Cx43-mediated sorting of miRNAs into extracellular vesicles

Abstract

Through the exchange of lipids, proteins, and nucleic acids, extracellular vesicles (EV) allow for cell-cell communication across distant cells and tissues to regulate a wide range of physiological and pathological processes. Although some molecular mediators have been discovered, the mechanisms underlying the selective sorting of miRNAs into EV remain elusive. Previous studies demonstrated that connexin43 (Cx43) forms functional channels at the EV surface, mediating the communication with recipient cells. Here, we show that Cx43 participates in the selective sorting of miRNAs into EV through a process that can also involve RNA-binding proteins. We provide evidence that Cx43 can directly bind to specific miRNAs, namely those containing stable secondary structure elements, including miR-133b. Furthermore, Cx43 facilitates the delivery of EV-miRNAs into recipient cells. Phenotypically, we show that Cx43-mediated EV-miRNAs sorting modulates autophagy. Overall, our study ascribes another biological role to Cx43, that is, the selective incorporation of miRNAs into EV, which potentially modulates multiple biological processes in target cells and may have implications for human health and disease.

Keywords: Connexin43; heterogeneous nuclear ribonucleoproteins; intercellular communication; miRNA; selective sorting.

Figure 1. Cx43 modulates the selective sorting of miRNAs into extracellular vesicles

1. EV produced by HEK293 cells expressing or not Cx43 (EV^Cx43+ or EV^Cx43−) were purified by differential ultracentrifugation. miRNAs expression profile analysis in EV^Cx43+ (represented as fold increase over EV^Cx43−).

2. Pathway enrichment analysis for the predicted targets of the over‐represented miRNAs in Cx43‐containing EV.

3. HEK293 cells were incubated with 5 µg of purified EV^Cx43− or EV^Cx43+ for 24 h, in the presence or absence of Bafilomycin A1 (Baf) in the last 6 h. The levels of LC3 were analyzed by WB and are depicted on graph (LC3‐II; n = 4 biological replicates).

4. miRNA‐centered regulatory network established by the miRNAs enriched in Cx43‐containing EV. The figure represents predicted targets that are simultaneously regulated by 3 or more of the selected miRNAs. The size of the symbols is proportional to the number of established functional connections. miRNA‐regulatory are represented by grey edges. Protein–protein interactions obtained from STRING database are depicted in black lines.

5. HEK293 cells were incubated with EV^Cx43− or EV^Cx43+ for 24 h before WB analysis of ATP2A2/SERCA2, a predicted target of miRNAs enriched in Cx43‐containing EV. Graph depicts quantification of ATP2A2/SERCA2 levels (n = 4 biological replicates).

Data information: P‐values were derived by Mann–Whitney test. Bars and error bars indicate mean ± SD in all graphs. Source data are available online for this figure.

Figure EV1. Cx43 modulates miRNAs content of extracellular vesicles

1. miRNAs expression profile analysis in HEK293^Cx43+ cells (represented as fold increase over HEK293^Cx43− cells).

2. EV produced by HEK293 cells expressing or not Cx43 (EV^Cx43+ or EV^Cx43−) were purified by differential ultracentrifugation. WB analysis of positive EV markers (Tsg101, Flotillin‐1 and GAPDH) and the negative marker Calnexin. WB for CD63 and CD81 was performed in non‐reducing conditions. Cellular extracts were used as control. 10 μg of total protein were loaded in each case.

3. Graph depicts total protein levels in purified EV produced by HEK293 cells expressing or not Cx43 (EV^Cx43+ or EV^Cx43−; n = 3 biological replicates).

4. Graph depicts total RNA levels in purified EV produced by HEK293 cells expressing or not Cx43 (EV^Cx43+ or EV^Cx43−; n = 4 biological replicates).

5. Representative NTA of concentration and size distribution of EV produced by HEK293 cells expressing or not Cx43 (EV^Cx43+ and EV^Cx43−; n = 2 biological replicates).

6. Representative TEM image of purified EV^Cx43+.

Data information: P‐values were derived by Mann–Whitney test. Bars and error bars indicate mean ± SD in all graphs. Source data are available online for this figure.

Figure EV2. Gene ontology (GO) enrichment analysis for the predicted targets of the over‐represented miRNAs in Cx43‐containing extracellular vesicles

1. GO Biological Process.

2. GO Molecular Function.

3. GO Cellular Component.

Figure EV3. Cx43‐containing extracellular vesicles modulate autophagy activity and ATP2A2/SERCA2 expression in recipient cells

1. HEK293 cells transfected with GFP‐LC3 were incubated with EV^Cx43− or EV^Cx43+ for 24 h. In the last 6 h, cells were serum‐starved before analysis of the number of GFP‐LC3 puncta. Scale bars 10 μm. Graph depicts the average number of GFP‐LC3 puncta per analyzed cell (n = 4 biological replicates).

2. Parental C33a cells were incubated with 5 µg of purified EV obtained from parental (EV^Cx43‐WT) or Cx43‐knockout cells (EV^Cx43‐KO) for 24 h, in the presence or absence of Bafilomycin A1 in the last 6 h. The levels of LC3 were analyzed by WB and depicted on graph (LC3‐II; n = 4 biological replicates).

3. Parental C33a cells were incubated with EV^Cx43‐WT or EV^Cx43‐KO for 24 h before WB analysis of ATP2A2/SERCA2, a predicted target of miRNAs enriched in Cx43‐containing EV. Quantification of the levels of ATP2A2/SERCA2 is depicted on graph (n = 4 biological replicates).

Data information: P‐values were derived by Mann–Whitney test. Bars and error bars indicate mean ± SD in all graphs. Source data are available online for this figure.

Figure 2. miRNAs enriched in Cx43‐containing extracellular vesicles are highly structured

Modelling of the 3D structure of mature miRNAs enriched in both EV^Cx43+ (red) and EV^Cx43− (green). Figure depicts each miRNAs structure, as well as the values of total energy [E(total)], van der Waals energy [E(VDW)] and electrostatic energy [E(elec)].

Figure 3. Cx43 interacts with hnRNPA2B1 and hnRNPQ

1. Cx43 was immunoprecipitated (IP) from HEK293^Cx43+ cells, after which interaction with hnRNPA2B1 and hnRNPQ was analyzed by WB.

2. Co‐localization between Cx43 (green) and hnRNPA2B1 or hnRNPQ (red) in C33a cells was analyzed by confocal microscopy. Scale bars 10 μm. Arrowheads in the inset images highlight co‐localization between Cx43 and hnRNPs.

Source data are available online for this figure.

Figure EV4. Cx43 interacts and co‐localizes with hnRNPA2B1 and hnRNPQ

1. IP of Cx43, hnRNPA2B1, or hnRNPQ was performed in C33a cells, as indicated. Interaction with hnRNPA2B1, hnRNPQ, and Cx43 was analyzed by WB.

2. hnRNPA2B1 or hnRNPQ were IP from HEK293^Cx43+ cells, where indicated, after which interaction with Cx43 was analyzed by WB.

3. Co‐localization between Cx43 (green) and hnRNPA2B1 or hnRNPQ (red) in HEK293^Cx43+ cells was analyzed by confocal microscopy. Scale bars 10 μm. Arrowheads in the inset images highlight co‐localization between Cx43 and hnRNPs.

Source data are available online for this figure.

Figure 4. The levels of miR‐133b are increased in Cx43‐containing extracellular vesicles

1. A–D

   EV‐producing cells expressing or not Cx43 (HEK293^Cx43+ or HEK293^Cx43−) were transiently transfected with shRNAs targeting hnRNPA2B1, hnRNPQ or non‐target shRNAs, for 48 h. miR‐133b levels were assessed in EV^Cx43+ and EV^Cx43− (A; n = 6–12 biological replicates) and EV‐producing cells (B; n = 4–9 biological replicates). miR‐509‐3p levels were assessed in EV^Cx43+ and EV^Cx43− (C; n = 6–15 biological replicates) and EV‐producing cells (D; n = 4–9 biological replicates).

2. E–H

   HEK293^Cx43+ or HEK293^Cx43− were transiently transfected with hnRNPA2B1‐GFP, GFP‐hnRNPQ or GFP alone, for 24 h. miR‐133b levels were assessed in EV^Cx43+ and EV^Cx43− (E; n = 4 biological replicates) and EV‐producing cells (F; n = 3 biological replicates). miR‐509‐3p levels were assessed in EV^Cx43+ and EV^Cx43− (G; n = 4 biological replicates) and EV‐producing cells (H; n = 3 biological replicates).

Data information: P‐values were derived by Mann–Whitney test. Bars and error bars indicate mean ± SD in all graphs.

Figure EV5. The levels of miR‐133b and miR‐199‐3p are increased in Cx43‐containing extracellular vesicles

1. A–H

   EV‐producing C33a cells expressing or not Cx43 (C33a‐WT or C33a‐KO) were transiently transfected with shRNAs targeting hnRNPA2B1, hnRNPQ or non‐target shRNAs, for 48 h. Levels of miR‐133b (A), miR‐199‐3p (B), miR‐410 (C) and miR‐509‐3p (D) were assessed in EV^Cx43‐WT and EV^Cx43‐KO (n = 4–5 biological replicates). Levels of miR‐133b (E), miR‐199‐3p (F), miR‐410 (G) and miR‐509‐3p (H) were assessed in C33a cells (n = 3 biological replicates).

Data information: P‐values were derived by Mann–Whitney test. Bars and error bars indicate mean ± SD in all graphs.

Figure 5. Cx43 participates in selective sorting of miR‐133b

1. A

   Cell free EV biogenesis was assessed by relative protection of Cx43‐luciferase. Reactions were incubated at 4 or 30°C in the presence of membranes derived from HEK293^Cx43− transiently transfected with Cx43‐luciferase and cytosol from HEK293^Cx43− cells, except in the negative control. 1% TX100 was used to disrupt membranes, where indicated (n = 3 biological replicates).

2. B

   Crude membranes and cytosol from broken HEK293^Cx43+ or HEK293^Cx43− cells were mixed in an ATP regenerating system, supplemented with synthetic miR‐133b or miR‐509‐3p, as indicated. Graph depicts the indicated miRNAs levels, assessed by RT‐qPCR and represented as percent protected after RNAse I digestion of unincorporated miRNAs (n = 7–8 biological replicates).

3. C

   Co‐immunoprecipitation of endogenous miR‐133b and miR‐509‐3p following UV crosslinking immunoprecipitation (CLIP) of Cx43 (n = 3–5 biological replicates).

4. D

   WB analysis for Cx43 in samples derived by miRNAs pulldowns performed with whole lysates of HEK293^Cx43+ cells and biotinylated miR‐133b and miR‐509‐3p (n = 5 biological replicates).

5. E–H

   EMSA was performed using either Cy3‐labeled miR‐133b (E,F) or miR‐509‐3p (G,H), and Cx43 produced by in vitro translation inserted into lipid nanodiscs, in the presence or absence of recombinant GST‐hnRNPA2B1 or GST‐hnRNPQ, as indicated. For competition EMSA, an excess of unlabeled miR‐133b or miR‐509‐3p was used (competitor). Empty nanodiscs were used to control unspecific binding.

6. I

   EMSA was performed using either Cy3‐labeled wild‐type or mutant miR‐199a‐3p, and Cx43‐enriched membranes prepared from C33a parental cells (Cx43⁺). Cx43⁻ membranes were used to control unspecific binding.

Data information: P‐values were derived by Mann–Whitney test. Bars and error bars indicate mean ± SD in all graphs. Source data are available online for this figure.

Figure 6. Interaction of Cx43 with miR‐133b in vitro is increased in the presence of hnRNPA2B1 and hnRNPQ

Representative maximum projection images of microscopy‐based miRNA‐protein interaction using beads coated with GST‐tagged NT/CT‐Cx43, hnRNPA2B1 or hnRNPQ, as indicated, following the addition of Cy3‐labelled miR‐133b or miR‐509‐3p. Graph depicts quantification of total fluorescence/bead, as a measure of protein‐miRNAs binding. Scale bars 50 μm (n = 2 biological replicates; average of 23 beads analyzed/condition). Data information: P‐values were derived by Kruskal–Wallis (Dunn’s post hoc) test. Bars and error bars indicate mean ± SD in all graphs. Source data are available online for this figure.

Figure 7. Cx43 facilitates the delivery of intraluminal RNAs to target cells

1. Equal amounts of EV^Cx43+ and EV^Cx43− labelled with SYTO® RNASelect™ green fluorescent stain were incubated with HEK293^Cx43+ cells for 30 min, at 37°C. Scale bars 10 μm (n = 5 biological replicates). P‐values were derived by Mann‐Whitney test.

2. Working model for the role of Cx43 in mediating the selective sorting of miRNAs into EV. Cx43 preferentially binds to miRNAs harboring a stable secondary structure, mediating its selective incorporation into EV. Although hnRNPs do not seem to directly participate in Cx43‐dependent loading of EV‐miRNAs, they strength the binding of miRNAs to cx43, suggesting that a multiprotein complex formed between RNA‐binding proteins and Cx43 is important to select miRNAs.