Functionally related proteins of AtCYP38 revealed by web-based tools
In order to search potential interacting proteins of AtCYP38, we first adopted the co-expression network program ATTED-II to search for proteins whose expression pattern is related to AtCYP38. The program revealed 20 proteins are tightly co-expressed with AtCPY38 (Fig. 1A and Table 1). Among these proteins, 8 proteins are localized in the thylakoid lumen, including FKBP18 (At1g20810), PPL (At1g76450), FKPB16-3 (At2g43560), PRP ( At2g44920), PPD6 (At3g56650), TAPX4 (At4g09010), TLP15 (At5g52970), and TLP17.4 (At5g53490); and 1 protein, HHL1 (At1g67700), resides on the thylakoid membrane. Considering the thylakoid lumen localization of AtCYP38, we assumed that these 9 proteins might be functionally related to AtCYP38. Meanwhile, the other 10 co-expression proteins, which are localized in the chloroplast stroma (Fig. 1A and Table 1), are not likely to physically interact with AtCYP38.
In order to evaluate the ATTED-II co-expression results, we also applied the functional protein association tool, STRING, to screen potential proteins related to AtCYP38. The results showed that there are 10 chloroplast proteins functionally related to AtCYP38 at the highest confidence value. Out of these proteins, 4 are localized in the chloroplast thylakoid lumen, including TLP18.3 (At1g54780), PRXQ (At3g26060), MPH2 (At4g02530), and FKBP13 (At5g45680), and 2 proteins, LQY1 (At1g75690) and ATPD (At4g09650), reside on the thylakoid membrane (Fig. 1B and Table 2). All the other proteins are not localized in the chloroplast thylakoid lumen. They might not physically interact with AtCYP38, though they could be functionally related.
After network analysis with these 2 different programs, we found that no protein was co-detected by both software. We assumed that the different results from these two web-based detections could be partially caused by their unique algorithms. Considering their localization and their functions in photosynthesis, we presumed that all these 15 lumen or membrane localized proteins might be functionally related to AtCYP38.
Target proteins screening against a library containing genome wide cDNA
Yeast two-hybrid (YTH) screening is a powerful and well-established tool to identify interested proteins which could be physically associated with a specific target protein. To identify potential targets and cofactors of AtCYP38, we launched a yeast two-hybrid screening with AtCYP38 as the bait protein. X-ray crystal structural analysis showed that the mature full-length AtCYP38 is composed of an N-terminal α-helical domain and a C-terminal cyclophilin β-sheet domain, and these two domains could possess different functions [23]. Therefore, we utilized three baits representing different forms of AtCYP38, including the mature full-length protein (CYP38F, without signal peptide), the N-terminal α-helical domain (CYP38N), and the C-terminal β-sheet domain (CYP38C) to perform yeast two-hybrid screening against an Arabidopsis genome wide cDNA library (Fig. 2A).
We screened approximately a total of 2.0×107 yeast transformants with these 3 bait proteins, respectively. When CYP38F is used as the bait, we obtained about 70 confirmed positive colonies. After data filtering and interaction confirmation, we found that 8 chloroplast proteins could be interacting partners of AtCYP38 (Fig. 2B, 3A and Table 3). Only 3 proteins were localized in the chloroplast thylakoid lumen, including PSBP-1 (At1g06880), TSP9 (At3g47070) and DEG8 (At5g39830). After screening with CYP38N as the bait, we obtained approximately 50 confirmed positive clones, and further narrowed down to 4 candidate proteins according to their sequence results. However, none of 4 candidate protein was localized in chloroplast (Fig. 2B, 3B and Table 4). Notably, there are a very limited number of positive colonies obtained from screening with CYP38F and CYP38N as baits, indicating that those two baits display a low activity in the YTH screening assay.
When screening the same library with the CYP38C domain, we then obtained about 200 confirmed positive colonies, which were much more than those obtained from screening with CYP38F or CYP38N. After data cleaning up and re-validation, we obtained 15 chloroplast proteins to be interacting candidates of AtCYP38 (Fig. 2B, 3C and Table 5). Interestingly, several chlorophyll binding proteins were identified in this screening, including LHCB1.1, LHCB1.2, LHCB1.5 and LHCB4.1. Other 3 thylakoid membrane proteins, the PSI subunit PASG, the Cytb6f complex subunit PETC, and ROD1[29], an enzyme involved in lipid metabolism, were also found to interact with AtCYP38. Three interacting proteins are thylakoid lumen localized proteins, including PSBP-1, PETE2, and AtCYP38. Notably, PSBP-1 is also a partner of CYP38F. In addition, we also found some proteins are not localized in the thylakoid lumen but still can be identified by yeast two-hybrid assay (Table 5). However, it is unlikely that these proteins physically interact with AtCYP38, considering the different compartmental locations of these proteins and AtCYP38.
It was suggested that the C-terminal domain plays its function to bind other proteins, and the N-terminal bundle domain acts as the regulatory domain, which suppressing AtCYP38 to bind its targets [23]. To test this hypothesis, we cross examined the identified proteins via cDNA library screening with different baits individually. We found that all target proteins screened by CYP38F and CYP38N could interact with CYP38C, but proteins identified from screening with CYP38C could not interact with CYP38F or CYP38N (Fig. 4), which is in agreement with the above assumption.
Altogether, after screening 2.0×107 colonies with three different forms of AtCYP38, we obtained a very limited number of potential interacting partners of AtCYP38. Only a few thylakoid lumen proteins were identified in the YTH screening, including PSBP-1, TSP9, DEG8, PETE2 and AtCYP38. We initially expected that a larger number of interacting proteins of AtCYP38 could be identified, given its the chaperone-like activity. The unexpected YTH screening results prompted us to think that screening against a genome wide cDNA library may not be an efficient way to identify the interested interacting proteins. Because proteins in the thylakoid lumen only occupy a very small fraction of the plant cell proteome, it is not surprising that cDNA for thylakoid lumen proteins are relatively rare in this library. Therefore, there exist too much noise background due to an overwhelming amount of unrelated proteins in this screening, and consequently, the chance would be very low for AtCYP38 to fish out target proteins which are located in the chloroplast thylakoid lumen.
Yeast two-hybrid screening against a thylakoid lumen protein mini library
Given that screening the genome wide cDNA library was not a very successful experience, we decided to launch an alternative screening strategy. Since AtCYP38 is a thylakoid lumenal protein, it would be a better choice to screen a library only harboring thylakoid lumen proteins and thylakoid membrane protein with lumenal domains, therefore massive unrelated proteins outside of this compartment would not interfere this screening process.
In order to achieve that, we constructed a yeast two-hybrid mini library harboring a majority of the previously identified lumenal proteins and lumenal domains of various thylakoid membrane proteins. The library was screened against baits including CYP38F, CYP38N and CYP38C, respectively. Interestingly, when we used CYP38F or CYP38N domain to perform the screening, we failed to find any positive potential interacting proteins.
When we used CYP38C to screen the YTH lumenal protein mini library, we obtained 15 potential interacting proteins, including 9 thylakoid lumen soluble proteins, and lumen fragments of 6 thylakoid membrane proteins (Fig. 5, Table 6). In detail, these soluble lumenal proteins are, At1g21500, At2g23670, At2g26340, CTPB (Carboxyl terminal peptidase, At3g57680), CTPA (At4g17740), TLP17.9 (At4g24930), DEG14 (At5g27660), CTPC (At5g46930), and TLP15 (At5g52970). Interestingly, 3 proteases responsible for process the C-terminal extension of D1 precursor (pD1), including CTPA, CTPB, and CTPC were all found to be interacting proteins of AtCYP38. And another protease, DEG14, was also found to interact with AtCYP38. 6 thylakoid membrane proteins whose lumen fragments could be bind to AtCYP38 included PSBM, CYTF, ALB3, CP43, CP47, and PSAF. We assumed that the membrane proteins and those proteases might be substrates of AtCYP38. The result from this screening is consistent with the in vivo function of AtCYP38 in the assembly of PSII supercomplex [16, 17].
Validating localization and in vitro pull-down confirmation assay.
AtCYP38 is known to be present in the chloroplast thylakoid lumen, and interacting proteins should be in this subcellular compartment as well. In order to verify the localization of candidate proteins, we selected three predicted lumenal proteins, At1g21500, TLP17.9 (At4g24930) and TLP15 (At5g52970), and fused the green fluorescent protein (GFP) tag to their C-terminus. The GFP-fused proteins were transiently expressed in leaf epidermal cells from Nicotiana benthamiana and analyzed by confocal microscopy. As we expected, At1g21500, TLP17.9 and TLP15 are localized in the chloroplast (Fig. 6A).
In order to further confirm the direct interaction between identified proteins with AtCYP38, we selected PSBP-1 and FKBP13 to perform the GST pull-down assays. As shown in Fig. 6B, GST-PSBP-1 and GST-FKBP13 were specifically pulled down by CYP38C-HIS but not GST, confirming the interaction between those two proteins and CYP38C in vitro. At the meantime, we also included two proteins with unknown function, At1g21500 and TLP17.9, to confirm their interaction with AtCYP38 by the same method. GST-21500 or GST-24930 also can be observed an obvious interaction with CYP38C in the pull-down assay (Fig. 6B). Overall, these results indicated that the interactions between AtCYP38 and our candidate proteins are valid, and our screening strategies are reliable and efficient.