3-25) available in each GSC culture, and not from their absolute amount.

In light of these observations, we rescheduled our strategy to obtain information about hcast 3-25 by selecting four GBM stem cultures showing low-medium-high transcription of hcast 3-25 (see Figure 2C) and explored the behavior of the calp/cast system following their differentiation. Indeed, the ability to differentiate (“non-GSCs”) is an expected although not essential property of GSCs. Moreover, the contribution of the stemness-to-differentiation axis to cellular features in GBM support the modulation of GSC tumor-propagating potential (Piccirillo et al., 2006; Stricker et al., 2013).

Differentiation involves calpain activity also in the CNS (Baudry and Bi, 2016; Shu et al., 2018; Nian and Ma, 2021) and transcription of cast Type III is upregulated during migration and differentiation of high-grade gliomas (Vo et al., 2019a). We found that except hcast 3-25, transcription of all the other components of the calp/cast system increased following differentiation. Concerning cast Type III, our observations are in agreement with the data obtained by Vo et al. (2019a), but we also found upregulation of Type I and Type II. Significantly, considering the transcription of the full-length Casts containing inhibitory domains (cast tot) in differentiated GBM cells, we observed that it increases up to 3-fold compared to the stem condition (see Figure 4), a minimal change compared to the upregulation of Calp-1 (up to 40-fold) and Calp-2 catalytic subunits. These data, combined with those obtained by WB analysis, indicate that differentiated cells compared to stem counterparts show: 1) increased Calp-1 activation, as demonstrated not only by the presence of the 75-kDa form but also by detection of low Mr inactive fragments (Ono and Sorimachi, 2012). The increased amounts of 80-kDa species indicate that Calp-1 consumption is efficiently counterbalanced by synthesis; 2) increased Calp-2 activation. Although autoproteolyzed Calp-2 was undetectable, the native protein decreased despite upregulated synthesis (Figure 5), suggesting that during differentiation, Calp-2 activation/degradation is faster than synthesis of new molecules (Baudry and Bi, 2016; Shu et al., 2018; Nian and Ma, 2021); 3) increased CSS1 mRNA level not matching with the corresponding protein amount. The lack of correspondence may be explained by several observations indicating that CSS1, in addition of being the regulatory subunit of calpain, is also related to tumor malignancy. Indeed, CSS1 plays a role in regulating the migration of several types of cancer cells (Li et al., 2014; Chen et al., 2019), including nasopharyngeal carcinomas cells infected by Epstein-Barr virus (Zheng et al., 2020). CSS1 is also targeted by miR-124, and its downregulation or miR-124 mimics suppress the migration and invasion of glioma cells in vitro (Cai et al., 2016). It is still unclear whether the CSS1 molecules acting in these tumors come from the pool that dissociated from calpain catalytic subunit following Ca²⁺ exposure or have been produced in excess and act independently from the calp/cast system; 4) increased Cast digestion. Cast has been cleaved by calpain since in all the four GBM differentiated cultures the amount of native protein is decreased compared to the stem condition. Since we used an anti-calpastatin antibody recognizing the fourth inhibitory domain, the low Mr fragments in GBM diff (Figure 5G) play an active role in maintaining the regulation of the calp/cast system. Indeed, even though both calpains are more activated following differentiation, the percentage of digested α-II spectrin is approximately comparable to that observed in the GBM stem (Supplementary Figure S5).

Although we have analyzed only four distinct GBM cultures, and the heterogeneity of the individual GBM may influence results, our data suggest that during differentiation calpains are differently activated and the calp/cast system seems to act in a physiological controlled condition. While differentiation induces upregulation of all the calp/cast system components, hcast 3-25 mRNA increases or decreases with respect to the stem state, depending on the GBM culture considered, suggesting that it plays a role finely tuned not exclusively related with differentiation.

We then induced overexpression of hcast 3-25 by transfecting hcast 3-25-V5 in GBM3 and GBM5 stem cells, the two GSC cultures that showed upregulation of hcast 3-25 following differentiation (Figure 4). Our observations demonstrate that hcast 3-25-V5 effectively competes with the endogenous calpastatins for the binding to calpain thus maintaining the protease active on its targets, ex vivo.

Transfection of primary and stem cells is often challenging, and we found it difficult to optimize methods for hard-to-transfect GBM3 and GBM5 stem cells, testing different commercially available transfection reagents. Efficient transfection of GBM3 was achieved with Lipofectamine™ Stem reagent, and Nucleofection technology was used for GBM5. Although the differences in hcast 3-25-V5 transcript levels between the two GBM cultures (hcast 3-25-V5 mRNA was increased 6.4-fold in GBM3 and 58.52-fold in GBM5 compared to respective empty-V5 controls), transient overexpression of hcast 3-25-V5 did not affect the transcription of the other calp/cast system components, GFAP and SOX2, as observed in 15-days-serum differentiated GBM3 and GBM5 cells. Significantly, in both transfected cultures, hcast 3-25-V5 overexpression affected the corresponding proteins since all the calpain targets analyzed were digested independently from the amount of hcast 3-25-V5 produced. Hcast 3-25 overexpression was also associated with significantly increased protein amounts of the astroglial marker GFAP and decreased levels of the stem marker SOX2. In this context, although GBM5 expressed the highest amount of hcast 3-25-V5 protein, we observed the same trend in both transfected stem cultures. As predictable, changes in GFAP and SOX2 mRNA and protein expression differ between 72 h-transfected cultures and 15-day FBS-differentiated counterparts in terms of absolute values since FBS contains a pool of various growth factors and components, which can modulate multiple molecular and cellular mechanisms underlying the differentiation process. Thus, it is possible to explain why a more efficient transfection of hcast 3-25 does not lead to increased cleavage of calpain targets and a more robust modulation of the stem and differentiation markers. Considering that upon FBS-induced differentiation, we observed that two GBM cultures out of four are characterized by upregulation of hcast 3-25, and the other two show downregulation (see Figure 4), we might hypothesize that this “anomalous” calpastatin is in some way involved in the differentiation of GSCs.

Finally, it is helpful to underline that cancer stem cells are not clonal entities, defined by distinct transcriptomic signatures and functional properties, but rather a cellular state driven by environmental conditions. In particular, stem cell-associated phenotypic heterogeneity is a feature of GBM; therefore in vitro inducible transition from a more undifferentiated to a differentiated state only mimics a complex in vivo dynamic process (Cui et al., 2023).

We suggest that hcast 3-25, preserving instead of inhibiting calpain activity, provides an additional regulatory mechanism of the calp/cast system acting in combination with the full-length calpastatins. Because it maintains calpain activity, hcast 3-25 could be important for prolonging calpain action in the presence of plenty of inhibitory units deriving from calpain-dependent cleavage. Interestingly, we previously demonstrated that the L-domain also plays a crucial role in regulating calpastatin’s inhibitory efficiency by determining the formation of complexes in which calpain can be fully inhibited or fully active (De Tullio et al., 2014).

This aspect is worth further investigation since calpastatin availability for calpain inhibition is related also to post-translational modifications, such as phosphorylation in the regulatory domain involved in calpastatin diffusion/aggregation (De Tullio et al., 2009). A recent paper reports that PKA-mediated phosphorylation of Ser 133 in human calpastatin domain I (in exon 8) enhances the inhibition of calpain and, consequently, postmortem calpain-dependent meat tenderization results decreased (Bai et al., 2023). The absence of inhibitory domains and the ability to extend the duration of calpain activity makes hcast 3-25 a promising tool for upregulating calpain in postmortem conditions.

In conclusion, a deeper understanding of the hcast 3-25 transcriptional regulation during GBM cell differentiation, a process leading to impairment of proliferation rate and invasiveness of GBM tumor cells could be critical for identifying new patient-specific GBM stem cell vulnerability and driving GBM stem cell fate. Due to the high undifferentiated profile of GBM, differentiation therapy has been proposed as a promising strategy for modulating GBM aggressiveness (Jin et al., 2017; Stevanovic et al., 2021; Liu et al., 2023). Thus, hcast 3-25 may represent an additional regulator of GBM stem cell phenotype. Therefore, deciphering novel molecular pathways such as the calp/cast system, sustaining GBM cell stemness, could prospectively serve as a tool for prognostic and/or therapeutic purposes, targeting therapy-resistant GBM stem cells, the population which fuels GBM initiation and recurrence.

In Figure 8 we propose a schematic model summarizing the calpain/calpastatin system in basal conditions considering GBM stem cultures, their differentiated counterparts, and hcast-3-25-overexpressing GBM stem cells. This model is based on previous experiments carried out in vitro (De Tullio et al., 2014; De Tullio et al., 2018; Sparatore et al., 2019) where we have observed that recombinant hcast 3-25 associates to native calpain competing for this binding with the Casts species containing the regulatory domain, and importantly, preserves the activity of calpain in the presence of Ca²⁺.