1.        Grow Bradyrhizobium strain in HM medium (Nieuwkoop et al., 1987) containing only arabinose as carbon source to an OD600nm= 0.4 to 0.8. Grow E. coli strain(s) in LB to approximately the same OD.
2.        Pellet 1 ml of both strains in a sterile microcentrifuge tube at 10,000 rpm for 3 min. If either stain was grown with antibiotics, then wash twice with 1 ml of HM medium and pellet again. For triparental mating, do the same to the strain with the helper plasmid.
3.        Decant the supernatant and resuspend the cell pellet in 50-100 µl of HM broth by pipetting (no vortexing).
4.        Mix the strain suspensions together and transfer the mixture onto a sterile 0. 45µM filter and place the filter (cell side up) onto a HM agar plate. Let the dry to remove any excess moisture and then incubate at 30oC for two days. Using a day-old HM plate helps to reduce the moisture. Set up controls with filters containing both donor and recipient alone.
5.        After incubation, lift the filter from the plate and place into a 1.5 ml microcentrifuge tube. Add 0.5-1.0 ml HM medium and vortex to suspend cells.
6.        Spread 100-200 µl of the cell suspension onto RDY agar plates (So et al., 1987) containing the appropriate antibiotics. The same is done for controls. Incubate plates at 30oC. Colonies usually appear in approximately 6-10 days.

References:

Nieuwkoop, A.J., Z. Banfalvi, N. Deshman, D. Gerhold, M.G. Schell, K. Sirotkin, and G. Stacey. 1987. A locus encoding host range is linked to the common nodulation genes of Bradyrhizobium japonicum. J. Bacteriol. 169: 2631-2638.

So, J.-S., A.L.M. Hodgson, R. Haugland, M. Leavitt, Z. Banfalvi, A.J. Nieuwkoop, and G. Stacey. 1987. Mol. Gen. Genet. 207: 15-23.