For each salinity level I think you could test all the proline concentrations. So you have 5 salinity levels and 6 proline concentrations = 5x6 = 30 treatments. You can add one treatment without salt (0 dS/m) and test all the proline concentrations as control. So you will have 36 treatments.
The best is to use 4 replicates for each treatment.
The best experimental design for pot culture is the completely randomized block design.
Thanks for sharing your question with us. If your treatments (proline conc. and salinity) have the same importance to you, and you want to study the effect of both Proline and salinity at the same time, you need to use factorial design in Completely Randomized Design (if you are doing the experiments under controlled condition like greenhouse). In factorial design with 3 replications, you will have 6 Prol x 5 Sal x 3 rep=90 treatments.
I hope these information will help you. Please do not hesitate to ask me if you need any further questions.
If your experiment conducted under control environments you can use Factorial experiment in CRD ,your experimental units will be 30(6*5)in every replication .If it conducted under field condition then you can use either factorial experiment in RCBD when you consider the two factors (salinity,and Proline)have the same importance degree ,and you may use split plot in RCBD if any of the two factors more importance than the other ,here you put one of them in the main plot ,and the other vin sub plots.