Recognition of nonproline N-terminal residues by the Pro/N-degron pathway

GID4 Can Recognize Nt-Sequence Motifs Lacking Pro.

GID4, a subunit of the GID Ub ligase and the main N-recognin of the Pro/N-degron pathway, has been shown to recognize (specifically bind to), both in vitro and in vivo, protein substrates that contain the Nt-Pro residue or a Pro at position 2, in the also required presence of cognate adjoining sequence motifs (Fig. 1) (43⇓–45). For example, in vivo yeast-based two-hybrid (Y2H) binding assays could detect the binding of S. cerevisiae GID4 to the Nt-Pro–bearing gluconeogenic enzyme Pro-FBP1 but not to its otherwise identical Nt-Ser–bearing mutant Ser-FBP1 (43). This recognition specificity of GID4 underlied the GID4-dependent in vivo degradation of Pro-FBP1, in contrast to metabolic stability of Ser-FBP1 (43). Yeast GID4 was also found to bind, in Y2H assays, to the Nt-sequence SPSV, which contained a Pro at position 2. That Nt-sequence was identical to the Nt-sequence SPSK of the short-lived S. cerevisiae gluconeogenic enzyme PCK1 (an in vivo substrate of GID4), save for the presence of Val, instead of Lys, at position 4 of SPSV (43).

Crystal structures of human GID4 bound to different Nt-Pro–containing short peptides showed that the main-chain α-imino group of Nt-Pro in these peptides is coordinated by five GID4-mediated hydrogen bonds, while the pyrrolidine side chain of Pro is bound to a hydrophobic patch in GID4 (44). It seemed possible that the pyrrolidine-contacting hydrophobic pocket of GID4 might tolerate some hydrophobic residues other than Pro. To address this conjecture, we used a mutation-scanning peptide SPOT array to characterize the substrate specificity of human GID4. Each residue of the PGLVKSAS peptide was substituted, systematically, with 1 of 18 amino acids of the genetic code, save for Cys and Trp, because of difficulties in using the latter residues to synthesize SPOT peptides. The array was incubated with full-length human GID4 N-terminally tagged with glutathione transferase (GST), and the binding of GID4 to peptide spots was detected using antibody to GST and chemilumenescence (Fig. 2).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Use of peptide arrays to analyze substrate recognition by human GID4. (A) Binding patterns of the full-length human GID4 to an array of the spotted Nt-Pro–bearing peptide PGLVKASAS (top row, denoted “wild-type” [wt]) and its indicated derivatives (Materials and Methods). (B) The array binding data were quantified and the relative binding of GID4 to each peptide is shown, with strong to weak binding scaled from black to white. (C) These substrate preferences of GID4 are also shown as a sequence logo.

We found that several hydrophobic residues (Ile, Leu, Val, and Phe) could substitute for Nt-Pro in the above SPOT sequence without eliminating the binding of resulting peptides to human GID4 (Fig. 2). In agreement with previous evidence (44), the binding of GID4 was stronger if a small side-chain residue was present at position 2. In addition, no significant binding of GID4 was observed with peptides bearing Gly, Asp, or Pro at position 3. GID4 exhibited at most a minor dependence of its binding to array’s peptides beyond position 3 (Fig. 2).

To further analyze these binding patterns, we synthesized a set of six-residue peptides (XGLWKS; X = any Nt-residue), and used an isothermal titration calorimetry (ITC) binding assay to quantify their interactions with the binding-competent GID4^116−300 fragment of the full-length human GID4¹⁻³⁰⁰ (44) (Fig. 3). Since the binding of XGLVKSAS peptides by GID4 did not exhibit a significant sequence specificity beyond position 3 (Fig. 2), the four-residue XGLW was extended by the Lys-Ser (KS) Ct-sequence to increase solubility of peptides that were used in ITC binding assays.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Affinities (K_ds) of XGLWKS peptides for the GID4^116–300 fragment of the full-length human GID4^1–300 (44). NDB, no detectable binding to GID4^116–300 (K_d > 0.5 mM). Superscript “a” (^a) signifies ITC measurements with CGLWKS performed using a cysteine-deficient mutant of GID4^116–300 (Materials and Methods and SI Appendix, Fig. S2).

In agreement with SPOT-array data (Fig. 2), GID4^116−300 exhibited significant affinities for the XGLWKS peptides in which X was either Ile (K_d of 16 μM), Leu (K_d of 21 μM), Val (K_d of 36 μM), or Phe (K_d of 40 μM). The affinity of GID4^116−300 for the otherwise identical but Nt-Pro–bearing PGLWKS (K_d of 1.9 μM) was still significantly stronger than for the other four peptides (Fig. 3 and SI Appendix, Fig. S1). While the binding of GID4^116−300 to the Nt-Met–bearing sequence MGLWKS could still be detected by ITC, the binding affinity (K_d of 72 μM) was even lower than the affinity of GID4^116−300 for the FGLWKS peptide (Figs. 1 and 3 and SI Appendix, Fig. S1).

Unexpectedly, we found that GID4^116−300 had a relatively high affinity (K_d of 7 μM) for the Nt-Cys–bearing CGLWKS peptide (SI Appendix, Fig. S2A). Inasmuch as there was no detectable binding of GID4^116−300 to the SGLWKS peptide, whose Nt-Ser is sterically and chemically similar to Nt-Cys (Fig. 3), we suspected that the binding to CGLWKS may have been artifactually strengthened by the formation of a disulfide bond between Nt-Cys of CGLWKS and a Cys residue of GID4^116−300. In the 3D crystal structure of GID4 (44), one of two Cys solvent-exposed residues, Cys-156 and Cys-261, might be involved in the disulfide bond formation with Nt-Cys of CGLWKS (SI Appendix, Fig. S2 D–F). To address this possibility, we constructed a human GID4^116−300 mutant in which these two cysteines were converted to Ser. In agreement with our assumption, the GID4^{116−300, C156S, C261S} mutant retained a nearly wild-type (high) affinity for the Nt-Pro–bearing PGLWKS peptide but did not exhibit a detectable binding to the CGLWKS peptide (SI Appendix, Fig. S2). In sum and for the reasons above, an Nt-Cys–bearing polypeptide is unlikely to be a physiologically relevant GID4 ligand.

Structural Basis for the Binding of Human GID4 to Non-Pro Nt-Residues.

To address molecular basis of the GID4-mediated recognition of Nt-sequences that bear non-Pro Nt-residues, we determined high-resolution crystal structures of human GID4^124−289 in complexes with the six-residue peptides IGLWKS and VGLWKS, respectively (Figs. 4 and 5A; see SI Appendix, Table S1 for crystal diffraction data and model refinement statistics). Detailed comparisons showed that the structures of GID4^124−289-IGLWKS and GID4^124−289-VGLWKS are very similar to that of GID4^124−289-PGLWKS, with backbone RMSDs of less than 0.5 Å (Fig. 4).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Structural basis of recognition of nonproline N-terminal residues by human GID4. (A) Superposition, in ribbon diagrams, of the overall crystal structures of complexes of GID4^124–289 with, respectively, PGLWKS (salmon, PDB ID code 6CDG), VGLWKS (green), and IGLWKS (cyan) peptides (see also Materials and Methods and SI Appendix, Table S1). (B) Close-up view of similar (conserved) interactions between the substrate-binding chamber of GID4 and the first four residues of the above peptides (PGLW, salmon; VGLW, green; and IGLW, cyan). The most relevant residues of GID4 are shown as gray/red/blue sticks vis-à-vis colored peptides. Hydrogen bonds are indicated as black dash lines. (C) Close-up view of the recognition of the Nt-Ile–bearing IGLWKS peptide in the substrate-binding chamber of human GID4. Denoted in red are the first four residues of the six-residue peptide. These residues are depicted as cyan/red/blue sticks. The most relevant residues of the GID4 chamber are either in black or in green (the hydrophilic Thr-173 residue), and are depicted via surface representations and gray sticks.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

(A) Plot of affinities (relative association constants, K_as) of the Nt-Ile–bearing peptide IGLWKS for GID4^116–300 and its indicated single-residue mutants, with the affinity of wild-type GID4^116–300 for IGLWKS taken as 100%. NDB, no detectable binding (K_d > 0.5 mM). (B) Plot of affinities (relative association constants, K_as) of the Nt-Ile–bearing peptide IGLWKS for GID4^116–300 and its indicated position-173 mutants, with the affinity of wild-type GID4^116–300 for IGLWKS taken as 100%. (C) Plot of affinities (relative association constants, K_as) of the Nt-Pro–bearing peptide PGLWKS for GID4^116–300 and its indicated position-173 mutants, with the affinity of wild-type GID4^116–300 for PGLWKS taken as 100%. In B and C, the highest-affinity levels for position-173 GID4 mutants (they differ for the IGLWKS versus PGLWKS peptides) are highlighted in red.

The α-amino groups of the above non–Nt-Pro peptides are coordinated by a set of conserved hydrogen bonds, similar to the set that coordinates the Nt–α-imino group of the Nt-Pro peptide. Specifically, the free α-amino group of Nt–Val-1 or Nt–Ile-1 forms hydrogen bonds with the side chains of Glu-237 and Tyr-258 of GID4, while the carbonyl oxygen of Val-1 or Ile-1 is further stabilized, through another hydrogen bond, by Gln-132 of GID4. The main chain of the second residue (Gly-2) of the six-residue peptides is anchored by the main chain of Ser-253 via two hydrogen bonds. The main chain of the third residue (Leu-3) of a bound peptide forms two hydrogen bonds with the side chain of Gln-282. Trp-4, the fourth residue of these peptides, interacts, through a hydrogen bond, with Gly-251 of GID4 (Figs. 4 and 5A). Inasmuch as bound peptides are located in the nearly enclosed binding channel of GID4, and since the first (Nt) residue of a peptide is deeply buried in the channel’s narrow end, the Pro residue is preferred, affinity-wise, as the Nt-residue of a GID4 substrate. But the binding cavity is still large enough to accommodate some other hydrophobic residues, such as Ile or Val. In this cavity of GID4, the side chain of a peptide’s Nt-residue is surrounded by the hydrophobic residues L159, V141, I161, L164, L171, I249, L240, and F254 (Figs. 4 and 5A).

Use of GID4 Mutants to Analyze Binding Specificity of GID4.

We used site-directed mutagenesis to convert to Ala residues – specific bulky hydrophobic residues of GID4 that surround the Nt-Ile residue of the GID4-bound IGLWKS peptide. These single-residue mutations caused a strong reduction of binding of the IGLWKS peptide to a mutant GID4, or even the loss of detectable binding (Fig. 5B and SI Appendix, Table S2).

Interestingly, an uncharged hydrophilic residue of GID4, Thr-173, is located near the hydrophobic pocket that binds to a hydrophobic Nt-residue of a substrate (Fig. 5A). Because of its spatial proximity to the hydrophobic pocket, Thr-173 might interfere with the binding of Nt-Pro or other substrates’ hydrophobic Nt-residues to wild-type GID4. To address this possibility, Thr-173 of GID4 was replaced with single hydrophobic residues, such as Ala, Val, Leu, and Ile. In agreement with the above prediction, all of these position-173 mutants bound to the IGLWKS peptide with a higher affinity than wild-type GID4. For example, the GID4^T173A mutant bound to IGLWKS fourfold more tightly than did wild-type GID4 (Fig. 5C and SI Appendix, Table S3). The results were similar with the binding of position-173 GID4 mutants to the otherwise identical but Nt-Pro–bearing PGLWKS, except that the strongest binder was the GID4^T173V mutant, in contrast to GID4^T173A vis-à-vis IGLWKS (Fig. 5 C and D). On the assumption (to be verified) that the increased affinity exhibited by some GID4 derivatives altered at position-173 is not accompanied by a lowered specificity of binding, such mutants may prove particularly helpful in screens for physiological substrates of the Pro/N-degron pathway.

GID4-Dependent Degradation of a Protein Bearing the Nt-Ile Residue.

Previous work has shown that the S. cerevisiae and human GID4 N-recognins are not only sequelogous [similar in sequence (111)] and contain the above-described hydrophobic binding chamber (Figs. 4 and 5A), but are also similar in their recognition specificities vis-à-vis test proteins that bear, in particular, Nt-Pro (43, 44, 47, 48). We asked, therefore, whether S. cerevisiae GID4 may also, similarly to human GID4, recognize a non-Pro Nt-residue of a protein and in addition target that protein for the in vivo degradation by the Pro/N-degron pathway.

Protein degradation was assayed in yeast using the promoter reference technique (PRT) (74). In this method, both a protein of interest and a long-lived reference protein (dihydrofolate reductase, DHFR) are expressed in S. cerevisiae from the same plasmid and identical constitutive promoters that contain additional (PRT-specific) DNA elements. Once transcribed, these elements form specific 5′-RNA aptamers that can bind to the added (cell-penetrating) tetracycline (Tc). As a result, Tc, which does not inhibit global translation in the cytosol, can selectively repress translation of the aptamer-containing mRNAs that encode both the reference and test proteins (74). Advantages of PRT include the accuracy-increasing reference protein and the avoidance of cytotoxic, artifact-prone global translation inhibitors, such as cycloheximide in chase-degradation assays (74). Following the addition of Tc, a decrease in the amount of a test protein relative to the reference protein during a chase would signify degradation of the test protein.

We used PRT to ask whether yeast GID4 could target for degradation a test protein that comprises the C-terminally triple-flagged S. cerevisiae FBP1_3f gluconeogenic enzyme in which the Nt-Pro–bearing (and targetable by GID4) Nt-sequence PTLV (43) was replaced by Nt-IGLW. The latter has been identified as the Nt-sequence that could bind to human GID4 in vitro with a significant affinity (K_d of 16 μM) (Fig. 3 and SI Appendix, Fig. S1B). That affinity was, however, not as high as the affinity of the otherwise identical Nt-Pro–bearing Nt-sequence PGLW (K_d of 1.9 μM) (Fig. 3 and SI Appendix, Fig. S1A).

The initially present Nt-Met residue of a test protein is cotranslationally cleaved off by ribosome-associated MetAPs if a residue at position 2 (it would become N terminal after cleavage) is not larger than Val (2, 112). For example, the Pro residue at position 2 does not impede the MetAP-mediated removal of the initially present Nt-Met, a step that yields Nt-Pro (2, 112). In contrast, Nt-Ile, the Nt-residue of the human GID4-binding IGLW peptide, is larger than Val, and therefore would not become N terminal if Ile follows Nt-Met in a nascent polypeptide. The Ub fusion technique bypasses this problem by making it possible to expose, in vivo, any desired residue (except Pro) at the mature N terminus of a protein of interest (113). In this method, a protein is expressed as a linear Ub-X–protein fusion. The fusion’s Ub moiety is cotranslationally cleaved off by a family of deubiquitylating enzymes (present in all eukaryotes), yielding X-protein that bears the desired Nt-X residue (113).

S. cerevisiae cells carrying a PRT-based plasmid that expressed the 187-residue DHFR reference protein (tagged with both flag and ha epitopes) and a test protein—either Ub-MPTLV-FBP1_3f, Ub-MSTLV-FBP1_3f, Ub-IGLW-FBP1_3f, or Ub-MPGLW-FBP1_3f (43)—were grown in a glucose-containing minimal medium. In this setting, PTLV becomes the Nt-sequence of wild-type FBP1 after the cotranslational removal of its Nt-Met by MetAPs. Because of the removal of Nt-Met, both the MPTLV-FBP1_3f and MSTLV-FBP1_3f moieties of the above Ub fusions were present in vivo as the short-lived wild-type PTLV-FBP1_3f (derived from MPTLV-FBP1_3f) and the otherwise identical but long-lived STLV-FBP1_3f (derived from MSTLV-FBP1_3f) (43). The Nt-Ser residue is not recognized by GID4, in contrast to Nt-Pro (43, 44). The above Ub fusion-based versions of MPTLV-FBP1_3f and MSTLV-FBP1_3f served as controls, to verify the suitability of using the Ub fusion technique (113) to assay GID4 substrates for degradation by the Pro/N-degron pathway.

The in vivo translation of both a test protein and the DHFR reference was selectively repressed by the addition of Tc, thereby initiating a chase (74). As expected, PTLV-FBP1_3f (but not STLV-FBP1_3f) was short-lived in vivo. Degradation of PTLV-FBP1_3f required GID4, as no degradation was observed in gid4Δ cells (Fig. 6 C and D). Ub-MPGLW-FBP1_3f (becoming PGLW-FBP1_3f in vivo), whose Nt-Pro–bearing sequence PGLW could bind to GID4 in vitro (Figs. 2 and 3 and SI Appendix, Fig. S1A), was also short-lived in vivo, in a GID4-dependent manner (Fig. 6 A and B).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

A protein bearing the GID4-binding, Prolacking N-terminal sequence Nt-IGLW can be targeted by the S. cerevisiae Pro/N-degron pathway for GID4-dependent degradation. The bands of test and reference (DHFR) proteins are indicated on the left. (A) Lane 1, kilodalton marker. Lanes 2 to 5, Tc/PRT-initiated chase assay (Results) with Ub-MPGLW-Fbp1_f3 in wild-type S. cerevisiae. Lanes 6 to 9, same as in lanes 2 to 5, but with gid4Δ cells. Lanes 10 to 13, same as in lanes 2 to 5, but with Ub-IGLW-Fbp1_f3. Lanes 14 to 17, same as in lanes 6 to 9 (gid4Δ cells) but with Ub-IGLW-Fbp1_f3. (B) Quantification of data in A. Identities of specific curves are indicated below lane numbers. (C) Lanes 1 to 4, Tc/PRT-chase assay with Ub-MPTLV-Fbp1_f3 in wild-type S. cerevisiae. Lanes 5 to 8, same as in lanes 1 to 4 but with gid4Δ cells. Lanes 9 to 12, same as lanes 1 to 4 but with Ub-MSTLV-Fbp1_f3. Lanes 13 to 16, same as lanes 5 to 8 (gid4Δ cells), but with Ub-MSTLV-Fbp1_f3. (D) Quantification of data in C. Identities of specific curves are indicated below lane numbers. All degradation assays were performed at least twice, yielding results that differed by less than 10%.

The IGLW-FBP1_3f protein (it became IGLW-FBP1_3f after the in vivo removal of Nt-Ub by DUBs), whose Nt-Ile–bearing Nt-sequence IGLW could bind to GID4 in vitro (Fig. 3 and SI Appendix, Fig. S1 B and G), was also short-lived in vivo (Fig. 6 A and B). The degradation of both PGLW-FBP1_3f and IGLW-FBP1_3f required GID4, as this degradation did not occur in gid4Δ cells (Fig. 6 A and B). Thus, the Pro-lacking Nt-sequence IGLW not only binds to GID4 but can also confer, on a test protein, a GID4-dependent degradation in S. cerevisiae (Figs. 3 and 6 A and B).

Interestingly, the GID4-dependent degradation of IGLW-FBP1_3f not only occurred but was even faster than the degradation of PGLW-FBP1_3f, although the in vitro affinity of human GID4 for the Nt-sequence IGLW (K_d of 16 μM) is significantly lower than the affinity of GID4 for the otherwise identical but Nt-Pro–bearing sequence PGLW (K_d of 1.9 μM) (Figs. 3 and 6 A and B and SI Appendix, Fig. S1 A, B, and G). Moreover, the early, prechase [“time-zero” (2, 35)] degradation of IGLW-FBP1_3f was also greater than the analogous degradation of PGLW-FBP1_3f (Fig. 6 A and B). Detailed reasons for the in vivo degradation of PGLW-FBP1_3f being slower than that of IGLW-FBP1_3f, despite the converse order of their in vitro affinities for GID4 remain to be understood. A plausible interpretation of these findings is as follows. Only one cotranslational in vivo cleavage (removal of the Nt-Ub moiety by DUBs) is required to produce the GID4-targetable IGLW-FBP1_3f protein from Ub-IGLW-FBP1_3f. In contrast, two obligatorily sequential cleavage steps, the removal of the Nt-Ub moiety and the removal of resulting Nt-Met by MetAPs, are required to produce the GID4-targetable PGLW-FBP1_3f protein from Ub-MPGLW-FBP1_3f. The resulting additional delay in generating PGLW-FBP1_3f (in contrast to no such delay in generating IGLW-FBP1_3f) may lead to a slower or incomplete removal of Nt-Met from MPGLW-FBP1_3f, the immediate precursor of PGLW-FBP1_3f.

In vivo levels of ribosome-associated MetAPs are significantly higher in the immediate vicinities of ribosomes than in the bulk solution. It is likely, therefore, that the two-step production of PGLW-FBP1_3f from the Ub-MPGLW-FBP1_3f is significantly less efficacious than the one-step (and MetAP-independent) production of IGLW-FBP1_3f from the Ub-IGLW-FBP1_3f. One likely reason for this inefficiency is that the N terminus of a nascent (being made) protein keeps moving away from a translating ribosome into bulk-solution regions that contain lower levels of MetAPs.

This model, to be addressed in future studies, predicts that the GID4-dependent degradation of PGLW-FBP1_3f that is produced, in one step, from MPGLW-FBP1_3f (Nt-Met can be cotranslationally removed if position 2 is occupied by Pro) would be faster than the degradation of the otherwise identical PGLW-FBP1_3f that is produced from Ub-MPGLW-FBP1_3f through the above two-step reaction. The degradation of PGLW-FBP1_3f that is produced through a single cleavage step is also predicted to be faster than the degradation of IGLW-FBP1_3f (which is produced from Ub-IGLW-FBP1_3f by DUBs), because of the difference in binding affinities of GID4 for Nt-PGLW vs. Nt-IGLW (Fig. 3 and SI Appendix, Fig. S1 A, B, and G).

Concluding Remarks.

We have shown that in the presence of cognate downstream motifs, human GID4, the main recognin of the Pro/N-degron pathway, can recognize not only the Nt-Pro residue but a few other hydrophobic Nt-residues as well (Figs. 1–3 and SI Appendix, Fig. S1). We solved crystal structures of human GID4 bound to peptides that bear either Nt-Ile or Nt-Val (Figs. 4 and 5A). We also used site-directed mutagenesis of GID4 to analyze the roles of GID4 residues in the binding of substrates by this N-recognin (Fig. 5 and SI Appendix, Tables S2 and S3). These approaches determined contributions of specific GID4 residues to the GID4-mediated recognition of the Nt-Pro residue versus hydrophobic residues other than Pro. We also found that the S. cerevisiae GID4 N-recognin can target a protein bearing an Nt-sequence, such as Nt-IGLW (which lacks both Nt-Pro and a Pro at position 2) for rapid degradation by the Pro/N-degron pathway in vivo (Fig. 6).

The Nt-Ile residue of the sequence such as Nt-IGLW can be recognized as a destabilizing one not only by GID4 of the Pro/N-degron pathway but also, for example, by S. cerevisiae UBR1, the N-recognin of the yeast Arg/N-degron pathway, except that Nt-Ile binds to UBR1 less strongly than other UBR1-recognized destabilizing Nt-residues (2). With that caveat, it is possible that some proteins (or protein fragments, since Nt-Ile would not be normally exposed at a nascent protein’s N terminus; see above) might be targetable, independently, by both the Pro/N-degron and Arg/N-degron pathways.

The known physiological substrates of the S. cerevisiae Pro/N-degron pathway are the gluconeogenic enzymes FBP1, ICL1, MDH2, and PCK1. All four of these proteins are conditionally polyubiquitylated and destroyed by the S. cerevisiae GID Ub ligase and the proteasome (Introduction). The GID ligase targets FBP1, ICL1, MDH2, and PCK1 via GID4, the main N-recognin of GID that binds to Nt-Pro residues (in FBP1, ICL1, and MDH2) or to a Pro at position 2 (in PCK1), in the also required presence of adjoining motifs (2, 43⇓–45, 48, 96).

The S. cerevisiae genome encodes about 300 proteins that are expected to bear Nt-Pro (43). However, at least a majority of these proteins do not appear to be substrates of GID4 N-recognin, in contrast to FBP1, ICL1, and MDH2 (43, 47, 48). In addition, it remains to be determined whether there are physiological substrates of S. cerevisiae GID4 other than PCK1 that are targeted, at least in part, through their Pro at position 2, analogously to PCK1.

In vertebrates, including mammals, the GID (CTLH) Ub ligase functions in primary cilia, in cell proliferation, and in other processes, including apparently, glucose sensing and energy homeostasis (refs. 96 and 102⇓⇓⇓⇓⇓⇓⇓–110 and references therein). While the Pro residue at position 2 is strongly conserved among both fungal and vertebrate PCK1 enzymes, the Nt-Pro residues of S. cerevisiae FBP1, ICL1, and MDH2 are weakly conserved even among fungi, and are absent, for example, from mammalian counterparts of these enzymes. In addition, the mammalian versions of S. cerevisiae FBP1, ICL1, and MDH2 are apparently not targeted for degradation by the GID (CTLH) Ub ligase (refs. 96 and 102⇓⇓⇓⇓⇓⇓⇓–110 and references. therein).

In sum, little is known about physiological substrates of S. cerevisiae GID4 outside the set of yeast gluconeogenic enzymes (Fig. 1) (43, 47, 48). As to human GID4, its Nt-Pro–bearing physiological substrates remain to be identified (44). It is also unknown whether some non-GID4 main subunits of the GID Ub ligase (Introduction) might also function as its targeting components (of currently unknown specificity), in addition to and independently of GID4.

One ramification of results in the present study is that they will facilitate discovery of physiological substrates of the Pro/N-degron pathway, since potential GID4 substrates are no longer confined to proteins that bear Nt-Pro or a Pro at position 2. In addition, our findings deepened the previously revealed similarity of target recognition by the S. cerevisiae and human GID4 N-recognins by expanding their substrate range from interactions with Nt-Pro or a Pro at position 2 to their demonstrated interactions, in a functionally relevant manner, with specific Nt-residues other than Pro (Figs. 1–6).