Non-symbiotic hemoglobins (nsHbs) are O ₂-binding proteins widely distributed in land plants, including rice ¹ . The nsHbs are classified into type 1 and type 2 (nsHbs-1 and nsHbs-2, respectively) based on sequence similarity and O ₂-affinity ^{2, 3} . The O ₂-affinity of nsHbs-1 is very high mostly because of an extremely low O ₂-dissociation ( k _off) rate constant ^{3– 5} resulting in that nsHbs-1 apparently do not release O ₂ after oxygenation ^{6, 7} . In contrast, the O ₂-affinity of nsHbs-2 is moderate mostly because of a moderate to high k _off rate constant for O ₂, thus apparently nsHbs-2 easily release O ₂ after oxygenation ^{2, 3, 6, 7} . Hence, it is possible that the in vivo function of nsHbs-1 is other than O ₂-transport and that nsHbs-2 function in vivo as O ₂-carriers.

Five copies ( hb1 to 5) of the nshb gene have been detected in the rice genome, which are differentially expressed in embryonic and vegetative organs from plants growing under normal and stress conditions ^{8– 11} . Based on the available information on the properties of rice nsHbs and data from the analysis of other plant and non-plant Hbs, it was proposed that rice nsHbs could exhibit a variety of functions in vivo, including O ₂-transport, O ₂-sensing, NO-scavenging and redox-signaling ^{6, 12, 13} . However, the in vivo activity of rice nsHbs has been poorly analyzed ¹² . An in vivo analysis for rice nsHbs is essential to elucidate the biological function(s) of these proteins. An approach to analyze the in vivo activity of nsHbs is generating knock out rice for individual nshb genes, however this is complicated because of the existence of five copies of nshb in the rice genome. An alternative approach to analyze the in vivo activity of rice nsHbs is examining individual rice nsHbs in a heterologous system, such as recombinant Escherichia coli. Rice Hb1 ⁴ and Hb2 ¹⁴ are nsHbs-1 that have been generated in recombinant E. coli TB1. The rice Hb1 and Hb2 amino acid sequence ⁴ , tertiary structure ¹⁵ and rate and equilibrium constants for the reaction of O ₂ ^{4, 14} are highly similar. Thus, it is possible that rice Hb1 and Hb2 function similarly in vivo. As an initial approach to test this hypothesis we analyzed the effect of the synthesis of rice Hb1 and Hb2 in the recombinant E. coli TB1 growth. Our results showed that synthesis of rice Hb1 and Hb2 inhibited the recombinant E. coli TB1 growth and that growth inhibition was stronger when recombinant E. coli TB1 synthesized rice Hb2 than when synthesized rice Hb1.

Untransformed (wild-type) and transformed (recombinant) E. coli TB1 (Invitrogen, CA, USA) containing the constitutive pEMBL18 ⁺::Hb1 ⁴ , pEMBL18 ⁺::Hb2 ¹⁴ , pEMBL18 ⁺::Lb a ¹⁶ , pEMBL18 ⁺::LbII ¹⁷ and pUC18::VHb ¹⁸ plasmids were grown in LB broth (Sigma-Aldrich, MO, USA) at 37°C with shaking at 200 rpm. Plasmids pEMBL18 ⁺::Lb a, pEMBL18 ⁺::LbII and pUC18::VHb were included as an O ₂-carrier control since they code for the synthesis of the O ₂-carrying soybean leghemoglobin a (Lb a), cowpea leghemoglobin II (LbII) ^{17, 19, 20} and Vitreoscilla Hb (VHb) ^{21, 22} , respectively. The existence of the VHb insert into the pUC18::VHb plasmid was verified by PCR (30 cycles at 55°C/30s for annealing, 72°C/30s for extension and 95°C/30s for denaturation) using specific oligonucleotides (VitHb/ATG: 5´-ATG TTA GAC CAG CAA ACC ATT-3´ and VitHb/TAA: 5´-TTA TTC AAC CGC TTG AGC GTA-3´) designed from the vhb sequence deposited in the Genbank database under the accession number X13516. The existence of the Hb1, Hb2, Lb a and LbII inserts into the pEMBL18 ⁺::Hb1, pEMBL18 ⁺::Hb2, pEMBL18 ⁺::Lb a and pEMBL18 ⁺::LbII plasmids, respectively, was verified by EcoRI- and NcoI (Invitrogen, CA, USA) -double digestion. Inserts were detected by electrophoresis in a 1.4% agarose gel. The existence of recombinant Hbs in cell soluble extracts was verified by SDS-PAGE in a 12.5% polyacrylamide gel. Evaluation of the effect of the Hb synthesis in the recombinant E. coli TB1 growth was performed in 50 ml cultures inoculated with ≈5 × 10 ⁸ colony forming units from a 20 ml overnight culture. Wild-type E. coli TB1 was included as control. All assays were performed in triplicate. Cell growth was quantitated by spectrophotometry using λ = 650 nm for an 8.5 h period.

Electrophoretic analysis of the PCR reaction and EcoRI- and NcoI-double digestions showed that plasmids isolated from recombinant E. coli TB1 contained inserts corresponding to the rice Hb1 ⁴ , rice Hb2 ⁴ , soybean Lb a ¹⁶ , cowpea LbII ¹⁷ and Vitreoscilla Hb ¹⁸ cDNAs ( Figure 1A). Likewise, analysis by SDS-PAGE showed that rice Hb1, rice Hb2, soybean Lb a, cowpea LbII and Vitreoscilla Hb existed in the soluble extracts of recombinant E. coli TB1 ( Figure 1B). This evidence indicated that rice Hb1, rice Hb2, soybean Lb a, cowpea LbII and Vitreoscilla Hb were synthesized by recombinant E. coli TB1.

Figure 2 shows that synthesis of rice Hb1, rice Hb2, soybean Lb a, cowpea LbII and Vitreoscilla Hb inhibited the recombinant E. coli TB1 growth. This was unexpected for soybean Lb a, cowpea LbII and Vitreoscilla Hb because these proteins would promote cell growth due to their O ₂-transport activity ^{17, 19– 22} . However, under the conditions tested in this work apparently soybean Lb a, cowpea LbII and Vitreoscilla Hb affected some aspects of the recombinant E. coli TB1 metabolism, possibly owed to the constitutive expression of these proteins into the host cells. Synthesis of rice Hb1 inhibited the recombinant E. coli TB1 growth similarly (∼37%) to the synthesis of soybean Lb a, cowpea LbII and Vitreoscilla Hb. This observation suggests that rice Hb1 could function in vivo similarly to O ₂-carrying Hbs. Likewise, synthesis of rice Hb2 also inhibited the recombinant E. coli TB1 growth. However, growth inhibition was stronger (∼61%) when recombinant E. coli TB1 synthesized rice Hb2 than when synthesized rice Hb1. This observation suggests that rice Hb2 could function in vivo by scavenging O ₂, possibly owing to its extremely low k _off rate constant for O ₂ ¹⁴ .

Results presented in this work suggest that in spite of the high similarity between rice Hb1 and Hb2 these proteins could function differently in vivo. In order to elucidate the apparent metabolic effects generated by the synthesis of rice Hb1 and Hb2, future work might focus on the physiological and biochemical characterization of recombinant E. coli TB1. This may include measuring cell respiratory rates and identifying cell proteins and metabolites using oximetry and proteomic and metabolomic approaches, respectively. Results from these analyses could provide valuable information to understand the in vivo function of rice nsHbs.