Fig 2: HERC2 recognizes chaperonin subunits using the adaptor ZNRD2(A) Cells treated with siRNAs against HERC2 or ZNRD2 were transfected with CCT4-3xFLAG, followed by non-denaturing IP using anti-FLAG affinity resin. Input and IP samples were analyzed by immunoblotting for the indicated proteins.(B) Cytosolic extracts from cells treated with siRNAs targeting HERC2, ZNRD2, or control were fractionated by sucrose gradient centrifugation to quantify the proportion of unassembled CCT4 and CCT7 subunits (see Figures S4A and S4B). Data are mean ± SD of data points (dots) from three independent experiments (color-coded). **p < 0.01 and ***p < 0.001 vs. control siRNA by one-way ANOVA with Dunn’s multiple comparisons test.(C) Cells treated with the indicated siRNAs were transfected with the dual-color CCT4 reporter and analyzed by immunoblotting (left) and flow cytometry (right).(D) Cells treated with siRNAs against ZNRD2 were transfected with the CCT4 reporter and siRNA-resistant ZNRD2, followed by immunoblotting (left) and flow cytometry (right).(E) Wild-type (WT) and ZNRD2-knockout (KO) Flp-In T-REx 293 cells were transfected with the dual-color CCT4 reporter and either 3xFLAG-HERC2, ZNRD2, or empty vector. Cells were analyzed by immunoblotting (top) and flow cytometry (bottom).(F) Diagram illustrating the domain structure of human HERC2 (4,834 amino acids) and overlapping HERC2 fragments tested for interaction with ZNRD2 in Figure S4G. HERC2 contains three RCC1-like domains (RLD1–3) and a C-terminal HECT E3 ligase catalytic domain (HECT). The HERC2 fragment containing RLD3 (F5; highlighted in red) interacts with ZNRD2 (see Figure S4G).(G) Cells treated with HERC2-targeting siRNA were transfected with the CCT4 reporter together with empty vector, siRNA-resistant 3xFLAG-HERC2 (WT), or its deletion mutant lacking RLD3 (?RLD3). Cells were analyzed by immunoblotting (left) and flow cytometry (right).See also Figure S4.

Fig 3: Recognition of unassembled chaperonin subunits via the adaptor ZNRD2, related to Figure 3(A and B) Cytosolic extracts (input) from HEK293T cells treated for 5 days with the indicated siRNAs were separated into 11 fractions on a 5%–45% sucrose gradient. Input and gradient fractions were analyzed by immunoblotting for CCT4 (A) and CCT7 (B), and the signal intensities of the bands in each fraction were measured by densitometry. The graphs below the blots depict the distributions of CCT4 and CCT7 across the gradient. The proportion of unassembled CCT4 and CCT7 migrating in fractions 2–4 was used for the quantification shown in Figure 3B. Assembled CCT migrates primarily in fractions 7–9.(C) Cells treated with control or ZNRD2-targeting siRNAs were transfected with the CCT4 reporter and analyzed by immunoblotting (left) and flow cytometry (right).(D) WT Flp-In T-REx 293 cells and ZNRD2-KO cells with stably integrated empty vector, untagged ZNRD2, or ZNRD2-3xFLAG were transfected with the CCT4 reporter. Cells were analyzed by flow cytometry (right), and total cell lysates were analyzed by immunoblotting (left).(E) HEK293T cells were treated with control or ZNRD2-targeting siRNAs, transfected with the CCT4 reporter and either empty vector or 3xFLAG-HERC2, and analyzed by flow cytometry.(F) HERC2-KO Flp-In T-REx 293 cells were treated with control or ZNRD2-targeting siRNAs, transfected with the CCT4 reporter and either empty vector or 3xFLAG-HERC2, and analyzed by flow cytometry.(G) The indicated HERC2 fragments with an N-terminal 3xFLAG tag (F1–F6) were transiently expressed in HEK293T cells and immunoprecipitated under non-denaturing conditions with anti-FLAG affinity resin (IP). Input (left) and IP samples (right) were analyzed by immunoblotting.(H) A dual-color NCOA4 reporter (as in Figure S1A) was assessed for degradation by flow cytometry in HEK293T cells knocked down for HERC2 or ZNRD2. NCOA4 degradation, which is normally very efficient as indicated by the low GFP:RFP ratio, was inhibited by HERC2 knockdown but not ZNRD2 knockdown.(I) HEK293T cells treated with control or HERC2-targeting siRNAs were transfected with the NCOA4 reporter together with empty vector or the indicated 3xFLAG-HERC2 constructs. In HERC2-knockdown cells, re-expression of HERC2 WT and ?RLD3, but not catalytically inactive C4762S mutant, restored NCOA4 degradation.

Fig 4: Physiological relevance of the HERC2-ZNRD2 pathway(A) ZNRD2-KO Flp-In T-REx 293 cells were complemented with either empty vector or the indicated ZNRD2 constructs at the unique FRT site in their genome. These cells were transfected with the CCT4 reporter and analyzed by immunoblotting (left) and flow cytometry (right). Wild-type (WT) Flp-In T-REx 293 cells were analyzed in parallel for comparison.(B) The exogenous ZNRD2 sequence at the FRT site was PCR-amplified from the genomic DNA of ZNRD2-KO cells complemented with ZNRD2 WT or its mutants (K82A or L42E), and the amplicons were sequenced. Shown are sequencing chromatograms for the WT and mutant cells at the regions around the K82 and K42 codons.(C) ZNRD2-KO cells expressing ZRND2 WT were mixed at a ratio of 1:1 with ZNRD2-KO cells expressing either the K82A mutant (left; WT vs. K82A) or the L42E mutant (right; WT vs. L42E) and then cultured for 4 weeks. An aliquot of each co-culture was sampled at the indicated time points, and the exogenous ZNRD2 sequences at the FRT site were amplified and sequenced as in (B). Sequencing chromatograms around the K82 and L42 codons are shown for the co-cultures containing the respective mutant cell lines.(D) Interface between ZNRD2 and RLD3 of HERC2 in the predicted structural model. RLD3 is shown in gray surface representation, while ZNRD2 is in blue cartoon representation. The close-up view highlights the E3994 residue (red) of HERC2.(E) HEK293T cells transiently expressing the indicated 3xFLAG-HERC2 constructs were subjected to anti-FLAG IP under non-denaturing conditions. Input and IP samples were analyzed by immunoblotting for the indicated proteins. The asterisk indicates a cross-reactive nonspecific band.(F) HERC2-KO cells were transfected with the CCT4 reporter (left) or the NCOA4 reporter (right), together with the indicated plasmids, and analyzed by flow cytometry.(G) Model for quality control (QC) during CCT biogenesis. Depending on the step at which biogenesis fails, qualitatively different types of QC act on the substrate. This includes ribosome-associated quality control (RQC) if CCT translation fails, protein QC if a CCT subunit fails to fold after translation, and assembly QC for several CCT subunits that fold but do not assemble into the CCT. Unassembled CCT subunits can be recognized by ZNRD2, which engages the RLD3 domain of the E3 ligase HERC2, and ubiquitinates the subunit(s) to target it for degradation. ZNRD2 can also recognize partially assembled complexes, but the triage of these complexes between degradation and assembly remains to be investigated. ZNRD2 cannot engage assembled CCT because its interaction surfaces are buried.

Fig 5: AlphaFold-predicted model of a putative CCT4-ZNRD2-HERC2 complex, related to Figure 5(A and B) AlphaFold-predicted model (A) and predicted alignment error (PAE) plot (B) of a putative ternary complex consisting of CCT4, ZNRD2, and HERC2-RLD3. The predicted structural model is colored by per-residue confidence score (predicted local distance difference test [pLDDT]).(C) Comparison of CCT4 in the predicted CCT4-ZNRD2-RLD3 complex (blue) with its experimentally determined structure (pink; PDB: 7lum [chain F]) as well as with the model of CCT4 predicted by AlphaFold2 in isolation (yellow).(D) Comparison of RLD3 of HERC2 in the predicted CCT4-ZNRD2-RLD3 complex (blue) with its experimentally determined structure (pink; PDB: 3kci).(E) Comparison of ZNRD2 in the predicted CCT4-ZNRD2-RLD3 complex (blue) with experimentally determined structure of an N-terminal region of ZNRD2 (pink; PDB: 6hcz) as well as with the model of ZNRD2 predicted by AlphaFold2 in isolation (yellow).(F) Surface representation of individual CCT subunits obtained from an experimentally determined structure of the CCT complex (PDB: 7lum), colored by hydrophobicity. Red dashed circles indicate a hydrophobic patch shared by all CCT subunits. When engaged with ZNRD2, this hydrophobic patch is predicted to be occupied by four highly conserved residues of ZNRD2 (see Figure 5B).(G) Predicted interaction between the W13 residue of ZNRD2 and CCT4. CCT4 is shown as surface colored by hydrophobicity, while ZNRD2 is as cartoon colored by conservation. A close-up view highlighting the W13 residue of ZNRD2 that lies in a shallow hydrophobic patch of CCT4 is shown in the inset.