Digital　Twin　(DT)　has　emerged　as　an　enabling　technology　for　sixth　generation　(6G)　vehicle-to-everything　(V2X)　communications.　However,　there　are　two　crucial　issues　on　leveraging　DT　for　6G　V2X　communications.　First,　what　kind　of　DT　capabilities　can　be　combined　with　the　6G　V2X　networks?　Second,　how　can　we　transform　the　DT　capabilities　into　practical　V2X　network　performance　gain?　Motivated　to　solve　these　problems,　this　article　investigates　the　DT　capabilities　under　a　DT　and　mobile　edge　computing　empowered　6G　V2X　network　architecture.　Specifically,　three　DT　capabilities　are　presented:　First,　strengthening　the　human-machine　interaction,　via　driving　behavior　analysis;　second,　improving　traffic　safety　via　knowledge-based　vehicle　fault　diagnosis;　and　third,　analyzing　spatial-temporal　traffic　characteristics,　via　data　aggregation.　Furthermore,　we　investigate　two　case　studies　for　illustrating　how　to　utilize　DT　capabilities　to　perform　task-efficiency　oriented　V2X　network　scheduling.　In　the　first　case　study,　the　driver　behavior　analysis　result　is　combined　with　the　V2X　channel　scheduling　strategy.　In　the　second　case　study,　a　deep　reinforcement　learning-based　vehicle　merging　decision　is　devised　in　the　DT　domain.　Then,　a　coalition-based　V2X　channel　scheduling　strategy　is　proposed,　to　help　accomplish　the　vehicle　merging　decision　task.　Finally,　we　evaluate　the　performance　of　our　proposed　task-efficiency　oriented　V2X　channel　scheduling　schemes,　and　highlight　the　future　research　directions.