Comparison of the catalytic roles played by the KMSKS motif in the human and Bacillus stearothermophilus trosyl-tRNA synthetases

The Class I aminoacyl-tRNA synthetases are characterized by two signature sequence motifs, "HIGH" and "KMSKS." In Bacillus stearothermophilus tyrosyl-tRNA synthetase, the KMSKS motif (230KFGKT234) has been shown to stabilize the transition state for tyrosine activation through interactions with the pyrophosphate moiety of ATP. In most eukaryotic tyrosyl-tRNA synthetases, the second lysine in the KMSKS motif is replaced by a serine or an alanine residue. Recent kinetic studies indicate that potassium functionally compensates for the absence of the second lysine in the human tyrosyl-tRNA synthetase (222KKSSS226). In this paper, site-directed mutagenesis and pre-steady state kinetics are used to determine the roles that serines 224, 225, and 226 play in catalysis of the tyrosine activation reaction. In addition, the catalytic role played by a downstream lysine conserved in eukaryotic tyrosyl-tRNA synthetases, Lys-231, is investigated. Replacing Ser-224 and Ser-226 with alanine decreases the forward rate constant 7.5- and 60-fold, respectively. In contrast, replacing either Ser-225 or Lys-231 with alanine has no effect on the catalytic activity of the enzyme. These results are consistent with the hypothesis that the KMSSS sequence in human tyrosyl-tRNA synthetase stabilizes the transition state for the tyrosine activation reaction by interacting with the pyrophosphate moiety of ATP. In addition, although they play similar roles in catalysis, the overall contribution of the KMSKS motif to catalysis appears to be significantly less in human tyrosyl-tRNA synthetase than it is in the B. stearothermophilus enzyme.     

Sequence, structural and evolutionary relationships between class 2 aminoacyl-tRNA synthetases

Class 2 aminoacyl-tRNA synthetases, which include the enzymes for alanine, aspartic acid, asparagine, glycine, histidine, lysine, phenylalanine, proline, serine and threonine, are characterised by three distinct sequence motifs 1,2 and 3 (reference 1). The structural and evolutionary relatedness of these ten enzymes are examined using alignments of primary sequences from prokaryotic and eukaryotic sources and the known three dimensional structure of seryl-tRNA synthetase from E. coli. It is shown that motif 1 forms part of the dimer interface of seryl-tRNA synthetase and motifs 2 and 3 part of the putative active site. It is further shown that the seven alpha 2 dimeric synthetases can be subdivided into class 2a (proline, threonine, histidine and serine) and class 2b (aspartic acid, asparagine and lysine), each subclass sharing several important characteristic sequence motifs in addition to those characteristic of class 2 enzymes in general. The alpha 2 beta 2 tetrameric enzymes (for glycine and phenylalanine) show certain special features in common as well as some of the class 2b motifs. In the alanyl-tRNA synthetase only motif 3 and possibly motif 2 can be identified. The sequence alignments suggest that the catalytic domain of other class 2 synthetases should resemble the antiparallel domain found in seryl-tRNA synthetase. Predictions are made about the sequence location of certain important helices and beta-strands in this domain as well as suggestions concerning which residues are important in ATP and amino acid binding. Strong homologies are found in the N-terminal extensions of class 2b synthetases and in the C-terminal extensions of class 2a synthetases suggesting that these putative tRNA binding domains have been added at a later stage in evolution to the catalytic domain.     

Catalytic activity of aminoacyl tRNA synthetases and its implications for the origin of life. I. Aminoacyl adenylate formation in tyrosyl tRNA synthetase

The changes in the catalytic activity resulting from amino acid substitutions in the active site region have been theoretically modeled for tyrosyl tRNA synthetase (Tyr-RS). The catalytic activity was calculated as the differential stabilization of the transition state using electrostatic approximation. The results indicate that charged residues His45, His48, Asp78, Asp176, Asp194, Lys225, Lys230, Lys233, Arg265, and Lys268 play essential roles in catalysis of aminoacyl adenylate formation in Tyr-RS, which is in general agreement with previously known experimental data for residues 45, 48, 194, 230, and 233. These catalytic residues have also been used to search for sequence homology patterns among class I aminoacyl RSs of which HIGH and KMSKS conserved sequence motifs are well known. His45 and His48 belong to the HIGH signature sequence of class I aminoacyl tRNA synthetases (aRSs), whereas Arg265 and Lys268 can constitute a part of the KMSKS charge pattern. Lys225, Lys230, and Lys233 may be part of the conservative substitution pattern [HKR]-X(4)-[HKR]-X(2)-[HKR], and Asp194 is part of the new GSDQ motif. This demonstrates that the three dimensional charge distribution near the active site is an essential feature of the catalytic activity of aRS and that the theoretical technique used in this work can be utilized in searches for the catalytically important residues that may provide a clue for a charge residue pattern conserved in evolution. The appearance of patterns I-IV in Arg-, Gln-, Met-, Ile-, Leu-, Trp-, Val-, Glu-, Cys-, and Tyr-RS indicates that all these enzymes could have the same ancestor.     

Differential G-alpha interaction capacities of the GoLoco motifs in Rap GTPase activating proteins

GoLoco motif proteins act as guanine nucleotide dissociation inhibitors (GDIs) for G-protein alpha subunits of the adenylyl cyclase-inhibitory (Galpha(i/o)) class. Rap1GAP2 is a newly identified GoLoco motif- and RapGAP domain-containing protein, and thus is considered a potential integrator of heterotrimeric and monomeric GTPase signaling. Primary sequence analysis indicated that the Rap1GAP2 GoLoco motif contains a lysine (Lys-75), rather than an arginine, at the crucial residue responsible for binding the alpha and beta phosphates of GDP and exerting GDI activity. To determine the functional outcome of this sequence variation we conducted a biophysical analysis of the human Rap1GAP2b/c GoLoco motif. We found that human Rap1GAP2b/c was deficient in GDI activity and Galpha interaction capability. Mutation of lysine-75 to arginine could not regain functional activity of the Rap1GAP2b/c GoLoco motif. Thus, the Rap1GAP2b/c GoLoco motif can be classed as inactive towards Galpha subunits. We also found that the Rap1GAP1a GoLoco motif, which lacks seven N-terminal amino acid residues present in canonical GoLoco motifs, does not interact with Galpha(i1). In contrast, the GoLoco motif of Rap1GAP1b, which is canonical in primary sequence, was found to interact with Galpha(i1).GDP.     

Analysis of the domain structure of elongation factor-2 kinase by mutagenesis.

A number of elongation factor-2 kinase (eEF-2K) mutants were constructed to investigate features of this kinase that may be important in its activity. Typical protein kinases possess a highly conserved lysine residue in subdomain II which follows the GXGXXG motif of subdomain I. Mutation of two lysine residues, K340 and K346, which follow the GXGXXG motif in eEF-2K had no effect on activity, showing that such a lysine residue is not important in eEF-2K activity. Mutation of a conserved pair of cysteine residues C-terminal to the GXGXXG sequence, however, completely inactivated eEF-2K. The eEF-2K CaM binding domain was localised to residues 77-99 which reside N-terminal to the catalytic domain. Tryptophan 84 is an important residue within this domain as mutation of this residue completely abolishes CaM binding and eEF-2K activity. Removal of approximately 130 residues from the C-terminus of eEF-2K completely abolished autokinase activity; however, removal of only 19 residues inhibited eEF-2 kinase activity but not autokinase activity, suggesting that a short region at the C-terminal end may be important in interacting with eEF-2. Likewise, removal of between 75 and 100 residues from the N-terminal end completely abolished eEF-2K activity.     

Rules for the recognition of dilysine retrieval motifs by coatomer

Cytoplasmic dilysine motifs on transmembrane proteins are captured by coatomer α‐COP and β′‐COP subunits and packaged into COPI‐coated vesicles for Golgi‐to‐ER retrieval. Numerous ER/Golgi proteins contain K(x)Kxx motifs, but the rules for their recognition are unclear. We present crystal structures of α‐COP and β′‐COP bound to a series of naturally occurring retrieval motifs—encompassing KKxx, KxKxx and non‐canonical RKxx and viral KxHxx sequences. Binding experiments show that α‐COP and β′‐COP have generally the same specificity for KKxx and KxKxx, but only β′‐COP recognizes the RKxx signal. Dilysine motif recognition involves lysine side‐chain interactions with two acidic patches. Surprisingly, however, KKxx and KxKxx motifs bind differently, with their lysine residues transposed at the binding patches. We derive rules for retrieval motif recognition from key structural features: the reversed binding modes, the recognition of the C‐terminal carboxylate group which enforces lysine positional context, and the tolerance of the acidic patches for non‐lysine residues.     

A newly identified, essential catalytic residue in a critical secondary structure element in the integrase family of site-specific recombinases is conserved in a similar element in eucaryotic type IB topoisomerases.

The integrase family of site-specific recombinases catalyzes conservative rearrangements between defined segments of DNA. A highly conserved tetrad (RHRY) of catalytic residues is essential for this process. This tetrad is dispersed in two motifs in the linear sequence, but is configured appropriately in the catalytic pocket to execute the strand cleavage and rejoining reactions. A third conserved motif has been identified in the Xer subgroup of the integrase family. Mutational analysis of 12 conserved residues in this motif in the XerD protein from Salmonella typhimurium led to the identification of an essential fifth catalytic residue (lysine 172) which is implicated in strand cleavage or exchange. This lysine residue occupies part of the turn of an antiparallel beta-hairpin which forms one side of the catalytic cleft in XerD, and is found at similar positions among evolutionarily diverse integrase family members. Related antiparallel beta-hairpins are present in eucaryotic type IB topoisomerase enzymes which also contain a critical lysine residue in the turn of the hairpin. In both the integrase family and eucaryotic type IB topoisomerases, the catalytic lysine residues are in close contact with the substrates and may play similar roles in influencing the reactivity of the phosphotyrosine intermediates formed during reactions catalyzed by both enzymes.     

Interaction between the N-terminal SH3 domain of Nck-alpha and CD3-epsilon-derived peptides: non-canonical and canonical recognition motifs

The first SH3 domain (SH3.1) of Nckalpha specifically recognizes the proline-rich region of CD3varepsilon, a subunit of the T cell receptor complex. We have solved the NMR structure of Nckalpha SH3.1 that shows the characteristic SH3 fold consisting of two antiparallel beta-sheets tightly packed against each other. According to chemical shift mapping analysis, a peptide encompassing residues 150-166 of CD3varepsilon binds at the canonical SH3 binding site. An exhaustive comparison with the structures of other SH3 domains able and unable to bind CD3varepsilon reveals that Nckalpha SH3.1 recognises a non-canonical PxxPxxDY motif that orientates at the binding site as a class II ligand. A positively charged residue (K/R) at position -2 relative to the WW sequence at the beginning of strand beta3 is crucial for PxxDY recognition. A 14-mer optimised Nckalpha SH3.1 ligand was found using a multi-substitution approach. Based on NMR data, this improved ligand binds Nckalpha SH3.1 through a PxxPxRDY motif that combines specific stabilising interactions corresponding to both canonical class II, PxxPx(K/R), and non-canonical PxxPxxDY motifs. This explains its higher capacity for Nckalpha SH3.1 binding relative to the wild type sequence.     

Mutational analysis of two putative catalytic motifs of the type IV restriction endonuclease Eco57I

The role of two sequence motifs (SM) as putative cleavage catalytic centers (77)PDX(13)EAK (SM I) and (811)PDX(20)DQK (SM II) of type IV restriction endonuclease Eco57I was studied by site-directed mutational analysis. Substitutions within SM I; D78N, D78A, D78K, and E92Q reduced cleavage activity of Eco57I to a level undetectable both in vivo and in vitro. Residual endonucleolytic activity of the E92Q mutant was detected only when the Mg(2+) in the standard reaction mixture was replaced with Mn(2+). The mutants D78N and E92Q retained the ability to interact with DNA specifically. The mutants also retained DNA methylation activity of Eco57I. The properties of the SM I mutants indicate that Asp(78) and Glu(92) residues are essential for cleavage activity of the Eco57I, suggesting that the sequence motif (77)PDX(13)EAK represents the cleavage active site of this endonuclease. Eco57I mutants containing single amino acid substitutions within SM II (D812A, D833N, D833A) revealed only a small or moderate decrease of cleavage activity as compared with wild-type Eco57I, indicating that the SM II motif does not represent the catalytic center of Eco57I. The results, taken together, allow us to conclude that the Eco57I restriction endonuclease has one catalytic center for cleavage of DNA.     

Mutational analysis of Chlorella virus DNA ligase: catalytic roles of domain I and motif VI.

A conserved catalytic core of the ATP-dependent DNA ligases is composed of an N-terminal domain (domain 1, containing nucleotidyl transferase motifs I, III, IIIa and IV) and a C-terminal domain (domain 2, containing motif VI) with an intervening cleft. Motif V links the two structural domains. Deletion analysis of the 298 amino acid Chlorella virus DNA ligase indicates that motif VI plays a critical role in the reaction of ligase with ATP to form ligase-adenylate, but is dispensable for the two subsequent steps in the ligation pathway; DNA-adenylate formation and strand closure. We find that formation of a phosphodiester at a pre-adenylated nick is subject to a rate limiting step that does not apply during the sealing of nicked DNA by ligase-adenylate. This step, presumably conformational, is accelerated or circumvented by deleting five amino acids of motif VI. The motif I lysine nucleophile (Lys27) is not required for strand closure by wild-type ligase, but this residue enhances the closure rate by a factor of 16 when motif VI is truncated. We find that a more extensively truncated ligase consisting of only N-terminal domain 1 and motif V is inert in ligase--adenylate formation, but competent to catalyze strand closure at a pre-adenylated nick. These results suggest that different enzymic catalysts facilitate the three steps of the DNA ligase reaction.     

A conserved non-canonical motif in the pseudoactive site of the ROP5 pseudokinase domain mediates its effect on Toxoplasma virulence

The ROP5 family is a closely related set of polymorphic pseudokinases that are critical to the ability of Toxoplasma to cause disease. Polymorphisms in ROP5 also make it a major determinant of strain-specific differences in virulence. ROP5 possesses all of the major kinase motifs required for catalysis except for a substitution at the catalytic Asp. We show that this substitution in the catalytic loop of ROP5 is part of a motif conserved in other pseudokinases of both Toxoplasma and human origin, and that this motif is required for the full activity in vivo of ROP5. This suggests evolutionary selection at this site for a biochemical function, rather than simple drift away from catalysis. We present the crystal structures of a virulent isoform of ROP5 both in its ATP-bound and -unbound states and have demonstrated that despite maintaining the canonical ATP-binding motifs, ROP5 binds ATP in a distorted conformation mediated by unusual magnesium coordination sites that would not be predicted from the primary sequence. In addition, we have mapped the polymorphisms spread throughout the primary sequence of ROP5 to two major surfaces, including the activation segment of ROP5. This suggests that the pseudoactive site of this class of pseudokinases may have evolved to use the canonical ATP-binding motifs for non-catalytic signaling through allostery.     

Structural snapshots of the KMSKS loop rearrangement for amino acid activation by bacterial tyrosyl-tRNA synthetase

Tyrosyl-tRNA synthetase (TyrRS) has been studied extensively by mutational and structural analyses to elucidate its catalytic mechanism. TyrRS has the HIGH and KMSKS motifs that catalyze the amino acid activation with ATP. In the present study, the crystal structures of the Escherichia coli TyrRS catalytic domain, in complexes with l-tyrosine and a l-tyrosyladenylate analogue, Tyr-AMS, were solved at 2.0A and 2.7A resolution, respectively. In the Tyr-AMS-bound structure, the 2'-OH group and adenine ring of the Tyr-AMS are strictly recognized by hydrogen bonds. This manner of hydrogen-bond recognition is conserved among the class I synthetases. Moreover, a comparison between the two structures revealed that the KMSKS loop is rearranged in response to adenine moiety binding and hydrogen-bond formation, and the KMSKS loop adopts the more compact ("semi-open") form, rather than the flexible, open form. The HIGH motif initially recognizes the gamma-phosphate, and then the alpha and gamma-phosphates of ATP, with a slight rearrangement of the residues. The other residues around the substrate also accommodate the Tyr-AMS. This induced-fit form presents a novel "snapshot" of the amino acid activation step in the aminoacylation reaction by TyrRS. The present structures and the T.thermophilus TyrRS ATP-free and bound structures revealed that the extensive induced-fit conformational changes of the KMSKS loop and the local conformational changes within the substrate binding site form the basis for driving the amino acid activation step: the KMSKS loop adopts the open form, transiently shifts to the semi-open conformation according to the adenosyl moiety binding, and finally assumes the rigid ATP-bound, closed form. After the amino acid activation, the KMSKS loop adopts the semi-open form again to accept the CCA end of tRNA for the aminoacyl transfer reaction.     

The double-length tyrosyl-tRNA synthetase from the eukaryote Leishmania major forms an intrinsically asymmetric pseudo-dimer

The single tyrosyl-tRNA synthetase (TyrRS) gene in trypanosomatid genomes codes for a protein that is twice the length of TyrRS from virtually all other organisms. Each half of the double-length TyrRS contains a catalytic domain and an anticodon-binding domain; however, the two halves retain only 17% sequence identity to each other. The structural and functional consequences of this duplication and divergence are unclear. TyrRS normally forms a homodimer in which the active site of one monomer pairs with the anticodon-binding domain from the other. However, crystal structures of Leishmania major TyrRS show that, instead, the two halves of a single molecule form a pseudo-dimer resembling the canonical TyrRS dimer. Curiously, the C-terminal copy of the catalytic domain has lost the catalytically important HIGH and KMSKS motifs characteristic of class I aminoacyl-tRNA synthetases. Thus, the pseudo-dimer contains only one functional active site (contributed by the N-terminal half) and only one functional anticodon recognition site (contributed by the C-terminal half). Despite biochemical evidence for negative cooperativity between the two active sites of the usual TyrRS homodimer, previous structures have captured a crystallographically-imposed symmetric state. As the L. major TyrRS pseudo-dimer is inherently asymmetric, conformational variations observed near the active site may be relevant to understanding how the state of a single active site is communicated across the dimer interface. Furthermore, substantial differences between trypanosomal TyrRS and human homologs are promising for the design of inhibitors that selectively target the parasite enzyme.     

The nrfI gene is essential for the attachment of the active site haem group of Wolinella succinogenes cytochrome c nitrite reductase

The cytochrome c nitrite reductase complex (NrfHA) is the terminal enzyme in the electron transport chain catalysing nitrite respiration of Wolinella succinogenes. The catalytic subunit NrfA is a pentahaem cytochrome c containing an active site haem group which is covalently bound via the cysteine residues of a unique CWTCK motif. The lysine residue serves as the axial ligand of the haem iron. The other four haem groups of NrfA are bound by conventional haem-binding motifs (CXXCH). The nrfHAIJ locus was restored on the genome of the W. succinogenes DeltanrfAIJ deletion mutant by integration of a plasmid, thus enabling the expression of modified alleles of nrfA and nrfI. A mutant (K134H) was constructed which contained a nrfA gene encoding a CWTCH motif instead of CWTCK. NrfA of strain K134H was found to be synthesized with five bound haem groups, as judged by matrix-assisted laser-desorption/ionization (MALDI) mass spectrometry. Its nitrite reduction activity with reduced benzyl viologen was 40% of the wild-type activity. Ammonia was formed as the only product of nitrite reduction. The mutant did not grow by nitrite respiration and its electron transport activity from formate to nitrite was 5% of that of the wild-type strain. The predicted nrfI gene product is similar to proteins involved in system II cytochrome c biogenesis. A mutant of W. succinogenes (stopI) was constructed that contained a nrfHAIJ gene cluster with the nrfI codons 47 and 48 altered to stop codons. The NrfA protein of this mutant did not catalyse nitrite reduction and lacked the active site haem group, whereas the other four haem groups were present. Mutant (K134H/stopI) which contained the K134H modification in NrfA in addition to the inactivated nrfI gene had essentially the same properties as strain K134H. NrfA from strain K134H/stopI contained five haem groups. It is concluded that NrfI is involved in haem attachment to the CWTCK motif in NrfA but not to any of the CXXCH motifs. The nrfI gene is obviously dispensable for maturation of a modified NrfA protein containing a CWTCH motif instead of CWTCK. Therefore, NrfI might function as a specific haem lyase that recognizes the active site lysine residue of NrfA. A similar role has been proposed for NrfE, F and G of Escherichia coli, although these proteins share no overall sequence similarity to NrfI and belong to system I cytochrome c biogenesis, which differs fundamentally from system II.     

Correlating amino acid conservation with function in tyrosyl-tRNA synthetase

Sequence comparisons have been combined with mutational and kinetic analyses to elucidate how the catalytic mechanism of Bacillus stearothermophilus tyrosyl-tRNA synthetase evolved. Catalysis of tRNA(Tyr) aminoacylation by tyrosyl-tRNA synthetase involves two steps: activation of the tyrosine substrate by ATP to form an enzyme-bound tyrosyl-adenylate intermediate, and transfer of tyrosine from the tyrosyl-adenylate intermediate to tRNA(Tyr). Previous investigations indicate that the class I conserved KMSKS motif is involved in only the first step of the reaction (i.e. tyrosine activation). Here, we demonstrate that the class I conserved HIGH motif also is involved only in the tyrosine activation step. In contrast, one amino acid that is conserved in a subset of the class I aminoacyl-tRNA synthetases, Thr40, and two amino acids that are present only in tyrosyl-tRNA synthetases, Lys82 and Arg86, stabilize the transition states for both steps of the tRNA aminoacylation reaction. These results imply that stabilization of the transition state for the first step of the reaction by the class I aminoacyl-tRNA synthetases preceded stabilization of the transition state for the second step of the reaction. This is consistent with the hypothesis that the ability of aminoacyl-tRNA synthetases to catalyze the activation of amino acids with ATP preceded their ability to catalyze attachment of the amino acid to the 3' end of tRNA. We propose that the primordial aminoacyl-tRNA synthetases replaced a ribozyme whose function was to promote the reaction of amino acids and other small molecules with ATP.     

Iterative database searches demonstrate that glycoside hydrolase families 27, 31, 36 and 66 share a common evolutionary origin with family 13

A catalytic sequence motif PDX10-30(E/D)XK is found in many restriction enzymes. On the basis of sequence similarities and mapping of the conserved residues to the crystal structure of NgoMIV we suggest that residues D160, K182, R186, R188 and E195 contribute to the catalytic/DNA binding site of the Ecl18kI restriction endonuclease. Mutational analysis confirms the functional significance of the conserved residues of Ecl18kI. Therefore, we conclude that the active site motif 159VDX21KX12E of Ecl18kI differs from the canonical PDX10-30(E/D)XK motif characteristic for most of the restriction enzymes. Moreover, we propose that two subfamilies of endonucleases Ecl18kI/PspGI/EcoRII and Cfr10I/Bse634I/NgoMIV, specific, respectively, for CCNGG/CCWGG and RCCGGY/GCCGGC sites, share conserved active site architecture and DNA binding elements.     

A minimal TrpRS catalytic domain supports sense/antisense ancestry of class I and II aminoacyl-tRNA synthetases

The emergence of polypeptide catalysts for amino acid activation, the slowest step in protein synthesis, poses a significant puzzle associated with the origin of biology. This problem is compounded as the 20 contemporary aminoacyl-tRNA synthetases belong to two quite distinct families. We describe here the use of protein design to show experimentally that a minimal class I aminoacyl-tRNA synthetase active site might have functioned in the distant past. We deleted the anticodon binding domain from tryptophanyl-tRNA synthetase and fused the discontinuous segments comprising its active site. The resulting 130 residue minimal catalytic domain activates tryptophan. This residual catalytic activity constitutes the first experimental evidence that the conserved class I signature sequences, HIGH and KMSKS, might have arisen in-frame, opposite motifs 2 and 1 from class II, as complementary sense and antisense strands of the same ancestral gene.     

The double lysine motif of tapasin is a retrieval signal for retention of unstable MHC class I molecules in the endoplasmic reticulum

Tapasin (tpn), an essential component of the MHC class I (MHC I) loading complex, has a canonical double lysine motif acting as a retrieval signal, which mediates retrograde transport of escaped endoplasmic reticulum (ER) proteins from the Golgi back to the ER. In this study, we mutated tpn with a substitution of the double lysine motif to double alanine (GFP-tpn-aa). This mutation abolished interaction with the coatomer protein complex I coatomer and resulted in accumulation of GFP-tpn-aa in the Golgi compartment, suggesting that the double lysine is important for the retrograde transport of tpn from late secretory compartments to the ER. In association with the increased Golgi distribution, the amount of MHC I exported from the ER to the surface was increased in 721.220 cells transfected with GFP-tpn-aa. However, the expressed MHC I were less stable and had increased turnover rate. Our results suggest that tpn with intact double lysine retrieval signal regulates retrograde transport of unstable MHC I molecules from the Golgi back to the ER to control the quality of MHC I Ag presentation.