Molecules to Behaviors


The Rhes protein in the striatum orchestrates the formation of cellular protrusions and travels between neurons 

When cells came into existence millions of years ago, they had a better chance of survival if they could communicate with one another. Cellular protrusions may have emerged as a secure and efficient form of direct communications.

Most cells harbor cellular protrusions(1). For example, bacteria interact with and enter animal cells through filamentous protrusions(2). Choanoflagellates, the living, single-celled relative of metazoa, form intercellular bridges between cells within the colonies(3). In the mammalian visual system, the highly dynamic filopodial protrusions play a critical role in synaptogenesis and circuit maturation(4). In nervous systems, cellular processes (dendrites and axons) between neurons control brain functions. Thus, cellular protrusions may have contributed not only to the welfare of solitary cells but also to the evolution of metazoan traits.

Nanoscaled, membranous, cellular protrusions, tunneling nanotubes (TNT), in which the cargoes are directly transferred from one cell to another, have been recently described in various cell types in vitro and in vivo(1, 5-7). TNT-like protrusions are involved in the delivery of diverse cargoes from cell to cell: vesicles, calcium signals (8), HIV, influenza propagation(9, 10), transport lysosomes, mitochondria, and mRNA(11-17). Notably, TNT-like protrusions are also involved in proteins involved in cancer progression, for example, P-glycoprotein, and in neurodegenerative diseases, for example, prions, tau, and α-synuclein(18-26).

Despite these studies, the presence of TNT-like protrusions or their roles in mammalian brain functions remains unknown. 

Rhes engineers TNT-like cellular protrusions and transport vesicular and protein cargoes between cells

We recently discovered that Rhes, which consists of small GTPase (1-171 aa) and SUMO E3-like domain (171-266 aa)(27, 28), play an essential role in the formation of TNT-like protrusions. Rhes is highly expressed in the MSNs and cholinergic interneurons in the striatum; it is also expressed to some extent in the cortex and the hippocampus(29-31). We ascribed several new roles for Rhes in the striatum in PD(32), inhibitory motor activity(33), and HD(34), which is consistent with independent reports(35-42). We found Rhes promotes posttranslational modification of mHTT with a small ubiquitin-like modifier (SUMO) and promotes cellular toxicity(35). Rhes also plays a critical role in mutant, tau-mediated pathology(43). In addition, a rare, highly conserved de novo mutation (R57H) in Rhes was detected in twins diagnosed on the autistic spectrum(44). Despite these studies, the mechanisms by which Rhes regulates distinct striatal functions remain unclear.

Recently, we found that Rhes travels from cell to cell as vesicular puncta from one cell (donor) to another cell (acceptor) via the TNT-like cellular protrusion “Rhes tunnel” (Fig. 1A, white arrow)(45). Rhes vesicular puncta (blue arrow) do not directly enter the lumen of the acceptor cell; rather, they slide along the plasma membrane (arrowhead) before entering the cytoplasm of the acceptor neuronal cells (Fig. 1A)(45). Scanning electron microscopy (SEM) revealed Rhes tunnels appeared to connect two cells, and their surface showed a seamless transition with the surface of connected cells (Fig. 1B)(45). Furthermore, we discovered that lysosomes, endosomes, and mHTT, the genetic cause of HD, are readily transported in Rhes tunnels(45) (Movie 1). The lipid modification domain of Rhes (C263) and SUMO E3 ligase domain contributes to the formation of the TNT-like Rhes tunnel and cargo transport(45).

Rhes travels from neuron to neuron in vitro and in vivo

To discover whether Rhes can be transported between neurons in vivo, we took advantage of the well-characterized Cre-recombinase system. We used D1RCre; D2REGFP mice and selectively expressed “Cre-On” RFP or RFP-Rhes in D1RCre and found numerous RFP-Rhes, but not RFP-alone, signals within D2REGFPneurons in the striatum (Fig. 1C). Collectively, these data indicated that Rhes travels between neurons in cultured cells and from the D1R neurons to the D2R neurons in vivo. To the best of our knowledge, Rhes is the first striatal-enriched protein demonstrated to travel between neurons in vitro and in vivo. 

Images that show striatal communication via the TNT-like membranous tunnels.

Rhes Tunnel – A Novel Communication Route Between Cells

Movie 1. Shows time-lapse confocal imaging of striatal neurons transfected with GFP-Rhes and mCherry-Rab5a]. The movie is rendered slow motion and texts were added using Movavi studio suite and arrows were added using image J. Ref: Subramaniam, 2020, Bioessays.

Outstanding questions

Our study, along with an increasing body of other studies, underlines that TNT-like protrusions may have a major role in intercellular protein transport; however, their role in the brain biology of the disease remains unknown. Identifying the mechanisms and in vivo functions of Rhes transport has the potential to transform our understanding of striatal neuron signaling and its abnormalities associated with the striatal disease. Our planned research aims to define molecular mechanisms and their functions; so, we are interested in the following. A) To identify and characterize TNT-like protrusions in the brain. B) To decipher whether neuron-to-neuron transport of Rhes mediates behavioral functions, using new Tg mouse models. For example, Rhes­–/– are sensitive to haloperidol-induced catalepsy. We have developed Rhes­–/–D1RCre; D2REGFP to express Rhes in D1R and investigate whether its transport to D2R neurons and alleviate sensitivity to haloperidol-induced catalepsy. We will develop inducible knock-in GFP-Rhes mice to track its movement in the striatum and beyond. C) To identify the protein composition of Rhes tunnels to discover potential “TNT markers,” which are currently unknown. D) To elucidate the entire volume of the Rhes tunnel and cargo (endosome and mHTT) delivery events in striatal cells, using correlative light and electron microscopy (CLEM) and focused ion beam milling, combined with SEM (FIB-SEM). E) To develop “Cre-On” mHTT AAV constructs to express in MSN in the presence and absence of Rhes, to investigate whether Rhes mediates the transports of mHTT in vivo and if so whether such intercellular transport promotes HD-related behaviors.

More than a century ago, whereas Cajal believed that brain cells are not physically connected and function as individual units (neuron theory), Golgi believed that brain cells are indeed physically connected (reticular theory). TNT-like protrusions form physical conduits between neuronal cells, and it is fair to assume that such physical contacts support the reticular theory and Golgi’s view. Moreover, it is also possible that these transient and highly fragile structures are vulnerable to Golgi’s staining and hence may have been disrupted and consequently unavailable for drawing by Cajal(46).

Our work, therefore, proposes an exciting new avenue of research and challenges the fundamental questions about neuron-to-neuron communication in the striatum and neurodegenerative disease, highlighting an unprecedented role of the direct delivery of proteins and cargoes to facilitate brain functions, via the TNT-like cellular protrusions.


Every day motor functions such as walking and running, polishing a nail, playing tennis, or pedaling a bicycle, are controlled by the brain circuits. The striatum, which is comprised of D1R-positive MSNs (also known as a direct pathway) and D2R-positive MSNs (indirect pathway), is central to this complex circuitry and relays signals to and from major parts of the brain(47). MSN dysfunction can lead to the motor abnormalities seen in HD and PD. The striatum is also the main site of action for many clinically approved drugs, such as L-DOPA and haloperidol, to treat psychiatric and neurological diseases(48-54). However, despite many advances, the precise striatal signaling mechanisms that regulate motor functions in the striatum remain unclear(55).

Rhes mediates motor behavior via a “Rhesactome” protein network in the striatum

During my routine long-distance running, I wondered whether there is a formation of protein-protein complexes in the striatum that help coordinate running functions. We knew that striatal protein Rhes acts as an inhibitor of motor activity and Rhes–/– exhibits enhanced behavioral responses, compared to WT, in response to psychostimulants and DA receptor ligands(56-59). We hypothesized that Rhes-mediated inhibitory control over motor behavior in the striatum involves the formation of a dynamic protein complex network(60). To test this, we adopted an amphetamine-induced behavioral activity model. The psychostimulant amphetamine acts at dopamine transporters to increase levels of extracellular dopamine, which in turn acts on striatal, post-synaptic dopamine receptors leading to stimulation of locomotor behaviors (Fig. 2). Amphetamine is widely prescribed to children diagnosed with ADHD, but it can elicit significant psychotic side effects(61). We reasoned that capturing the in vivo protein interaction of Rhes during amphetamine injection would provide a better understanding of the mechanisms by which Rhes functions as a suppressor of amphetamine-induced motor behavior in the striatum.

After 15 mins of the amphetamine injection (at which time the mice reached maximum locomotion), we isolated the striatum, and its protein lysates were subjected to immunoprecipitation, using Rhes IgG or control IgG, followed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis(60). We found dynamic interaction of Rhes with specific proteins “Rhesactome,” which included PDE2A (phosphodiesterase 2A, a protein associated with major depressive disorder), LRRC7 (leucine-rich repeat-containing 7, a protein associated with PD and ADHD), DLG2 (disks large homolog 2, a protein associated with chronic pain), and SHANK3 (SH3 and multiple ankyrin repeat domains 3, a protein implicated in intellectual disability). This result indicated a clear association of protein complexes with motor activity, and we found that this association is regulated by a striatal GEF, RasGRP1, which acted as an inhibitor of Rhes(60). We found that partial Rhes-deficient (Rhes+/–) mice had an enhanced locomotor response to amphetamine, this phenotype being attenuated by coincident depletion of RasGRP1, which affected the composition of the amphetamine-induced Rhesactome(60). Thus, we identified dynamic protein complexes in the striatum whose formation is linked to inhibitory motor activity in mice.

In our planned research, we are interested in understanding the D1R-MSN and D2R-MSN specific inhibitory function and Rhesactome protein complexes of Rhes. As Rhes is highly enriched in the synaptic fractions(60), we are particularly interested in deciphering the role of synapse vs. non-synaptic protein complexes and the underlying mechanisms by which they diminish motor activity, by biochemical and gene knockout strategies. Such studies may aid the development of novel approaches and intracellular “signalosomes” targets to treat various neurological and psychological disorders associated with striatal dysfunction (Fig. 2).

The RasGRP1 upregulation in the striatum sets the stage for propagating motor abnormalities in Parkinson Disease

Administration of L-DOPA provides immediate symptomatic relief to the PD patient. L-DOPA, which increases dopamine in the striatum, activates signaling in the striatum that helps patients to stand up and walk. L-DOPA continues to be a mainstream drug for PD patients, but, within five years, half of the patients using L-DOPA develop an irreversible condition—involuntary repetitive, rapid and jerky movements—commonly called L-DOPA-induced dyskinesia (LID). This paradoxical and bidirectional effect unearths the dilemma in the actions of this drug in neurodegenerative diseases: the complex and incompletely understood striatal signaling mechanisms.

            Our lab research into fundamental biology that promotes LID led us to the discovery of cellular targets, where a mechanism-based intervention can be applied to prevent debilitating motor abnormalities in a mouse model of PD(32, 62). We discovered that RasGRP1 is selectively induced in the striatum by L-DOPA treatment in a Hemi-Parkinson Disease mouse model, as well as in a macaque model of PD(62). We confirmed that RasGRP1 upregulation by L-DOPA is effectively linked to the generation of LID because we found that RasGRP1-null mice show markedly diminished LID compared to WT control PD mice(62).  Importantly, RasGRP1 deletion did not affect the therapeutic benefits of L-DOPA. In terms of signaling mechanisms, we found that RasGRP1 promoted ERK and mTOR as two independent and parallel pathways (62). RasGRP1 is known to induce ERK via H-Ras(63, 64), but we found RasGRP1 also acts as a GEF for RHEB and signal mTOR. Previous studies showed that RasGRP1 can activate distinct small GTPases, such as HRas, M-Ras, R-Ras, K-Ras, and TC21, in vitro(65-67). We found that RasGRP1 and its effectors are expressed in a spatially distinct manner (H-Ras is predominantly synaptic, whereas RHEB is cytoplasmic) in the striatum, which may potentially reduce crosstalk under LID. But the underlying mechanisms by which RasGRP1 triggers LID remain unknown.

We are interested in the D1R- and D2R-specific roles of RasGRP1 in orchestrating LID. For example, LID induces morphological changes in both D1R and D2R MSN. We found RasGRP1 is upregulated in D1R-MSN(62), and we will determine whether RasGRP1 affects D1R and D2R changes in LID. Based on previous studies, we propose to test E2F-1, which is upregulated in the LID striatum (unpublished), as a potential regulator of RasGRP1 and LID. As H-Ras and RHEB are spatially separated in the striatum, we will identify distinct RasGRP1/GTPase complexes that may regulate ERK and mTOR signaling in a spatially compartmentalized manner in the MSNs under LID (Fig. 2).

The mTOR kinase mediates haloperidol-induced cataleptic behavior

The mammalian target of rapamycin (mTOR) is a ubiquitously expressed serine/threonine kinase protein complex (mTORC1 or mTORC2) that orchestrates diverse functions, ranging from embryonic development to aging. However, its brain tissue-specific roles remain less explored. Here, we have identified that the depletion of the mTOR gene in the mice striatum completely prevented the extrapyramidal motor side-effects (catalepsy) induced by the D2R antagonist haloperidol(68), which is the most widely used typical antipsychotic drug. Conversely, a lack of striatal mTOR in mice did not affect catalepsy triggered by the D1R antagonist SCH23390. Along with the lack of cataleptic effects, the administration of haloperidol in mTOR mutants failed to increase the striatal phosphorylation levels of ribosomal protein pS6(S235/236), as seen in control animals(68). To confirm the observations of the genetic approach, we used a pharmacological method and determined that the mTORC1 inhibitor rapamycin has a profound influence upon post-synaptic, D2R-dependent functions. We consistently found that pretreatment with rapamycin entirely prevented (in a time-dependent manner) the haloperidol-induced catalepsy and pS6K(T389) and pS6(S235/236) signaling upregulation in WT mice(68). Collectively, our data indicate that striatal mTORC1 blockade may offer therapeutic benefits with regard to the prevention of D2R-dependent, extrapyramidal motor side-effects of haloperidol in psychiatric illness (Fig. 2).  However, the cell-type-specific role of mTOR complexes or their regulators in the striatum remains unknown.  

            We will generate D1R/D2R-specific deletion of Raptor and D1R/D2R-specific deletion of Rictor for systematic assessment of the MSN cell type-specific role(s) in striatal motor functions.  We found, using LC/MS/MS, several novel binding partners of mTOR in the brain, such as ATP5B, Mic60, Uqcrc1/2, Slc25a3, Ndufs1, PDH-E1, and Cox6c, that bound to mTOR, at an abundance equivalent to the established mTORC1-binding partner, PRAS40. We are interested in understanding the molecular link between mitochondria, mTOR, and motor activity in the brain.

Striatal conductors of motor abnormalities.


The huntingtin protein impedes the movement of ribosomes and inhibits protein synthesis

Ribosomes, which are around 200,000 in a typical mammalian cell, need to move along mRNA (15,000–60,000/cell) to decode the information into proteins(69). Because ribosomes perform this monumental task, it has been suspected that ribosomal movement relies on evolutionarily sculpted molecular mechanisms(70). Ribosome stalling can trigger neurodegeneration(71). However, what controls the ribosomal movement and underlying mechanisms in neurodegenerative disease is unclear.

            Traditional biochemical experiments in yeast and bacteria have implicated many factors, such as codon usage, peptide properties, mRNA structure, tRNA availability, or solely a continuously changing cellular demand, that may be implicated in ribosomal pause(72). We identified huntingtin (HTT), an evolutionarily conserved protein(73), as a potential regulator of ribosome movement. HTT inhibits ribosome movement and reduces protein synthesis. Polyglutamine mutation in HTT (mHTT), which causes HD, gains inhibitory effects and further impedes ribosomal movement, and diminishes protein synthesis(74). These unexpected findings have led us to propose aberrant ribosome stalling as a potential mechanism for the progressive motor and degenerative phenotype in HD.

Diminished protein synthesis and ribosome stalling in HD cells

Polysome profiling results showed HD cells had a bigger polysome (PS) to monosome ratio compared to control cells, but diminished protein synthesis(74). Ribosome run-off experiments with harringtonine(75-77) indicated that ribosomes run more slowly in HD cells (Fig. 3A) and have diminished translation (protein synthesis) (Fig. 3B). These data indicated that ribosome stalling is an ideal mechanism to explain this phenomenon, as it strongly reflects diminished protein synthesis phenotype in HD. Additional experiments revealed that normal HTT is a physiological inhibitor of ribosome movement and inhibitor of protein synthesis, which is exacerbated by mHTT(74) (Fig. 3C).

Next, we isolated mRNAs from the slowly translating PS in HD-homo and control cells, using a harringtonine-based ribosome run-off assay (RRA) followed by mRNA-Seq (PS-RRA-mRNA-Seq). We found that there were ~1,157 targets that showed significantly high mRNA abundance in the PS of HD-homo cells (p < 0.05) compared to the control polysome(74). These data suggest that the translation pool of mRNAs is quite distinct in slowly moving ribosomes in HD.

Furthermore, by employing the high-resolution Ribo-Seq tool, we were able to demonstrate that there are widespread alterations of ribosome occupancy (ribosome-protected fragments, RPF), codon-specific pauses, and altered center of ribosome density in HD cells compared to control cells(74). Intriguingly, global Ribo-Seq also revealed targets such as cGAS, a DNA sensor(78), which showed enhanced ribosome occupancy on exon1 (Fig. 3D). Thus, besides a global slowdown of translation, certain transcripts such as cGAS show enhanced protein synthesis in HD. We confirmed that cGAS is upregulated in the human HD striatum and promotes inflammatory and autophagy responses in cellular HD models(79).

A graph of ribosome traffic

Outstanding questions

Previous evidence supports translational deficits in HD (80, 81), but the mechanisms are unknown. By employing elegant biochemical and molecular biology tools, our lab has demonstrated a new role of HTT and mHTT in controlling protein synthesis, involving slow ribosomal movements. As HTT and mHTT bind to ribosomes, we are interested in identifying their ribosomal interactors. We are particularly interested in identifying “pause-fixing factors,” molecular complexes on the mHTT mRNA, which show enhanced ribosome occupancy on the exon1 preceding CAG repeats(79). HTT consists of 28-32 HEAT [huntingtin, elongation factor 3 (eEF3) 1, protein phosphatase 2A (PP2A) 2, and the yeast PI3‐kinase TOR1] repeats that span the entire protein. Many translation regulators, such as eEF3, eIF4Gs, p97DAP5, GCN1, and FRAP/mTOR, contain HEAT repeats, which interact with rRNA and ribosomal proteins of the small ribosomal subunit, and it is proposed that this plays a role in the translocation of aminoacyl-tRNA from the A site to the P site on the ribosome(82-84). We are interested in elucidating the potential role of HEAT repeats in HTT on ribosome stalling. We are particularly interested in intracellular differences in protein synthesis in heterozygous HD conditions. We test the hypothesis that depending upon the localization and stoichiometry of HTT vs mHTT within the neurons, the translation is either up or downregulated in a compartmentalized manner.