Tetraspanins are evolutionarily-conserved membrane proteins that associate dynamically with numerous partner proteins in tetraspanin enriched microdomains (TEMs) and regulate cell morphology, motility, and signaling.1
Many of the 33 mammalian tetraspanins are expressed in the nervous system, playing a role in neurite outgrowth, myelination and glia cell number control.2
Among them TSPAN7, encoded by the TM4SF2
gene, is directly associated with cognitive defects in humans. TM4SF2
mutations, including TM4SF2
inactivation by an X:2 balanced translocation, a premature stop codon TGA (gly218-to-ter),3
and a 2-bp deletion (564delGT) resulting in a premature stop codon at position 1924
are directly associated with non-syndromic intellectual disability. The gly218-to-ter nonsense mutation and the 2-bp deletion predict a truncated protein lacking the fourth transmembrane domain and cytoplasmic C-terminal tail.
In our recent paper5
we have identified TSPAN7 as a new player in synaptic maturation and function, showing that it localizes at excitatory synapses in cultured embryonic rat neurons.
In particular, we found that TSPAN7 overexpression promotes the formation of filopodia and dendritic spines, while TSPAN7 knockdown causes a reduction in spine head size, and increases spine motility and turnover–consistent with the results of an in vivo study showing that small filopodial-like spines are more dynamic and undergo more rapid remodelling than larger spines.6
These morphological abnormalities are similar to those reported in some mentally retarded patients7
and also in various animal models of mental disorders7,8
including oligophrenin-1-, Il-1RAPL1-, and PAK3- knockout mice;9
similar morphological abnormalities have also been observed in dissociated neuronal cultures using specific siRNAs to knockdown the transcripts of the same proteins.9
It is noteworthy that changes in spine morphology are closely linked to changes in the strength of synaptic connections.10
Thus, mushroom spines have larger, more complex postsynaptic densities (PSDs), with higher glutamate receptor densities, than smaller spines;11
furthermore spine size correlates with PSD size.12
The fact that TSPAN7 knockdown affects spine morphology suggests it is causing defects in PSDs and reducing the stability of postsynaptic structures. Consistent with this, TSPAN7 knockdown reduced the expression level of postsynaptic markers, in particular of AMPARs, and such decreases were accompanied by a reduction in the number of functional synaptic AMPARs as shown by a decrease in mEPSCs amplitude and frequency. The measure of PPR is not altered upon TSPAN7 knockdown, consistent with an increase in the number of silent synapses, i.e., lacking AMPAR, in silenced neurons. By contrast NMDAR currents were not affected.
It is important to examine how TSPAN7 may affect PSD composition and AMPAR currents. As tetraspanins associate with several proteins and regulate their trafficking and signaling,1
the function of these molecules has to be considered in the light of their specific interacting/binding proteins. It has previously been reported that TSPAN7 associates with phosphatidylinositol 4-kinase (PI4K) type II activity.13
In this paper, we identified other TSPAN7 associated proteins: PICK1, that is a direct interactor and GluA2 AMPAR subunit and β1-integrin that are associated to TSPAN7 (). We showed that the C-terminal tail of TSPAN7 interacts with the PICK1 PDZ domain. We also found that TSPAN7 associates with GluA2/3 and β1-integrin, and that following TSPAN7 knockdown, PICK1 associated more tightly with AMPAR, while β1-integrin no longer did so. These findings suggest that these proteins form a single macromolecular complex, whose composition depends on TSPAN7.
Figure 1. The cartoon shows a synaptic complex composed by TSPAN7 and the proteins: PICK1, AMPAR, integrins and PI4K. In particular, TSPAN7 associates with AMPAR GluA2 subunit, β1-integrin, PI4K and, through its C-terminal tail, directly (more ...)
TSPAN7 could, therefore, regulate signaling pathways important for synaptic transmission and plasticity via its association with β1-integrin, PICK1 and AMPA. PICK1 is a well-established regulator of AMPAR trafficking and of the number of AMPA receptors at synapses.14
It has been involved in AMPAR-regulated trafficking in long-term depression (LTD) in hippocampal neurons15,16
and the cerebellum.17,18
Changes in AMPAR trafficking are crucial to synaptic plasticity and therefore to learning and memory. In fact, in various synapse types, LTD involves endocytosis of AMPARs from the plasma membrane, whereas long-term potentiation (LTP) involves insertion of additional AMPARs to the surface.19,20
We found that GluA2 underwent increased internalization in response to TSPAN7 knockdown, with reduction in surface GluA2, that was accompanied by decreased mEPSCs; by contrast endocytosis block with dynasore restored GluA2 surface levels, pointing out AMPAR trafficking as the mechanism involved. The modulation of PICK1 and TSPAN7 expression independently leads to opposite effects on GluR2 distribution, with PICK1 overexpression promoting—and PICK1 silencing reducing—GluA2 internalization;21,22
while the TSPAN7 expression level positively correlates with surface localization of GluA2 containing AMPAR. Rebalancing the relative expression of the two proteins by double knockdown or double overexpression experiments reinstate GluA2 distribution to control levels. Interestingly TSPAN7 deleted for its C-terminus and thus lacking the amino acids responsible for PICK1 binding, has no effect on GluA2 trafficking. These data are consistent with a model in which TSPAN7 competes with AMPAR for PICK1 binding and identify TSPAN7 as a new modulator of AMPAR trafficking through PICK1.
TSPAN7’s direct binding to PICK1 suggests a ready explanation for the defects in AMPAR trafficking observed upon TSPAN silencing, but does not explain the effects of TSPAN7 on filopodia and dendritic spine morphology. A possible explanation is via the regulation of actin filaments. The initiation and elongation of filopodia as well as spine formation are known to depend on the precisely regulated polymerization and cross-linking of actin filaments.23
Although PICK1 is also involved in actin polymerization,24
we showed that PICK1 and TSPAN7 regulate spine morphology independently. A possibility is that TSPAN7 influences actin filaments via an association with PI4K13
and/or via its association with β1-integrin—as shown in this and other studies.25
Both PI4K and β1-integrin are known regulators of actin dynamics.26,27
By linking these two proteins, TSPAN7 could recruit them to a complex on the plasma membrane that coordinates actin dynamics with the assembly or organization of postsynaptic AMPA receptors.
We investigated the presence of β1-integrin in the TSPAN7 complex since tetraspanins are known to interact with integrins (especially those containing the beta1 chain)25
and β1-integrin is involved in LTP,28
suggesting involvement in learning and memory. In fact, mice with postnatal deletion of β1-integrin are impaired in working memory.28
We found that β1-integrin is part of the TSPAN7-AMPAR complex.
Other studies have shown that regulatory interactions between integrins and ion channels are common and often involve formation of a multi-protein complex2
and that glutamate receptors are targets for integrins.2
In dissociated neurons and cultured hippocampal slices, AMPA treatment has been shown to increase the surface expression of β1- and α5-integrin.29
Our data indicate that TSPAN7 serves as a functional connector between integrins and AMPARs with implications for synaptic transmission and plasticity particularly since AMPARs mediate the majority of fast excitatory synaptic transmission in the brain, and changes in AMPAR number or function, can result in alterations in synaptic strength19
The changes we have observed in synapse morphology and physiology resulting from TSPAN7 silencing point to TSPAN7 having a critical role in synapse function, consistent with the intellectual disability found in some patients with TSPAN7 mutations. Our findings allow us to speculate that TSPAN7-related intellectual disability may occur because TSPAN7 loss results in alteration of dendritic filopodia and spine morphology and synapse physiology, likely mediated by reduced stabilization of AMPAR at dendritic spines and also by lack of binding of β1-integrin to the AMPAR-PICK1 complex.