Spot 42 RNA

Last updated
Spot 42 RNA
RF00021.jpg
Identifiers
SymbolSpot_42
Alt. SymbolsSpot42
Rfam RF00021
Other data
RNA type Gene; sRNA
Domain(s) Bacteria
SO SO:0000389
PDB structures PDBe

Spot 42 (spf) RNA is a regulatory non-coding bacterial small RNA encoded by the spf (spot forty-two) gene. [1] Spf is found in gammaproteobacteria and the majority of experimental work on Spot42 has been performed in Escherichia coli [2] [3] and recently in Aliivibrio salmonicida. [4] In the cell Spot42 plays essential roles as a regulator in carbohydrate metabolism and uptake, and its expression is activated by glucose, and inhibited by the cAMP-CRP complex. [5] [6] [7] [8] [9]

Contents

The sRNA is transcribed from a separate promoter and binds to messenger RNA targets through imperfect base pairing. The half-life of Spot42 in vivo is 12 to 13 minutes at 37 °C. [5] When grown in media supplemented with glucose, each cell contains 100–200 Spot42 copies.[ citation needed ] The corresponding level is however reduced 3–4-fold when cells are grown in succinate or when cAMP is added to cells grown in glucose. [8]

Discovery

Spot42 was first described in 1973 as an unstable RNA species of 109 nucleotides in Escherichia coli . It was discovered by polyacrylamide gel electrophoresis and 2-D fingerprinting in an attempt to study the accumulation of small RNAs in E. coli during amino acid starvation. [2] [3] In these experiments the electrophoretic mobility of Spot42 was similar to that of 5S rRNA. In 1979 Spot42 was found to accumulate under growth in the presence of glucose (i.e., when adenosine 3′,5′-cyclic monophosphate (cAMP) is low). During growth with a non-glucose carbon source (i.e., when cAMP concentrations are high) the Spot42 concentrations were found to be significantly lower. [7]

Later experiments showed that over-expression of Spot42 (~10 fold increase) resulted in impaired growth and lowered ability to adapt to shifts to richer media. [10] Further, shift from glucose to succinate as the carbon source resulted in a long lag period and slow growth rate. It was also stated that the reason for the abnormal responses was caused by an elevated number of excessive Spot42 RNA gene products rather than excess of the gene itself. A deletion study of spf in E. coli cells resulted in viable spf null mutants, which indicates that Spot42 is non-essential, at least under controlled lab conditions. [11]

Genomic localization and natural distribution

The natural distribution of the spf gene is restricted to 5 orders of gammaproteobacteria; Enterobacteriales, Aeromonadales, Vibrionales, Alteromonadales, Chromatiales. [12]

Enterobacteriaceae

The spf gene is highly conserved in Escherichia , Shigella , Klebsiella , Salmonella , Yersinia genera within the family Enterobacteriaceae. [6] In E. coli the spf gene is flanked by polA (upstream) and yihA (downstream). [13] [14] A CRP binding sequence and -10 and -35 promoter sequences are found upstream of spf.

Vibrionaceae

Spf is also highly conserved within the family Vibrionaceae, and was recently identified in all 76 available Vibrionaceae genomes (e.g., Vibrio , Aliivibrio, Photobacterium and Grimontiagenera). [4] In e.g., Vibrio cholerae , Vibrio vulnificus, Aliivibrio fischeri and Aliivibrio salmonicida the spf gene is flanked by polA (upstream) and a sRNA gene encoding the novel VSsRNA24 (downstream).

Biological function and specific targets

It was for some years unclear if the function of Spot 42 was mediated through the 109 nucleotide RNA itself or if the function was mediated through the 14 amino acids long peptide which is hypothetically encoded from within the sRNA sequence. This was based on the observation that Spot42 contains structural features similar to other non-coding RNAs found in E. coli (such as 6S RNA and lambda bacteriophage), as well as features that are typically found in mRNAs (i.e., polypurine sequence followed by AUG, 14 amino acids and an UGA terminator). [1] Using a filter binding assay and other methods showed that Spot 42 is not an mRNA. In this approach the affinity between Spot42 and the 70S ribosome was tested. [15] Here, Spot 42 showed very inefficient binding to purified 70S ribosomes, which lead to the conclusion that the function of Spot 42 is mediated by the RNA itself. Bækkedal and Haugen made a Spot42 consensus secondary structure based on all known "spf" sequences at the time (2015) and found that the spot42 gene is highly conserved across the 5 orders it is identified. [12] The secondary structure has highly conserved nucleotide positions that have the potential to participate in binding with known mRNA targets.

Biological function of Spot42 in Escherichia coli

In E. coli Spot 42 accumulates under growth in the presence of glucose (i.e., when adenosine 3′,5′-cyclic monophosphate (cAMP) is low). [7] The direct responsiveness of Spot 42 levels to glucose and cAMP is due to repression of spf expression by a cAMP-CRP (cAMP-receptor protein) complex. [8] Spot42 is found in 100–200 copies per cell when cells are grown in glucose, and is reduced 3–4 folds when cells are grown in succinate (a secondary carbon sources). The reduction of Spot42 in cells grown in secondary carbon sources is a result of binding of the cAMP-CRP complex to the spf promoter, which negatively regulates transcription of Spot42. Later, the proximity of spf to polA (gene encoding DNA polymerase I) led Polayes and co-workers to test whether the products of these genes could influence each other. [13] They found that by reducing levels of Spot 42, either by deletion of spf or by manipulating the growth conditions, the DNA pol A activity was reduced. The underlying mechanism for this observation remains however unknown.

Spot42 targets in Escherichia coli

Spot42 can interact directly with mRNA targets through base pairing. The first Spot 42 target was discovered by Møller et al. who showed that Spot 42 specifically binds to a short complementary region at the translation initiation region of galK (encodes a galactoinase). [6] galK is the third gene in the galactose operon, which contains four genes (galETKM) and produces a polycistronic mRNA. Spot 42 mediates discoordinate expression of the gal operon (i.e., the individual genes in the operon are not similarly expressed) by binding to the galK Shine-Dalgarno region, thereby blocking ribosome binding and translation of the galK gene. The physiological significance of the coordinate expression is unclear, but suggests that Spot 42 plays a role in fine-tuning gene expression to optimize the utilization of carbon sources.

Beisel and Storz demonstrated with microarray analysis and reporter fusions that Spot 42 plays a broader role in metabolism by regulating at least 14 operons. [5] These operons contain a number of genes involved in uptake and catabolism of non-favoured carbon sources. During overexpression of Spot 42 sixteen different genes show 2-fold elevated levels of mRNA. The identified genes are mostly involved in central and secondary metabolism, as well as uptake and catabolism of non-preferred carbon sources and oxidation of NADH.

A comparative genomics approach was able to broaden the link of Spot 42 with the Escherichia coli TCA cycle. [16] Next to the previously reported, TCA affiliated target gltA [5] both icd and sucC were computationally predicted and subsequently experimentally verified as direct targets of Spot 42. In addition this approach detected gdhA as direct target of Spot42. gdhA codes for the glutamate dehydrogenase and links citrate cycle and nitrogen metabolism.

Biological function and targets of Spot42 RNA in Salmonella

A study combining a transcriptome wide binding map of the Hfq protein with a comparative target prediction [17] has aided in identifying the mglB (STM2190) mRNA as a direct target of Spot42. [18]

Biological function and targets of Spot42 RNA in A. salmonicida

The observation that A. salmonicida contains the spf gene (which encodes Spot 42), but lacks the galK operon (the natural Spot 42 target in E. coli), have inspired scientists to study the role of Spot 42 in this fish pathogen. [4] A. salmonicida is unable to utilize galactose (lacks gal operon) in minimal medium and addition of galactose has little effect on the growth rate. When cells are grown in glucose the level of Spot42 is increased 16–40 folds, but is in contrast decreased 3 folds when cAMP is added, indicating that Spot42 probably have similar roles as in E. coli (i.e., in carbohydrate metabolism). It has been hypothesized that Spot 42 works in concert with a novel sRNA gene, called VSsrna24, located 262 nt downstream of spf. The VSsrna42 RNA is approximately 60 nt in length and has an expression pattern opposite to that of Spot42. Furthermore, in a spf deletion mutant a gene encoding a pirin-like protein was upregulated 16 folds. Pirin has key roles in the central metabolism by regulating the activity of pyruvate dehydrogenase E1 and therefore select if pyruvate will be fermented or go through respiration through the TCA cycle and electron transport.

Related Research Articles

<span class="mw-page-title-main">Cyclic adenosine monophosphate</span> Cellular second messenger

Cyclic adenosine monophosphate is a second messenger important in many biological processes. cAMP is a derivative of adenosine triphosphate (ATP) and used for intracellular signal transduction in many different organisms, conveying the cAMP-dependent pathway.

In genetics, an operon is a functioning unit of DNA containing a cluster of genes under the control of a single promoter. The genes are transcribed together into an mRNA strand and either translated together in the cytoplasm, or undergo splicing to create monocistronic mRNAs that are translated separately, i.e. several strands of mRNA that each encode a single gene product. The result of this is that the genes contained in the operon are either expressed together or not at all. Several genes must be co-transcribed to define an operon.

<i>lac</i> operon Set genes encoding proteins and enzymes for lactose metabolism

The lactose operon is an operon required for the transport and metabolism of lactose in E. coli and many other enteric bacteria. Although glucose is the preferred carbon source for most bacteria, the lac operon allows for the effective digestion of lactose when glucose is not available through the activity of beta-galactosidase. Gene regulation of the lac operon was the first genetic regulatory mechanism to be understood clearly, so it has become a foremost example of prokaryotic gene regulation. It is often discussed in introductory molecular and cellular biology classes for this reason. This lactose metabolism system was used by François Jacob and Jacques Monod to determine how a biological cell knows which enzyme to synthesize. Their work on the lac operon won them the Nobel Prize in Physiology in 1965.

<span class="mw-page-title-main">Nucleoid</span> Region within a prokaryotic cell containing genetic material

The nucleoid is an irregularly shaped region within the prokaryotic cell that contains all or most of the genetic material. The chromosome of a prokaryote is circular, and its length is very large compared to the cell dimensions, so it needs to be compacted in order to fit. In contrast to the nucleus of a eukaryotic cell, it is not surrounded by a nuclear membrane. Instead, the nucleoid forms by condensation and functional arrangement with the help of chromosomal architectural proteins and RNA molecules as well as DNA supercoiling. The length of a genome widely varies and a cell may contain multiple copies of it.

<i>trp</i> operon Operon that codes for the components for production of tryptophan

The trp operon is a group of genes that are transcribed together, encoding the enzymes that produce the amino acid tryptophan in bacteria. The trp operon was first characterized in Escherichia coli, and it has since been discovered in many other bacteria. The operon is regulated so that, when tryptophan is present in the environment, the genes for tryptophan synthesis are repressed.

The galactose permease or GalP found in Escherichia coli is an integral membrane protein involved in the transport of monosaccharides, primarily hexoses, for utilization by E. coli in glycolysis and other metabolic and catabolic pathways (3,4). It is a member of the Major Facilitator Super Family (MFS) and is homologue of the human GLUT1 transporter (4). Below you will find descriptions of the structure, specificity, effects on homeostasis, expression, and regulation of GalP along with examples of several of its homologues.

The gene rpoS encodes the sigma factor sigma-38, a 37.8 kD protein in Escherichia coli. Sigma factors are proteins that regulate transcription in bacteria. Sigma factors can be activated in response to different environmental conditions. rpoS is transcribed in late exponential phase, and RpoS is the primary regulator of stationary phase genes. RpoS is a central regulator of the general stress response and operates in both a retroactive and a proactive manner: it not only allows the cell to survive environmental challenges, but it also prepares the cell for subsequent stresses (cross-protection). The transcriptional regulator CsgD is central to biofilm formation, controlling the expression of the curli structural and export proteins, and the diguanylate cyclase, adrA, which indirectly activates cellulose production. The rpoS gene most likely originated in the gammaproteobacteria.

The L-arabinose operon, also called the ara or araBAD operon, is an operon required for the breakdown of the five-carbon sugar L-arabinose in Escherichia coli. The L-arabinose operon contains three structural genes: araB, araA, araD, which encode for three metabolic enzymes that are required for the metabolism of L-arabinose. AraB (ribulokinase), AraA, AraD produced by these genes catalyse conversion of L-arabinose to an intermediate of the pentose phosphate pathway, D-xylulose-5-phosphate.

cAMP receptor protein

cAMP receptor protein is a regulatory protein in bacteria. CRP protein binds cAMP, which causes a conformational change that allows CRP to bind tightly to a specific DNA site in the promoters of the genes it controls. CRP then activates transcription through direct protein–protein interactions with RNA polymerase.

<span class="mw-page-title-main">Maltose-binding protein</span>

Maltose-binding protein (MBP) is a part of the maltose/maltodextrin system of Escherichia coli, which is responsible for the uptake and efficient catabolism of maltodextrins. It is a complex regulatory and transport system involving many proteins and protein complexes. MBP has an approximate molecular mass of 42.5 kilodaltons.

fis

fis is an E. coli gene encoding the Fis protein. The regulation of this gene is more complex than most other genes in the E. coli genome, as Fis is an important protein which regulates expression of other genes. It is supposed that fis is regulated by H-NS, IHF and CRP. It also regulates its own expression (autoregulation). Fis is one of the most abundant DNA binding proteins in Escherichia coli under nutrient-rich growth conditions.

<span class="mw-page-title-main">GcvB RNA</span>

The gcvB RNA gene encodes a small non-coding RNA involved in the regulation of a number of amino acid transport systems as well as amino acid biosynthetic genes. The GcvB gene is found in enteric bacteria such as Escherichia coli. GcvB regulates genes by acting as an antisense binding partner of the mRNAs for each regulated gene. This binding is dependent on binding to a protein called Hfq. Transcription of the GcvB RNA is activated by the adjacent GcvA gene and repressed by the GcvR gene. A deletion of GcvB RNA from Y. pestis changed colony shape as well as reducing growth. It has been shown by gene deletion that GcvB is a regulator of acid resistance in E. coli. GcvB enhances the ability of the bacterium to survive low pH by upregulating the levels of the alternate sigma factor RpoS. A polymeric form of GcvB has recently been identified. Interaction of GcvB with small RNA SroC triggers the degradation of GcvB by RNase E, lifting the GcvB-mediated mRNA repression of its target genes.

<span class="mw-page-title-main">SgrS RNA</span>

SgrS is a 227 nucleotide small RNA that is activated by SgrR in Escherichia coli during glucose-phosphate stress. The nature of glucose-phosphate stress is not fully understood, but is correlated with intracellular accumulation of glucose-6-phosphate. SgrS helps cells recover from glucose-phosphate stress by base pairing with ptsG mRNA and causing its degradation in an RNase E dependent manner. Base pairing between SgrS and ptsG mRNA also requires Hfq, an RNA chaperone frequently required by small RNAs that affect their targets through base pairing. The inability of cells expressing sgrS to create new glucose transporters leads to less glucose uptake and reduced levels of glucose-6-phosphate. SgrS is an unusual small RNA in that it also encodes a 43 amino acid functional polypeptide, SgrT, which helps cells recover from glucose-phosphate stress by preventing glucose uptake. The activity of SgrT does not affect the levels of ptsG mRNA of PtsG protein. It has been proposed that SgrT exerts its effects through regulation of the glucose transporter, PtsG.

The gal operon is a prokaryotic operon, which encodes enzymes necessary for galactose metabolism. Repression of gene expression for this operon works via binding of repressor molecules to two operators. These repressors dimerize, creating a loop in the DNA. The loop as well as hindrance from the external operator prevent RNA polymerase from binding to the promoter, and thus prevent transcription. Additionally, since the metabolism of galactose in the cell is involved in both anabolic and catabolic pathways, a novel regulatory system using two promoters for differential repression has been identified and characterized within the context of the gal operon.

<span class="mw-page-title-main">Autoinducer-2</span> Chemical compound

Autoinducer-2 (AI-2), a furanosyl borate diester or tetrahydroxy furan, is a member of a family of signaling molecules used in quorum sensing. AI-2 is one of only a few known biomolecules incorporating boron. First identified in the marine bacterium Vibrio harveyi, AI-2 is produced and recognized by many Gram-negative and Gram-positive bacteria. AI-2 arises by the reaction of 4,5-Dihydroxy-2,3-pentanedione, which is produced enzymatically with boric acid, and is recognized by the two-component sensor kinase LuxPQ in Vibrionaceae.

Bacterial small RNAs (bsRNA) are small RNAs produced by bacteria; they are 50- to 500-nucleotide non-coding RNA molecules, highly structured and containing several stem-loops. Numerous sRNAs have been identified using both computational analysis and laboratory-based techniques such as Northern blotting, microarrays and RNA-Seq in a number of bacterial species including Escherichia coli, the model pathogen Salmonella, the nitrogen-fixing alphaproteobacterium Sinorhizobium meliloti, marine cyanobacteria, Francisella tularensis, Streptococcus pyogenes, the pathogen Staphylococcus aureus, and the plant pathogen Xanthomonas oryzae pathovar oryzae. Bacterial sRNAs affect how genes are expressed within bacterial cells via interaction with mRNA or protein, and thus can affect a variety of bacterial functions like metabolism, virulence, environmental stress response, and structure.

Diauxic growth, diauxie or diphasic growth is any cell growth characterized by cellular growth in two phases. Diauxic growth, meaning double growth, is caused by the presence of two sugars on a culture growth media, one of which is easier for the target bacterium to metabolize. The preferred sugar is consumed first, which leads to rapid growth, followed by a lag phase. During the lag phase the cellular machinery used to metabolize the second sugar is activated and subsequently the second sugar is metabolized.

<span class="mw-page-title-main">Ars operon</span>

In molecular biology, the ars operon is an operon found in several bacterial taxon. It is required for the detoxification of arsenate, arsenite, and antimonite. This system transports arsenite and antimonite out of the cell. The pump is composed of two polypeptides, the products of the arsA and arsB genes. This two-subunit enzyme produces resistance to arsenite and antimonite. Arsenate, however, must first be reduced to arsenite before it is extruded. A third gene, arsC, expands the substrate specificity to allow for arsenate pumping and resistance. ArsC is an approximately 150-residue arsenate reductase that uses reduced glutathione (GSH) to convert arsenate to arsenite with a redox active cysteine residue in the active site. ArsC forms an active quaternary complex with GSH, arsenate, and glutaredoxin 1 (Grx1). The three ligands must be present simultaneously for reduction to occur.

The gua operon is responsible for regulating the synthesis of guanosine mono phosphate (GMP), a purine nucleotide, from inosine monophosphate. It consists of two structural genes guaB (encodes for IMP dehydrogenase or and guaA apart from the promoter and operator region.

<i>gab</i> operon

The gab operon is responsible for the conversion of γ-aminobutyrate (GABA) to succinate. The gab operon comprises three structural genes – gabD, gabT and gabP – that encode for a succinate semialdehyde dehydrogenase, GABA transaminase and a GABA permease respectively. There is a regulatory gene csiR, downstream of the operon, that codes for a putative transcriptional repressor and is activated when nitrogen is limiting.

References

  1. 1 2 Sahagan BG, Dahlberg JE (July 1979). "A small, unstable RNA molecule of Escherichia coli: spot 42 RNA. I. Nucleotide sequence analysis". J. Mol. Biol. 131 (3): 573–592. doi:10.1016/0022-2836(79)90008-1. PMID   390161.
  2. 1 2 Ikemura, Tochimichi; Dahlberg (25 July 1973). "Small ribonucleic acids of Escherichia coli. II. Noncoordinate accumulation during stringent control". The Journal of Biological Chemistry. 248 (14): 5033–5041. doi: 10.1016/S0021-9258(19)43667-3 . PMID   4577762.
  3. 1 2 Ikemura, T; Dahlberg (25 July 1973). "Small ribonucleic acids of Escherichia coli. I. Characterization by polyacrylamide gel electrophoresis and fingerprint analysis". Journal of Biological Chemistry. 248 (14): 5024–5032. doi: 10.1016/S0021-9258(19)43666-1 . PMID   4577761.
  4. 1 2 3 Hansen, Geir; Ahmad (24 January 2012). "Expression profiling reveals Spot 42 small RNA as a key regulator in the central metabolism of Aliivibrio salmonicida". BMC Genomics. 13: 37. doi:10.1186/1471-2164-13-37. PMC   3295665 . PMID   22272603.
  5. 1 2 3 4 Beisel CL, Storz G (2011). "The Base-Pairing RNA Spot 42 Participates in a Multioutput Feedforward Loop to Help Enact Catabolite Repression in Escherichia coli". Mol Cell. 41 (3): 286–297. doi:10.1016/j.molcel.2010.12.027. PMC   3072601 . PMID   21292161.
  6. 1 2 3 Moller, T; Franch T; Udesen C; Gerdes K; Valentin-Hansen P (2002). "Spot 42 RNA mediates discoordinate expression of the E. coli galactose operon". Genes Dev. 16 (13): 1696–1706. doi:10.1101/gad.231702. PMC   186370 . PMID   12101127.
  7. 1 2 3 Sahagan, Barbara; Dahlberg (5 July 1979). "A small, unstable RNA molecule of Escherichia coli: spot 42 RNA. II. Accumulation and distribution". Journal of Molecular Biology. 131 (3): 593–605. doi:10.1016/0022-2836(79)90009-3. PMID   229230.
  8. 1 2 3 Polayes DA, Rice PW, Garner MM, Dahlberg JE (July 1988). "Cyclic AMP-cyclic AMP receptor protein as a repressor of transcription of the spf gene of Escherichia coli". J. Bacteriol. 170 (7): 3110–3114. doi:10.1128/jb.170.7.3110-3114.1988. PMC   211256 . PMID   2454912.
  9. El Mouali, Y; Gaviria-Cantin, T; Sánchez-Romero, MA; Gibert, M; Westermann, AJ; Vogel, J; Balsalobre, C (June 2018). "CRP-cAMP mediates silencing of Salmonella virulence at the post-transcriptional level". PLOS Genetics. 14 (6): e1007401. doi:10.1371/journal.pgen.1007401. PMC   5991649 . PMID   29879120.
  10. Rice, PW; Dahlberg (December 1982). "A gene between polA and glnA retards growth of Escherichia coli when present in multiple copies: physiological effects of the gene for spot 42 RNA". Journal of Bacteriology. 152 (3): 1196–1210. doi:10.1128/jb.152.3.1196-1210.1982. PMC   221627 . PMID   6183252.
  11. Hatful, GF; Joyce (June 1986). "Deletion of the spf (spot 42 RNA) gene of Escherichia coli". Journal of Bacteriology. 166 (3): 746–750. doi:10.1128/jb.166.3.746-750.1986. PMC   215189 . PMID   2940230.
  12. 1 2 Bækkedal, Cecilie; Haugen, Peik (September 2015). "The Spot 42 RNA: A regulatory small RNA with roles in the central metabolism". RNA Biology. 12 (10): 1071–1077. doi:10.1080/15476286.2015.1086867. PMC   4829326 . PMID   26327359.
  13. 1 2 Polayes; Rice (May 1988). "DNA polymerase I activity in Escherichia coli is influenced by spot 42 RNA". Journal of Bacteriology. 170 (5): 2083–2088. doi:10.1128/jb.170.5.2083-2088.1988. PMC   211090 . PMID   2452153.
  14. Joyce; Grindley (December 1982). "Identification of two genes immediately downstream from the polA gene of Escherichia coli". Journal of Bacteriology. 152 (3): 1211–1219. doi:10.1128/jb.152.3.1211-1219.1982. PMC   221628 . PMID   6183253.
  15. Rice; Polayes (August 1987). "Spot 42 RNA of Escherichia coli is not an mRNA". Journal of Bacteriology. 169 (8): 3850–3852. doi:10.1128/jb.169.8.3850-3852.1987. PMC   212481 . PMID   2440852.
  16. Wright PR, Richter AS, Papenfort K, Mann M, Vogel J, Hess WR, Backofen R, Georg J (2013). "Comparative genomics boosts target prediction for bacterial small RNAs". Proc Natl Acad Sci U S A. 110 (37): E3487–E3496. Bibcode:2013PNAS..110E3487W. doi: 10.1073/pnas.1303248110 . PMC   3773804 . PMID   23980183.
  17. Wright PR, Georg J, Mann M, Sorescu DA, Richter AS, Lott S, Kleinkauf R, Hess WR, Backofen R (2014). "CopraRNA and IntaRNA: predicting small RNA targets, networks and interaction domains". Nucleic Acids Res. 42 (Web Server): W119–23. doi:10.1093/nar/gku359. PMC   4086077 . PMID   24838564.
  18. Holmqvist E, Wright PR, Li L, Bischler T, Barquist L, Reinhardt R, Backofen R, Vogel J (2016). "Global RNA recognition patterns of post-transcriptional regulators Hfq and CsrA revealed by UV crosslinking in vivo". EMBO J. 35 (9): 991–1011. doi:10.15252/embj.201593360. PMC   5207318 . PMID   27044921.