Fragile X syndrome (FXS) is the leading inherited intellectual disability and the most prevalent single-gene cause of autism. It arises from a lack of FMRP expression, which significantly impacts the synthesis of synaptic proteins and particularly the morphology of dendritic spines. In FXS neurons dendritic spines tend to be long and filopodial, indicating an immature state. Current research suggests that FXS and similar neurodevelopmental disorders can be viewed as synapse-related diseases.

Matrix metalloproteinase 9 (MMP-9) is an enzyme produced at the synapse and secreted to the extracellular space in an neuronal activity-dependent manner. MMP-9 degrades various substrates including gelatin, protein components of the extracellular matrix and synaptic cell adhesion molecules. The local release of MMP-9 at synapses facilitates targeted extracellular proteolysis of specific substrates, driving morphological alterations in dendritic spines. Synaptic translation of MMP-9 mRNA is regulated by FMRP.

In Fmr1 knockout mice the excessive protein synthesis at the synapse results in heightened enzymatic activity of MMP-9 potentially accounting for the observed spine abnormalities in FXS. Interestingly, inhibiting MMP-9 activity by application of minocycline, the tetracycline analog, to Fmr1 knockout mice can rescue the spine phenotype both in vivo and in cultured neurons. This effect was accompanied by behavioral improvement in young knockout mice towards better performance in general cognition, activity and anxiety tests.

The extracellular localization of MMP-9 makes it an especially attractive potential target for therapeutic pharmacological intervention. Understanding of the molecular mechanism of synaptic MMP-9 expression and its dysregulation in FXS has led to rational therapeutic approaches using MMP-9 inhibitors such as minocycline.

Synapses are highly specialized sites of interaction between two neurons. Adhesion proteins ensure a stable synaptic connection. Neuroligins and neurexins, for instance, are crucial not only in maintaining this physical link but also in orchestrating the synaptic structure and ensuring effective synaptic transmission (Sudhof, 2008, Sudhof, 2021).

Synaptic functions are strictly connected to dendritic spine structure, and many prevalent neurological disorders are linked to the structural changes of dendritic spines (Bosch and Hayashi, 2012, Rochefort and Konnerth, 2012). Basically, synapse morphology reflects the strength and stability of a synapse. Typically, spines with larger heads signify more robust synaptic transmission and are generally more stable than their smaller, long filopodial-type counterparts (Yoshihara, De Roo, & Muller, 2009). Dendritic spines are very dynamic structures, and the activity of neuronal connections can modulate their shape (Tonnesen & Nagerl, 2016).

Characterizing FXS is the loss of the fragile X mental retardation protein (FMRP), a vital player in mRNA metabolism in neurons (Hagerman et al., 2017). At synapses FMRP, an RNA binding protein, regulates translation elongation through reversible stalling of ribosomes (Darnell et al., 2011). One of the characteristic features of FXS neurons is the abnormal morphology of their dendritic spines (Rudelli et al., 1985). When observed under the microscope in brain slices or cultured Fmr1 knockout mice neurons, the dendritic spines appear longer and thinner than in wild-type neurons which is a signature of being immature (Comery et al., 1997).

On a molecular scale, a profound consequence of the FMRP gene’s loss of expression is the dysregulation of synaptic protein synthesis (Bassell & Warren, 2008). The level of local translation is elevated in Fmr1 knockout mice at the basal conditions and, importantly, it cannot be induced by neuronal stimulation (Lu et al., 2004, Muddashetty et al., 2007, Zalfa et al., 2003). Another distinguishing feature of FXS is the altered synaptic plasticity triggered by metabotropic glutamate receptors (mGluRs), leading to an enhancement of LTD (Bear, Huber, & Warren, 2004). Many identified FMRP target mRNAs encode postsynaptic and presynaptic proteins including mGluR5 and NMDA receptor components, structural and scaffolding proteins, neurotransmitter receptors, and signaling molecules (Darnell et al., 2011). The increased local synthesis of MMP-9 mRNA in FXS plays a pivotal role in shaping the morphology of dendritic spines and will be discussed below.

Local regulation of protein synthesis in neuronal dendrites has emerged as a leading research focus mainly because of its importance in synaptic homeostasis, plasticity, and memory formation (Sun et al., 2021). Synaptic translation ensures precise spatial and temporal control over protein synthesis, coupled with the rapid regulatory impact of the newly synthesized proteins on spine morphology and receptor signaling. This facilitates a unique response from each synapse to stimulation, underpinning their plasticity. Dysregulation of local protein synthesis results in profound dendritic spine pathology and contributes to abnormalities in specific neuronal circuits, which in turn may underlie the socio-cognitive impairments characteristic of neurodevelopmental disorders (Liu-Yesucevitz et al., 2011, Penzes et al., 2011, Richter et al., 2015).

Matrix metalloproteinase-9 (MMP-9) belongs to the family of endopeptidases that is expressed in neurons and secreted at the synapse in response to neuronal activity (Dzwonek et al., 2004, Rivera et al., 2010). It regulates the pericellular environment through cleavage of its protein components and plays a critical role in regulating spine morphology and synaptic plasticity. MMP9 is involved in the activity-dependent reorganization of spine architecture (Dityatev et al., 2010, Michaluk et al., 2011).

The transcription of MMP-9 mRNA in neurons is regulated by the activity-related transcription factor c-Fos (Jaworski, Kalita, & Knapska, 2018). Therefore the expression of MMP-9 was correlated with different forms of synaptic plasticity induced either by physiological stimulation or, particularly strongly, by seizures (Szklarczyk, Lapinska, Rylski, McKay, & Kaczmarek, 2002). It was also shown that MMP-9 is rapidly and transiently activated in an NMDA receptor and protein synthesis-dependent manner during the late phase of long-term potentiation (L-LTP) (Bozdagi et al., 2007, Gorkiewicz et al., 2015, Nagy et al., 2006, Wang et al., 2008).

The first observation showing the presence of mRNA encoding MMP-9 in dendrites was published in 2002 by Szklarczyk and colleagues. The authors demonstrated that kainite-induced seizures result in the upregulation of MMP-9 mRNA, protein, and enzymatic activity in the molecular layer of rat hippocampal dentate gyrus (Szklarczyk et al., 2002). This finding was substantiated by high-resolution electron microscopy showing MMP-9 mRNA, detected by non-radioactive in situ hybridization, within dendritic spines of hippocampal neurons after kainite (Konopacki et al., 2007). The fact that MMP-9 mRNA was observed in neuronal processes suggested that it could be locally translated in synapses as has been shown for other synaptic proteins (Biever et al., 2020).

Further research has produced direct evidence of MMP-9 mRNA synaptic translation in response to neuronal stimulation, providing major insight into mechanisms of MMP-9-mediated synaptic plasticity (Dziembowska et al., 2012). The activity-dependent transport of MMP-9 mRNA to the dendrites of dentate gyrus granule cells following LTP induction was visualized using fluorescent in situ hybridization. MMP-9 mRNA was more abundant in the dendrites of dentate gyrus granule cells two hours after LTP induced by high-frequency stimulation (HFS) of the medial perforant path indicating that MMP-9 mRNA was actively transported to the dendrites after the stimulation resulting in synaptic plasticity. MMP-9 mRNA was also shown to be associated with synaptic polyribosomes and de novo MMP-9 protein synthesis was observed in isolated synaptoneurosomes, followed by the secretion of the newly synthesized enzyme on the synapse (Dziembowska et al., 2012). The enzyme, proMMP-9, is secreted in its latent form and requires proteolytic cleavage by proteases like tPA for activation (Michaluk and Kaczmarek, 2007, Michaluk et al., 2011). Additionally, its activity can be modulated through co-secretion with its natural inhibitor, tissue inhibitor of metalloproteinases-1 (TIMP1) (Roderfeld et al., 2007).

As noted earlier, synaptic cell-adhesion molecules that bridge pre- and post-synaptic neurons play a pivotal role in the formation and maintenance of synapses. To modify the shape of dendritic spines, these connections need partial dismantling and reconstruction. The local release of MMP-9 at synapses facilitates targeted extracellular proteolysis of specific substrates, driving morphological alterations in dendritic spines. This process is enhanced in response to neuronal stimulation (Dziembowska et al., 2012).

The substrates of MMP-9 include protein components of the extracellular matrix and synaptic cell adhesion molecules such as neuroligins 1–3 (Chmielewska, Kuzniewska, Milek, Urbanska, & Dziembowska, 2019), nectin 3 (van der Kooij et al., 2014), ICAM5 (Tian et al., 2007), beta dystroglycan and proteolytic activation of cell surface receptors, such as beta1 integrins (Bajor and Kaczmarek, 2013, Michaluk et al., 2011, Peixoto et al., 2012, Wang et al., 2008).

The number and diversity of MMP-9 substrates argue for its role in synaptic changes associated with synapse structure and function. In this context, MMP-9 can be considered a unique mediator of both structural and functional forms of synaptic plasticity.

The Fmr1 knockout mice lacking the expression of FMRP provide an excellent animal model for the studies of fragile X syndrome (Santos, Kanellopoulos, & Bagni, 2014). In 2009 Iryna Ethel’s group discovered enhanced gelatinolytic activity of MMP-9 in the Fmr1 knockout mice brains compared to their wild-type littermates (Bilousova et al., 2009). Also, the treatment of cultured hippocampal neurons with MMP-9 induced the appearance of immature, filopodial spine profile. Subsequently, our findings indicated that MMP-9 mRNA undergoes local translation at the synapse upon stimulation (Dziembowska et al., 2012). The next question that we asked was if the elevated level of MMP-9 observed in Fmr1 knockout mice the effect of dysregulated local MMP-9 translation at the synapse. Indeed, our studies established that MMP-9 mRNA is an FMRP target and the increased MMP-9 gelatinolytic activity observed in synapses isolated from Fmr1 knockout mice results from enhanced synaptic translation of MMP-9 mRNA. In silico analysis of the murine MMP-9 mRNA sequence revealed a set of paired guanines (G), which can form an intramolecular G-quartet within the 3’UTR of the transcript. The paired guanines are RNA sequence motifs bound by FMRP (Darnell et al., 2001, Schaeffer et al., 2001, Subramanian et al., 2011). The direct interaction between MMP-9 mRNA and FMRP in the brain, was confirmed by RNA co-immunoprecipitation with anti-FMRP 7G1–1 antibody on synaptoneurosomes isolated from the cerebral cortex and hippocampi of WT and Fmr1 knockout mice (Janusz et al., 2013). Moreover, the in situ hybridization (for MMP-9 mRNA) combined with immunofluorescence with anti-FMRP antibody showed that the RNA-protein complexes dissociate in response to stimulation with the mGluR1 metabotropic glutamate receptor agonist DHPG (3,5-dihydroxyphenylglycine). The translational regulation of MMP-9 mRNA by FMRP was studied by polyribosome profiling. After DHPG stimulation in wild types, the MMP-9 mRNA shifted to the actively translating heavy polyribosomal fraction, in agreement with the occurrence of fine-tuned local protein synthesis. However, in synaptoneurosomes from Fmr1 knockout mice there was no increase of MMP-9 mRNA translation in response to DHPG stimulation, indicating a lack of response to synaptic stimulation. These findings indicate that the absence of FMRP leads to increased basal translation and impaired activity-mediated translation of MMP-9 mRNA at the synapse (Janusz et al., 2013).

The studies on animal models of FXS were further substantiated by the discovery of elevated MMP-9 levels in postmortem brain from FXS patients. In the same samples the authors found increased phosphorylation of the mRNA 5′ cap binding protein, eukaryotic initiation factor 4E (eIF4E). Genetic or pharmacological reduction of eIF4E phosphorylation rescued core behavioral deficits, synaptic plasticity alterations, and dendritic spine morphology defects via reducing exaggerated translation of Mmp9 mRNA in Fmr1(−/y) mice, whereas MMP-9 overexpression produced several FXS-like phenotypes (Gkogkas et al., 2014).

Other studies confirmed the role of MMP-9 in structural plasticity pointing to the importance of tight regulation of local translation of its mRNA. Sidhu, Dansie, Hickmott, Ethell, and Ethell (2014) have confirmed our data arguing for the role of MMP-9 in the development of FXS-associated defects in Fmr1 knockout mice. Genetic disruption of Mmp-9 in Fmr1 knockout mice (double Fmr1 and MMP-9 knockout mouse) rescued key aspects of Fmr1 deficiency, including dendritic spine abnormalities and abnormal mGluR5-dependent LTD, as well as aberrant behaviors in open field and social novelty tests. Remarkably, MMP-9 deficiency also corrected non-neural features of Fmr1 deficiency, specifically macroorchidism, indicating that MMP-9 dysregulation contributes to FXS-associated abnormalities outside the CNS. Further, MMP-9 deficiency suppressed elevations of Akt, mammalian target of rapamycin, and eukaryotic translation initiation factor 4E phosphorylation seen in Fmr1 knockout mice, which are also associated with other autistic spectrum disorders. These findings establish that MMP-9 is critical to the mechanisms responsible for neural and non-neural aspects of the FXS phenotype.

To date, probably most of the mRNAs regulated by FMRP have been identified predominantly in neurons, as highlighted by Darnell et al. (2011). However, the precise role of many of these mRNAs in modulating synaptic plasticity remains to be elucidated. MMP-9, on the other hand, is a protein distinctly recognized for its role in governing structural plasticity. In Fmr1 knockout mice, a dysregulated local translation of MMP-9 mRNA at the synapses leads to increased enzymatic activity of MMP-9. This elevation might be a primary factor contributing to the observed morphological abnormalities in the dendritic spines of FXS (Janusz et al., 2013).

Further studies of molecular mechanisms underlying synaptic expression of MMP-9 revealed the presence of a putative binding site for the microRNA 132 (miR-132) in the 3’UTR of the transcript. Interestingly, it was located close to the G pair-rich sequence (G-quartet) bound by FMRP. Edbauer and coworkers (2010) have shown the interaction of FMRP with many microRNAs in neurons using immunoprecipitation with an anti-FMRP antibody. In this experiment, miR-132 directly interacts with FMRP (Edbauer et al., 2010). The overexpression of miR-132 in neurons led to the reduction of the MMP-9 level secreted by the cells to the culture media, indicating that miR-132 overexpression downregulated the level of MMP-9 (Jasinska et al., 2016).

Expression of miR-132 in neurons is induced by the transcription factor CREB, which was first demonstrated in rat cortical neurons and then in mouse neurons stimulated with neurotrophins (Remenyi et al., 2010, Vo et al., 2005). CREB is known to be involved in neuronal survival, maturation, differentiation and function, but it also controls developmental plasticity, memory formation, adaptive behavior, and drug addiction and regulates circadian rhythms (Lonze & Ginty, 2002). Therefore, it is unsurprising that miR-132 was studied mainly in the neuronal context and is now one of the best described microRNA with regard to a function in neuronal plasticity.

The first microRNA shown to regulate dendritic spine morphology was miR-134 (Schratt et al., 2006). The authors showed that miR-134 targets the Limk1 kinase, which regulates ADF/cofilin interactions with the actin cytoskeleton, and that its overexpression leads to the shrinkage of dendritic spines. Other microRNAs have also been shown to affect spine architecture (Eacker, Dawson, & Dawson, 2013). Only one microRNA—miR-132—has been described to control dendritic spine enlargement and increase synaptic transmission. miR-132 was shown to induce neurite outgrowth and to modulate the dendritic morphology of both immature cortical and hippocampal neurons by controlling the expression of its targets such as p250GAP (Impey et al., 2010, Wayman et al., 2008), a GTPase-activating protein involved in neuronal differentiation or methyl CpG-binding protein 2 (MeCP2) (Klein et al., 2007).

A subset of miRNAs are localized to dendrites (Kye et al., 2007, Schratt et al., 2006), where local translation may affect dendritic spine morphology. Knockdown of miR-132 activity in newborn neurons in the adult hippocampus or in neurons of the visual cortex leads to a reduction in stable mushroom spines and an increase in filopodia-type spines (Edbauer et al., 2010, Luikart et al., 2011, Mellios et al., 2011). These structural changes were accompanied by decreased frequency, but not amplitude of mEPSCs. Conversely, infusion of miR-132 mimics into the visual cortex following monocular deprivation increased mushroom-type spines and eliminated ocular-dominance-associated plasticity (Tognini, Putignano, Coatti, & Pizzorusso, 2011). Our preliminary data strongly suggest that MMP-9 can be the new target protein contributing to the miR-132-dependent structural plasticity of dendritic spines.

Interestingly, similar to MMP-9, miR-132 has been implicated in the regulation of structural plasticity of dendritic spines in neurons. Generally, the overexpression of miR-132 in hippocampal neurons increases the width of the dendritic spines heads (Edbauer et al., 2010). This effect can be explained by inhibiting MMP-9 activity, which causes elongation of the dendritic spines (Michaluk et al., 2011).

The exact molecular mechanism responsible for microRNAs’ regulation of mRNA stability is unknown and can differ depending on the cell type. Generally, microRNAs regulate their target mRNAs by RISC-mediated deadenylation and decay or by the reversible inhibition of translation (Filipowicz, Bhattacharyya, & Sonenberg, 2008). In neurons where mRNAs are transported to be locally translated in dendrites it is easy to imagine the translational repression as a potent mechanism that controls spatial and temporal protein expression. Indeed, such regulation was shown for PSD-95 mRNA and miR-125a (Muddashetty et al., 2011). The overexpression of miR-132 in neurons significantly affected dendritic spines shape and morphology. The inhibition of MMP-9 endogenous activity in the Fmr1 knockout neurons by overexpressing miR-132 led to an increase in the width of dendritic spines and corrected their aberrant morphology (Jasinska et al., 2016).

The studies on cultured neurons confirmed that miR-132 facilitates local translation of the MMP-9 protein in cooperation with FMRP. However, the regulation of mRNA expression by microRNAs is very complex—one type of microRNA can be targeted to and regulate many different mRNA molecules. Therefore, we decided to go further in our studies and directly prove that miR-132 regulates the synthesis of MMP-9 in the brain. To this end, we constructed a transgenic mouse with a mutation in the 3’UTR region of MMP-9 mRNA where miR-132 binds,that prevents the interaction of these two molecules. Thanks to this new unique mouse model, we were able to identify the involvement of miR-132 in the regulation of MMP-9 expression in visual cortex plasticity during postnatal mouse development. Our results show for the first time a direct link between synaptic plasticity with the regulation of local MMP-9 mRNA translation by miR-132 (Jasinska et al., 2016).

The discovery of an additional mechanism for regulating the local expression of MMP-9 at the synapse by microRNAs suggests the precise space, time, and protein level is tightly controlled in neurons.

Dysregulation of MMP-9 mRNA local translation in Fmr1 knockout mice leads to excessive protein synthesis at the synapse, which can at least partially explain spine dysmorphology in FXS. Applying minocycline, the tetracycline analog inhibiting MMP-9 activity, to Fmr1 knockout mice rescued the abnormal spine phenotype in vivo and in cultured neurons (Sidhu et al., 2014). This effect was accompanied by behavioral improvement in young knockout mice towards better performance in general cognition, activity, and anxiety tests. Moreover, the reduction of exaggerated MMP-9 mRNA translation in Fmr1(−/y) mice by the inhibition of translation initiation (reduction of eIF4E phosphorylation) also rescued core behavioral deficits of these mice as well as synaptic plasticity alterations and dendritic spine morphology defects (Gkogkas et al., 2014) (Fig. 1).

The first clinical trials were conducted to test the minocycline in patients with fragile X and patients with autism, suggest that minocycline is relatively well tolerated and provides functional benefits to patients (Paribello et al., 2010, Utari et al., 2010). The clinical study, conducted at the MIND Institute (Sacramento, USA) with the main goal of determining if minocycline had an effect on cognition or behavior in FXS patients, reported high plasma levels of MMP-9 in individuals with FXS when compared to healthy subjects (appropriately matched for age control group) (Dziembowska et al., 2013).

The randomized, double-blind, placebo-controlled clinical trial with minocycline was conducted at the MIND institute. The results of this study suggest that, in humans, the levels of MMP-9 are lowered by minocycline and that changes in expression are positively associated with improvement based on clinical measures, in some cases. This trial demonstrated benefits from minocycline in the majority of patients in the Clinical Global Impression Scale (CGI-I) and in 2 categories of the visual analog scales (VAS) including anxiety and mood and a miscellaneous category that included social benefits and toilet training. Only a small number of patients did not show improvements with minocycline treatment, and side effects were not different than what was experienced by placebo (Dziembowska et al., 2013).