Interactions of CAF1-NOT complex components from Trypanosoma brucei [version 1; peer review: 2 approved with reservations]

The CAF1-NOT complex of Trypanosoma brucei , like that of other eukaryotes, contains several NOT proteins (NOT1, NOT3, NOT3/5, NOT10, and NOT11), NOT9/CAF40, and the CAF1 deadenylase, which targets 3' poly(A) tails. Again like other eukaryotes, deadenylation is the first step in the degradation of most trypanosome mRNAs. In animal cells, destruction of unstable mRNAs is accelerated by proteins that bind the RNA in a sequence-specific fashion, and also recruit the CAF1-NOT complex. However, this has not yet been demonstrated for T. brucei . To find interaction partners for the trypanosome NOT complex, we did a genome-wide yeast two-hybrid screen, using a random shotgun protein fragment library, with the subunits CAF40, NOT2, NOT10 and NOT11 as baits. To assess interaction specificity, we compared the results with those from other trypanosome proteins, including the cyclin-F-box protein CFB1. The yeast 2-hybrid screen yielded four putatively interacting proteins for NOT2, eleven for NOT11, but only one for NOT9/CAF40. Both CFB1 and NOT10 had over a hundred potential interactions, indicating a lack of specificity. Nevertheless, a detected interaction between NOT10 and NOT11 is likely to be genuine. We also identified proteins that co-purify with affinity tagged NOT9/CAF40 by mass spectrometry. The co-purifying proteins did not include the 2-hybrid partner, but the results confirmed NOT9/CAF40 association with the CAF1-NOT complex, and suggested interactions with expression-repressing RNA-binding proteins (ZC3H8, ZC3H30, and ZC3H46) and the deadenylase PARN3.


Introduction
The CCR4/CAF1-NOT complex (NOT complex) represses translation and deadenylates polyadenylated mRNAs, thus priming them for degradation. It has been found in all eukaryotes examined so far. CCR4 and CAF1 are deadenylases; NOT1 is a large protein upon which the whole complex is assembled, and it binds directly to most other components. NOT3 or NOT5 (two proteins in yeast), NOT2 and CAF40 assemble on one side of the complex, while CAF1, CCR4, NOT10, and NOT11 are on the other. 1,2 . Another component, called Not4p in yeast, is less evolutionarily conserved; it participates in protein quality control by acting as an E3 ligase 3 . Artificial attachment of any of the complex components to a reporter mRNA represses expression, presumably via recruitment of the whole complex 4 . Analysis of unstable mRNAs in animal cells has revealed that in many cases, destabilizing elements in the 3'-untranslated region are recognized by specific RNA-binding proteins (RBPs), which in turn recruit components of the NOT complex via binding to different subunits 5-8 .

Results
Yeast two-hybrid screens Our trypanosome yeast two-hybrid prey library was made by random shotgun genomic cloning. The library has several million independent yeast clones, each of which expresses a different protein fragment (with roughly 1/12 being within an open reading frame and in-frame) 15 . NOT2, NOT10, NOT11 and CAF40 were used as baits to screen the library by mating. We also included another protein of interest, CFB1. CFB1 is expressed in bloodstream-form trypanosomes. Its function is somewhat enigmatic, though it is probably required for optimal cell growth 16 . All results are in Table S1. For the NOT complex components, at least one million diploid progeny were subjected to selection, resulting in between 100 and 800 surviving colonies, from which inserts were amplified and subjected to high-throughput sequencing. To find interacting proteins, we selected open reading frames for which there were at least two different in-frame sequences represented by at least 20 reads (Table S1, sheet 2). This procedure gives false positives and false negatives. Poor folding could result in a false positive due to aberrant exposure of hydrophobic surfaces, or in a false negative due to incorrect formation of a folding-dependent interaction domain. In addition, proteins for which the whole protein, or a (near-) complete N-terminus, are required for interaction would be excluded.
We obtained 6, 158, 15 and 3 interaction partners for NOT2, NOT10, NOT11 and CAF40 respectively; CFB1 interacted promiscuously with over 800 partners. (These did not include the only validated partner, MKT1, but we had previously found that MKT1 seems only to interact as a complete protein 15 .) To judge the likely specificity of the interactions, we compared them with those of MKT1 15 , RBP10 (Mugo and Clayton; unpublished study, manuscript submitted), 4EIP and Tb927.7.2780 (unpublished studies; Terrao, ZMBH). The result for NOT10 strongly suggested a tendency for non-specific interactions, although not as severe as for CFB1. The most specific interactions for NOT2, NOT10 and NOT11 are shown in Table 1. It was notable that no interactions with either CAF1 or NOT1 were detected; this might be because the fragments encoded in our prey library are too small. NOT11 had two unique interactions, with a DNAj domain protein (Tb927.9.1560) and mitochondrial EF-Tu; the latter is unlikely to be physiological because of the protein location. There were also two interactors that were shared only with the promiscuous CFB1 -a protein phosphatase and a protein of unknown function, both of which are probably in the cytoplasm. NOT11 interacted with itself, and shared 7 other interactions with NOT10. Some of these, such as the interaction between NOT10 and NOT11, are likely to be genuine -NOT10 interacts with NOT11 in yeasts 4,17 . Apart from NOT11, none of these proteins interacts with bloodstream-form trypanosome mRNA 14 or has any effect in the tethering assay 14,18 .
To supplement results from the random fragment library, we used recombinant CAF40 to screen a cDNA protein expression array and a similar array displaying proteins from full-length open reading frames of proteins involved in mRNA metabolism 14,19 . Two other proteins, Aurora kinase B and T. brucei polo-like kinase, were included as bait controls to exclude "sticky" proteins. Using the open reading frame array, no specific interaction partners were found. For the cDNA array, selected clones were sequenced, then checked in a pairwise yeast-two hybrid assay. The only confirmed interaction was with the nascent polypeptide associated complex alpha subunit-like protein (Tb927.9.8100/8130). An interaction between the yeast NOT complex and the nascent polypeptide associated complex was previously reported in yeast; the beta subunit Egd1p, had a weak two-hybrid interaction with Caf40p, but much stronger interactions were detected between Egd1p and some other NOT complex components 20 . The significance of the possible interaction is unclear since the nascent polypeptide associated complex was not found in the affinity Table 1. Most specific interaction partners of NOT2, NOT11 and CAF40. For details see Table S1. Only proteins that interacted in less than four screens are shown; none were positive with MKT1 or 4EIP. The columns for NOT2, NOT10, NOT11 and CAF40 show the number of different interacting protein fragments. The 4EIP, 7.2780 and CFB1 screens were done at different times, with different sequencing depth, so that the numbers are not strictly comparable. These are therefore designated "y" for "yes" and "0" for less than 2 interacting fragments. "Loc" indicates the subcellular location, when known, either from the TrypTag project (http://tryptag.org) or from other information. "cyt" = cytosol, "mit" = mitochondrion", "nuc" = nucleolus. POMP37 was a mitochondrial membrane protein by proteomics but the GFP-tagged protein in TrypTag has an ER-type pattern.

Tb927
Function Proteins that co-purify with V5-tagged CAF40 To find proteins that co-purify with CAF40, we integrated a sequence encoding a V5 tag in frame with the 5' end of the open reading frame in the genome of bloodstream form trypanosomes 21 . The resulting parasites are expected to express N-terminally V5tagged CAF40 at an approximately normal level. Eluates from two independent pull-downs -one with RNase inhibitor, and the other with RNase -were analysed by mass spectrometry, in parallel with a single control from wild-type cells (no tag) (Table S2, sheet 1). As additional controls we used the results from three other tandem affinity purifications of tagged GFP. To detect proteins that are frequent contaminants after affinity purification we compared the results with those from many other experiments from our and other labs. Since our CAF40 experiment included only two replicates, all of the results need further confirmation before definitive conclusions can be drawn. Nevertheless, V5-tagged CAF40 showed specific, reproducible RNA-independent pull down of the NOT1, NOT10, NOT11 and CAF1 subunits of the CAF1-NOT complex (Table S2, sheet 2), while co-purification of NOT2 and NOT5 was diminished by RNAse treatment (Table S2, sheet 3).
The results also suggested RNA-independent association with three potential RNA-binding proteins, ZC3H8, ZC3H30 and ZC3H46. Interestingly, all three of these RNA-binding proteins were found associated with mRNA 14 and each one repressed expression in tethering screens 14,18 . We therefore speculate that they may repress expression by recruiting the NOT complex via CAF40. Some of the proteins that were pulled down in the absence of RNase also appeared interesting; they included 7 known or potential RNAbinding proteins (CSBII, DRBD4 (PTB2), PUF6, ZC3H28 and ZC3H41, HNRNPH and Tb927.11.14090), and the 5'-3' exoribonuclease XRNA (at rather low coverage). This was, however, only a single experiment, so the results will not be discussed further.

Conclusions
The yeast 2-hybrid data by themselves provided no clues concerning functional partners of the NOT complex components investigated, beyond confirming the likely interaction between NOT11 and NOT10. Nevertheless, they may be useful in conjunction with other results; for example, the protein phosphatase Tb927.10.4930 might be implicated in regulation. Our mass spectrometry of CAF40, in contrast, did point to some proteins that might be involved in recruiting the NOT complex to mRNAs; this is especially likely for those proteins that also repressed expression in the tethering assay. We therefore hope that the results will be useful to others who may wish to investigate the modes of action of these proteins in detail.

Yeast two-hybrid analysis
The Matchmaker GAL4 Two-Hybrid System3 (Clontech) was used, with Gateway cloning of amplified open reading frames to create bait plasmids. The trypanosome prey library, sequencing methods and bioinformatic analysis have been previously described 15 . Briefly, bait plasmids were transformed into AH109 yeast, and the pool of prey plasmids were transformed into the Y187 strains. The cells were allowed to mate, plated on SD agar plates lacking tryptophan, leucine, or both (double drop-out medium) to calculate mating efficiency, then plated without tryptophan, leucine, adenine and histidine (quadruple drop-out medium), and incubated for 3 to 5 days at 30°C. The resulting colonies were re-plated on quadruple drop-out SD plates with 40 μg/ml X-α-Gal and 3-amino triazole (3-AT, 0.5 to 2mM), and incubated for 3 to 5 days. Blue colonies were then pooled and grown overnight in quadruple drop-out medium with 3-amino triazole for plasmid isolation. The plasmid inserts were amplified with barcoded primers, pooled and sent for library preparation by David Ibbersson (BioQuant, Heidelberg, Germany), then sequenced (Illumina Hi-Seq, EMBL, Heidelberg., Germany). The numbers of reads obtained are in Table S1, Sheet 1.
Reads were aligned to the T. brucei 927 genome, as described 15 . We selected sequences that were present at least 10 times, and that had annotated open reading frames that were in frame with the DNA-binding domain (Table S1, sheets 3-5). We then chose a list of unique open reading frames with only one copy each of repeated genes 22 . We also restricted the selection to open reading frames that included no more than 13 codons upstream of the start codon, and included at least 6 codons before the stop codon. Individual results are in Table S1, sheets 4-7 and final results are summarized in Table S1, sheet 2. Sheet 3 includes locations found in screens with more than three different baits.
Affinity purification of V5-CAF40 and mass spectrometry About 10 10 bloodstream form cultured bloodstream-form T brucei carrying endogenously V5-tagged CAF40 were subjected to immunoprecipitation, as previously described 13 . The eluate was run 1 cm in 10% SDS PAGE gel, and was cut in 6 pieces for Mass-Spectrometry analysis. The piece were transferred to a 96-well plate and reduced, alkylated and digested with trypsin, as described 23 , except that triethylammoniumbicarbonate buffer was used instead of ammoniumbicarbonate buffer. Following digestion, tryptic peptides were extracted from the gel pieces with 50% acetonitrile/0.1% TFA, and concentrated nearly to dryness in a speedVac vacuum centrifuge. Peptides were separated on an analytical column (75um × 150mm) with a flow rate of 300nl/min (nanoAcquity, Waters) using 1% formic acid and an acetonitrile gradient increasing from 3% acetonitrile to 37% acetonitrile over 50 minutes. The UPLC system was on-line coupled to an ESI LTQ Orbitrap XL MS (Thermo Fisher). One survey scan (res: 60000) was followed by 5 information dependent product ion scans in the LTQ. Only doubly and triply charged ions were selected for fragmentation.
Data were analyzed using Proteome Discoverer 1.4 and Mascot (Matrix Science; version 2.4). Mascot was set up to search the the TriTrypDB_9.0_Tbrucei_2015Jan database (20400 entries) using trypsin as protease, a fragment ion mass tolerance of 0.50 Da and a parent mass tolerance of 100 ppm. Iodoacetamide derivative of cysteine was specified in Mascot as a fixed modification. Deamidation of asparagine and glutamic acid, as well as oxidation of methionine, were specified in Mascot as variable modifications. Target decoy PSM validator was set to a target FDR (strict) of 0.01. Results are in Table S2.

Competing interests
No competing interests were disclosed.

Grant information
This work was funded by core funding to CC, and by the Deutsche Forschungsgemeinschaft (grant Cl112/24).
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Open Peer Review
In this study the authors try to identify proteins interacting with specific components of the CAF1-NOT complex in Trypanosome brucei. They used the 2 hybrid method for this and analyzed CBF1 as a control bait to determine specificity. Since they did not identify any significant partner for one of their baits, CAF40, they used alternative approaches, one of which, the affinity purification of CAF40 from Trypanosome brucei followed by mass spectrometry, gave some results.

General comments
This work really does not go beyond the description of the screening and identified potential partners with a very minimal description of these potential partners.
The filtering used to identify which clones are more likely to be true interaction partners has its limitations. The authors use a protein of interest CBF1, find 800 partners and conclude that Cbf1 is promiscuous. We have no information about this protein, or why the identification of 800 partners means that it is promiscuous. It is also not clear why comparing partners of the CAF1-NOT proteins to those of CBF1 is useful, and the authors end up using data that is not in this manuscript, namely partners of others proteins that were screened, to determine which proteins might be specific partners of the CAF1-NOT proteins. It seems that in this case, it would just be more appropriate to provide all the clones isolated for the specific baits and then use a letter in the table to indicate that clones were isolated also in many other unrelated screens. The data exists in a supplementary table, but it could be put in a reader-friendly manner like Table 1 in the main text.
It is curious that no CAF1-NOT components were isolated. I am not sure that the size of the library fragments is a good explanation. This could be tested directly, by using the structural information available and make a very small NOT1 clone. But in any event, it seems that the more likely explanation is that the screening was not saturated.
Previous data has indicated that chaperones co-purify with NOT10-11, hence the finding of many partners may not be indicative of non-specific interactions but may have significance.
In conclusion, I do think that the results of a screen like this would be useful to the community. It could be that if we wait for the authors to try to follow up on all the clones and get relevant biological data, this may take a lot of time and hence some of the data will end up not being made visible. However I feel that the effort the authors make to validate some clones over others is not a good idea. In any event, if they chose to do so, then there should be more description about these clones.

Minor comment
The way by which Not4 acts in protein quality control is somewhat controversial, I would not write that it is by acting as an E3 ligase.

If applicable, is the statistical analysis and its interpretation appropriate? Partly
Are all the source data underlying the results available to ensure full reproducibility? Yes Are the conclusions drawn adequately supported by the results? Yes would need to be repeated twice more, then individual partners followed up by other means. We had obviously hoped that some CAF40 interaction partners would emerge from both screens, but that did not happen.
We decided to publish as is because my lab does not currently have the resources available for further follow-up and it seemed silly to just leave the data on a private hard disk. We thought that at least, it could help other researchers who are doing pull-downs or twohybrid. First, it will prevent other people from wasting time doing the 2-hybrid screens again. Secondly, it might help other people to decide which results to follow up in their own interaction experiments. For example, if a lab is working on a protein X, and discovers that it pulls down a NOT complex subunit, they could look at our results and see if their protein X was in one of our datasets. If it was, that would encourage them to investigate further.
different. I understand that these were the most specific interaction partners, but I was confused a mitochondrial protein was among those. Overall, I felt that the presentation of the data in the tables was a little 'raw'. In their conclusion, the authors state that 'The Y2H data by themselves provided no clues confirming functional partners of the Not complex'.
The affinity purification of Caf40 ±RNAse identified several interesting proteins, although the one CAF40 interaction partner identified in the Y2H was not among these. However, these results were weakened by the authors themselves as the paragraph ends with 'This was, however, only a single experiment, so the results will not be discussed further'. As with the Y2H data, I also felt that the data presentation was rather raw.
I'm sorry not to be more positive with this manuscript. I do realise that the Y2H was a lot of work, but it is nonetheless important to increase confidence in the results using the controls/replicates suggested above. An additional approach such as BioID would also be a good control, although I realise that this is beyond the scope of this manuscript.

Are sufficient details of methods and analysis provided to allow replication by others? Yes
If applicable, is the statistical analysis and its interpretation appropriate?

Not applicable
Are all the source data underlying the results available to ensure full reproducibility? Yes

Are the conclusions drawn adequately supported by the results? Yes
Competing Interests: No competing interests were disclosed.
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias • You can publish traditional articles, null/negative results, case reports, data notes and more • The peer review process is transparent and collaborative • Your article is indexed in PubMed after passing peer review • Dedicated customer support at every stage • For pre-submission enquiries, contact research@f1000.com