Neurons-derived extracellular vesicles promote neural differentiation of ADSCs: a model to prevent peripheral nerve degeneration

Potential mechanisms involved in neural differentiation of adipocyte derived stem cells (ADSCs) are still unclear. In the present study, extracellular vesicles (EVs) were tested as a potential mechanism involved in the neuronal differentiation of stem cells. In order to address this, ADSCs and neurons (BRC) were established in primary culture and co-culture at three timepoints. Furthermore, we evaluated protein and transcript levels of differentiated ADSCs from the same timepoints, to confirm phenotype change to neuronal linage. Importantly, neuron-derived EVs cargo and EVs originated from co-culture were analyzed and tested in terms of function, such as gene expression and microRNA levels related to the adult neurogenesis process. Ideal neuron-like cells were identified and, therefore, we speculated the in vivo function of these cells in acute sciatic nerve injury. Overall, our data demonstrated that ADSCs in indirect contact with neurons differentiated into neuron-like cells. Neuron-derived EVs appear to play an important role in this process carrying SNAP25, miR-132 and miR-9. Additionally, in vivo neuron-like cells helped in microenvironment modulation probably preventing peripheral nerve injury degeneration. Consequently, our findings provide new insight of future methods of ADSC induction into neuronal linage to be applied in peripheral nerve (PN) injury.

Adipose-derived stem cells (ADSCs) are a type of mesenchymal stem cell (MSC) able to diffferentiated into mesodermal lineages 16 and ectodermal lineages as neuron-like cells 17 . In addition, ADSCs can be modulated by other cells to start its differentiation process 18 and can be an alternative cellular therapy to be used in nerve regeneration and microenvironment modulation after nerve injury [18][19][20][21] .
Due to the above context, In the present study, we sought to evaluate: if small EVs were involved in the neural differentiation induction process of ADSCs when in co-culture with neurons (BRC) 18 ; if the neuron differentiation followed the same gene pathway followed by neural stem cells; and if differentiated ADSCs (neuron-like cells) were functional in vivo as an alternative cell type to be used in nerve injury modulation during the regenerative process.
To answer these questions, we first investigated if neurons could induce the neural differentiation process in ADSCs in a transwell co-culture system 18 , evaluating the microtubule protein beta tubulin III (TUBIII) 22 , as well as the structural protein present in neurons, synaptosomal-associated protein 25 (SNAP25) 23 , a protein required for synapse vesicle formation. Gene expression of Snap25 and microtubule-associated protein 2 (Map2) 24 and transcript levels of microRNA involved in the neuronal differentiation pathway were quantify and analyzed at three time points.
After we had proof of ADSCs neuronal differentiation capacity in co-culture and timeline progression, we next investigated if EVs were involved in this process. In order to do so we looked at the cellular communication by small EVs 4,25,26 , measuring transcript levels of mRNA and microRNAs related with adult neurogenesis 27 , in co-cultured differentiated ADSCs induced by neurons, and in differentiated ADSCs induced only by neuron-derived EVs, to find the most neuron-like cells, which could be used in the in vivo study.
Following this screening, we applied the neuron-like cells to a neuron differentiated ADSCs in vivo in a neurotmesis injury. In vivo analysis of the injured nerve was performed to validate if these cells could assist in PN regeneration or reduce PN degradation during acute period after an extreme injury.
Our results indicate that, by co-culture, neurons stimulated ADSC differentiation. The EVs appear to play a role in this process carrying SNAP25, miR-9 and miR-132, and have neural induction capacity, though lower than the transwell co-culture system. When applied in vivo, axon density of a normal sciatic nerve and the group that received the neuron-like cells were similar, suggesting that these cells reduced PN degeneration after injury.
We were able to detect Map2 and SNAP25 mRNAs, with an increase of Map2 levels in ADSC-CCs on D7 compared to D3 (Fig. 1d) and lower levels of Map2 on D14 than the other days. Importantly, we were not able to detect Snap25 in ADSC-CCs on D3; however, we detected some transcripts on D7 and D14 (Fig. 1d).
Interestingly, SNAP25 with 25 kDa protein band was detected only in neurons and a very weak band of SNAP25 with 25 kDa was detected only in EVs from neurons and from the co-culture media ( Fig. 2e and Supplementary Fig. S2), but strong bands of SNAP25 with 50 kDa and 55 KDa was detected in EVs from neurons and from the co-culture media, but not in EVs from ADSCs (Fig. 2e). Furthermore, after 3 days of culture about 30% of ADSCs cultured with small EVs isolated from BRC were positive for SNAP25 (Fig. 2f), and when in co-culture about 40% of ADSC-CC were SNAP25 + .
Regarding the EVs tracking analysis, PalmtdTomato reporter was efficiently inserted into EVs from BRC +PalmtdTomato (Exo-BRC) as demonstrated in Fig. 2g. EVs from BRC +PalmtdTomato cultured with ADSCs were observed together with ADSCs but at this point it was not possible to affirm if there was EVs uptake by ADSC or if EVs were in the membrane of ADSCs (Fig. 2h,i). In addition, the transwell co-culture system allowed transfer of EVs from BRC +PalmtdTomato to ADSCs (Fig. 2j).

Discussion
In the presented work, we emphasized the importance of looking at the neuron-derived EVs' capacity to induce neuronal differentiation in ADSCs. Our results demonstrated that there were more differentiated ADSC-CCs at D7, and this process appears to be mediated by neuron secretions. Importantly, SNAP25, miR-9, and miR-132 were found in neuron-derived EVs, which were taken up by ADSC-CCs, as shown by the EV +PalmtdTomato system. Lastly, we analyzed the in vivo proprieties of differentiated ADSCs. The ADSC-CCs D7 applied in vivo improved limb functionality after neurotmesis.
In accordance with our findings, recent studies using chemical compounds demonstrated similar expression of neural markers, such as Map2 and SNAP25 in stem cells differentiated into neuronal lineages 24,28,29 . Regarding the extracellular vesicles findings, the neuron-secreted EVs were involved in ADSCs neuronal differentiation as recently pointed out by Takeda et al. 30 .
A novel aspect of this current work is the delivery of SNAP25 by neuron-secreted EVs, which might play a role in neural differentiation. The SNAP25 protein normally weighs 25 kDa, but our findings demonstrated the presence of 50 kDa and 55 kDa protein bands in EVs, which is described in other studies as probably a protein complex 31 .
Another very interesting finding of our study was the confirmation of a crosstalk between neuron-derived EVs and ADSCs by the presence of neuron-secreted EVs +PalmtdTomato within ADSC-CCs. In fact, studies have confirmed that EV-mediated communication is a dynamic and multidirectional mechanism of cell communication, mediating the delivery of functional components 12,13 .
In the context of the transcripts related to adult neurogenesis, in ADSCs co-cultured with BRCs, levels of Numbl and Mbd1 were higher in ADSC-CCs D3 than in all of the other induced ADSCs, which is an indication of neurogenesis as it can be observed in NSCs 28,32 . Importantly, Mbd1, which helps to maintain the potentiality of neural stem cells by restricting the differentiation genes and initiating the activation of these genes 33 , was higher in ADSC-CCs D3 and could be an indication of activation of a similar adult neurogenesis pathway 33 . Additionally, ADSC-CCs D7 presented higher levels of Ezh2, a gene responsible for the differentiation pathway of NSCs 34 , than ADSC-CCs D3, which indicates an important transcript change in these cells after a certain period of co-culture under neuron induction. Furthermore, Mib1, responsible for the neural differentiation process 35 , and Creb, involved with neuron maturation and function 36 , were most highly expressed in ADSC-CCs D7, suggestive of neuron maturation of these co-cultured cells.
When the same transcripts were measured in the ADSCs induced by neuron-derived EVs, other transcript profiles were observed.
Ptbp1 is an RNA regulatory protein and assists in the control of alternative splicing during the process of neural differentiation in progenitor stem cells 37 , controls synapse-related protein genes 38 , and is a target gene of miR-124 39 and levels of Bdnf, which is a neurotrophin that regulates the entire neuronal environment, aiding in the process of neuronal maturation present in neurogenesis 40 were higher in the induced ADSCs by EVs; which could be an indication that the neural differentiation process was being activated.
Importantly, Rest was in low levels in both cells, which indicates that such cells are possibly involved in the process of neurogenesis. Rest 41 and the CoRest/MeCP2 42 complex regulate genes related to quiescence and proliferation prior to the onset of differentiation of NSCs into mature neurons 43 , and when Rest is silenced, there are higher rates of reprogrammed fibroblasts for neural differentiation 41 . It is known that Rest inhibition is partially modulated by up-regulation of miR-9 and miR-124 41 , but this inhibitory pathway was not observed in the induced ADSCs.
Neuron-secreted EVs appear to play a role in this phenotype change from mesenchymal stem cell to neuron-like cells. MiR-9 is highly detected in the brain 44,45 and its high levels can reduce Tlx expression, improving the differentiation of NSCs 46 . In this study, the ADSCs which received neuron-secreted EVs, had high levels of miR-9 and Tlx on D3 and miR-9 was detected in the MVs from the neurons, suggesting the transfer of this miRNA. Additionally, miR-132 was highly abundant in all induced ADSCs and presented in all neuron-secreted EVs.
Therapies for the regeneration of peripheral nerve have mostly been evaluated using mesenchymal stem cells under or already differentiated in neuron-like cells 47 , similar to this study, but a highlight of our study was that probably there was a delay of the Wallerian degeneration process during the acute period after an extreme nerve injury by using the neuron-like cells developed in this study. Several studies have shown that even after surgery, full recovery of the peripheral nerve is not a reality 48,49 , even when using cells or allografts 50 in an effort to delay the degenerative process as an alternative to gain time and improve recovery after surgery.
In conclusion, neurons are able to induce the neural differentiation of ADSCs; confirmed by ADSC-CCs phenotypic changes, as well as the expression of neuronal transcripts. Furthermore, we were able to demonstrate the transfer of neuron-derived EVs to ADSCs using an EV-reporter construct, and the delivery of SNAP25, miR-9, and miR-132, suggesting that cell-secreted vesicles from neurons were related to the neural differentiation of ADSCs. Lastly, the strategy to reduce the process of degeneration using differentiated MSC as the cell type developed in this study can be useful when an increased time during pre-surgery is required due to problems such as infection at the site of injury or non-vascularization, or when the total section of the peripheral nerve is necessary. The use of pre-differentiated cells in neuron-like cells may be an alternative in these cases, since it appears that they aid in improving the function of the affected limb after injury.

Experiment 1-ADSCs differentiation into neuronal lineage and neuron-derived EVs involvement in neural differentiation. Neurons promote ADSC differentiation when in indirect contact.
Neurons and mesenchymal stem cell isolation and culture. We used the FVB mouse strain from the Hemocenter of Ribeirao Preto's vivarium, Ribeirao Preto, SP, Brazil to collect ADSCs and neurons (BRC) (N = 10, 2 to 3-month old). Cells were isolated and primary cultured was performed. In addition, ADSC was characterized as mesenchymal stem cell by colony unit forming, flow cytometry for mesenchymal markers, and neural differentiation of the ADSCs to prove stem capacity before conducting experiments. For more details: Supplementary Information.
Immunocytochemistry and flow cytometry analysis. In order to determine the change in phenotype, we performed immunocytochemistry and flow cytometry of SNAP25 and TUBIII (N = 7). For immunocytochemistry, quantification of the percentage of positive cells for SNAP25 and TUBIII markers was obtained according to the literature 51 . For more details: Supplementary Information.
Transduction with eGFP and sorting. Lentiviral production with eGFP was performed by lipofection of 293FT cells (Invitrogen, Carlsbad, California, USA). The ADSCs were cultured in 60 mm petri dishes until attaining 60% confluence, followed by 3 mL incubation with 6 μg/mL of virions particles, and then sorted by flow cytometry using FACSAria Cell Sorter supported by DiVa V.6.1.2 software (BD Biosciences, San Jose, CA, USA) using filters adjusted to the light emission of 525 nm for FITC.
Messenger RNA and micro RNA extraction, reverse transcription of messenger or micro RNA and RT-qPCR analysis. To analyze genes and pathways related to the neurogenesis (Supplementary information Tables 1 and 2) messenger RNA and micro RNAs quantification was performed. For these procedures, three biological and two or more technical replicates were used. For more details: Supplementary Information. Total RNA was isolated from all differentiated ADSCs, control cells (ADSC and BRC), and EVs using TRIzol LS (Life Technologies, Carlsbad, California, USA), following the manufacturer's protocol. RNA quality was determined by Nanodrop (Thermo Fisher Scientific, Carlsbad, California, USA) based on 260 and 280-ratio analysis.
In summary, for mRNA cDNA synthesis was performed using the High Capacity Reverse Transcription Kit (Applied Biosystems, Carlsbad, CA, USA), following the manufacturer's protocol (genes and primers are listed in Supplementary Information, Tables 1 and 2). For microRNA, miScript PCR System (num. 218193, Qiagen, Hilden, Germany) was used according to the manufacturer's protocol, with a total of 100 ng of total RNA (microRNAs listed, and primers listed, Tables 3 and 4

of Supplementary Information).
Neuron-derived EVs promote neural differentiation of ADSC. Isolation and characterization of small EVs. From control media and co-culture media of the three periods (D3, D7, D14) small EVs were isolated using ExoQuick, their shape was analyzed using transmitted microscopy and ALIX, CD63, CD9, HSP70, cytochrome, calnexin, COX IV was also analyzed by western blot [52][53][54][55] . For nanoparticle quantification we used NanoSight NS300 (Malvern Instruments, Malvern, UK) with a laser of 405 nm and NanoSight NTA software v2.3 to evaluate the EVs from D3 and D7. EVs were diluted in DPBS (1:100). Five videos of 30 seconds were recorded with 14-camera level at 37 °C for each sample after calibration with 100 nm of beads (Malvern Instruments, Malvern, UK) to ensure equipment accuracy as indicated by the manufacturer. The concentration was defined from the average of the recorded movies. For more details: Supplementary Information. PalmtdTomato labeled EVs. We used the plasmid vector kindly provided by Dr. Charles P. Lai (National Tsing Hua University, Hsinchu, Taiwan) 56 , containing the tandem dimer Tomato (tdTomato) fused at NH2-termini with a palmitoylation signal (PalmtdTomato) to label EVs. To summarize, neurons were transduced with a lentivirus vector encoding PalmtdTomato for 24 hours, then the media was changed and EV experiments were performed.
Small EVs funtion analysis. To confirm EVs communication from neurons to ADSCs, we performed two culture conditions: (i) ADSC cultured with small extracellular vesicles (Exo) +PalmtdTomato from BRC +PalmtdTomato , Exos were isolated using ExoQuick and added to ADSC GFP+ or ADSC GFP-; (ii) co-culture of ADSC GFP-and BRC +PalmtdTomato using transwell system.
Additionally, the number of ADCS positive for SNAP25 from these two culture conditions and the presence or absence of SNAP25 in the small EVs were measured.
EVs messenger RNA and micro RNA extraction, reverse transcription of messenger or micro RNA and RT-qPCR analysis. To analyze genes and pathways related to the neurogenesis (suplemmentary information Tables 1 and 2) transcript level analysis of ADSC cultured with neuron-derived small extracellular vesicles (ADSC + BRCExo) and ADSC cultured with neuron-derived large vesicles (ADSC + BRCMV) from D3 and D7 was performed. We selected D3 because it was the suggested day when ADSCs were starting the neuronal differentiation and D7 because it was the day with a higher rate of differentiated ADSC-CC and according to our transcript analyzes with ADSC similar to BRC. Transcript levels of specific microRNAs in control cells (ADSC and BRC) and in BRC EVs was used as baseline.