A mitochondrial insertion directs substrate selection and engagement by the protein unfoldase ClpX Authors: Adam DeCosta, Rebecca Barrick, Julia R Kardon Published: 2026 ABSTRACT AbstractThe force-generating ATPases Associated with diverse cellular Activities (AAA+) ATPase domain of protein unfoldases is specified for many substrates and other functional partners through elaboration with accessory domains. Mitochondrial homologs of the unfoldase ClpX contain an insertion within the AAA+ domain that is absent in bacterial homologs. This mitochondrial insertion (MI) maps to the substrate-encountering face of ClpX, leading us to hypothesize that the MI directs interactions of ClpX with mitochondrial substrates, the best-characterized of which is the first enzyme in heme biosynthesis, 5-aminolevulenic acid synthase (ALAS). We find that the MI is critical for both recruitment and activation of ALAS by S. cerevisiae ClpX. The MI was dispensable for heme-induced, adaptor-mediated degradation of ALAS by the human CLPX-CLPP protease complex (CLPXP) but contributed to adaptor-independent recruitment of the model substrate casein for degradation. Although truncation of the MI moderately perturbed ATPase activity in both yeast and human ClpX, this effect could be uncoupled from the requirement for the MI in the efficiency of ALAS activation by targeted mutagenesis. The MI therefore can serve both to recruit a substrate to mitochondrial ClpX and to accelerate its processing by the AAA+ motor.Keywords: ATPase Associated with diverse cellular activities (AAA), mitochondria, heme, enzyme kinetics, adaptor protein, ATP-dependent protease, protein degradation AAA+ family unfoldases provide essential control of protein biogenesis, repair, regulation, and quality control (1). These enzymes share a highly conserved AAA+ ATPase domain that grips protein substrates to apply the mechanical force that drives protein unfolding (2). This core mechanism must be controlled and directed for specific substrates and functions. Intensive study of several bacterial AAA+ unfoldases has demonstrated that like several other well-studied members of the larger AAA+ family, accessory domains are responsible for directing the AAA+ domain to specific functions. For AAA+ unfoldases, these accessory domains directly mediate specific interactions with substrates or with adaptor proteins and other functional partners (3). Mitochondrial AAA+ unfoldases are closely related to those of their bacterial relatives, but in many cases, these accessory domains have diverged widely between bacterial and mitochondrial homologs, for which substrate repertoires and recruitment strategies are much less characterized. One example of such divergence is the AAA+ unfoldase ClpX, which is widely conserved in bacteria and as a mitochondrial protein in eukaryotes. In both bacteria and mitochondria, ClpX can act to regulate substrates on its own by remodeling them (4, 5) and by unfolding coupled to degradation in complex with its partner peptidase ClpP (2, 6). ClpX is essential for viability in many bacteria (7), its deletion is early-embryonic lethal in mice (8, 9), and mutations in ClpX or ClpP underly two mitochondrial syndromes (10, 11), underscoring the broad importance of the regulation of bacterial and mitochondrial proteomes by ClpX. Proteomic, cell biological, and biochemical data suggest that the substrate repertoire of mitochondrial ClpX homologs has diverged widely from known bacterial substrates (12, 13, 14). A broadly conserved function of bacterial ClpX, the degradation of stalled translation products directed by the ssrA tag and the adaptor protein SspB, appears to have been lost in the evolution of mitochondria, which lack genes encoding SspB, and with few exceptions, the ssrA tag (15). Although both bacterial and mitochondrial ClpX homologs recruit adaptor proteins by a related N-terminal zinc-finger domain, the single demonstrated adaptor protein for mitochondrial ClpX (16) appears to be restricted to metazoans, and no bacterial adaptors appear conserved as mitochondrial proteins. Mitochondrial ClpX homologs also contain an element of ∼30 to 70 amino acids that is absent in bacterial homologs (Fig. 1). This mitochondrial insertion (MI) is inserted between two helices of the core AAA+ domain, emerging from the substrate-engaging face of the ClpX hexamer (Fig. 1, A–C). We therefore hypothesized that the MI could contribute to substrate recruitment by mitochondrial ClpX. The MI has been reported to impair oligomerization, ATPase activity, and degradation of the model substrate casein by human CLPXP, although these effects were observed with a truncation of CLPX that removed elements beyond the C-terminal end of the MI as defined here, extending into regions conserved among all ClpX homologs (Fig. 1, A and D) (17). Mitochondrial ClpX homologs contain an element inserted within the AAA+ domain.A, domain diagrams of bacterial (E. coli) and mitochondrial (S. cerevisiae and H. sapiens) ClpX homologs. Zinc-finger N-domain is displayed in orange, mitochondrial insertion in blue, and AAA+ ATPase domain in gray. B, a model of a hexamer of S. cerevisiae Mcx1 generated with AlphaFold3. The AAA+ domain (pLDDT >70 is depicted as a gray surface and ribbon diagrams of the mitochondrial insertion and N-terminal element are colored by AlphaFold3 pLDDT backbone scores (very high, pLDDT >90; confident, 90 > pLDDT >70; low, 70 > pLDDT >50; very low, pLDDT < 50). C, a model of a hexamer of human CLPX generated with AlphaFold3 and displayed in the same orientation (from structural alignment) as the model of Mcx1 in (B). The AAA+ domain and zinc-finger N-domain are depicted as light and dark gray surfaces, respectively, and were modeled at pLDDT >70. The ribbon diagram of the mitochondrial insertion is colored by AlphaFold3 pLDDT score as in (B). Very-low-confidence coil at the extreme C-terminus of Mcx1 and N- and C-terminus of CLPX is not displayed. D, Multiple sequence alignment of bacterial and mitochondrial ClpX homologs in the region of the mitochondrial insertion. Alignment was generated and colored using ClustalOmega with standard settings. Region displayed corresponds to residues 69 to 134 and 199 to 287 in S. cerevisiae and H. sapiens ClpX homologs, respectively. H. sapiens and S. cerevisiae homologs, the focus of this study, are highlighted with boxed species names and the mitochondrial insertion sequences truncated in this study are highlighted in blue (S. cerevisiae, 81-127; H. sapiens, 210-280). The best-characterized substrate of mitochondrial ClpX is the first enzyme in heme biosynthesis, ALAS. In S. cerevisiae, ClpX activates ALAS by ATP-driven partial unfolding to accelerate incorporation or replacement of its cofactor, pyridoxal phosphate (PLP) (5, 12). In metazoans, heme also induces negative-feedback degradation of ALAS by ClpXP (11, 16, 18). The adaptor protein polymerase delta-interacting protein 2 (POLDIP2) recognizes a heme-bound motif in ALAS and induces degradation of ALAS by ClpXP (16). Although this appears to be the dominant activity of ClpX toward ALAS in humans, this negative feedback mechanism is absent in S. cerevisiae, which lacks both POLDIP2 (which is restricted to metazoans) and ClpP. Here, we establish that the MI is critical for activation of ALAS by S. cerevisiae ClpX. The MI is not required for adaptor-mediated degradation of ALAS by human ClpX, but modulates adaptor-independent recognition of the model substrate casein. Our kinetic analysis of ALAS activation indicates that the MI contributes both to recruitment of ALAS and, unexpectedly, to the maximal rate of activation. These data indicate that the MI can direct the activity of ClpX both through substrate selection and by contributing to efficient engagement of substrate by the core AAA+ unfoldase. Results The mitochondrial insertion (MI) is a ubiquitous, small (∼30-70 amino acids) sequence insertion among mitochondrial ClpX homologs, although its sequence diverges substantially across the eukaryotic phylogeny (Fig. 1D). The MI is predicted to project from the face of the mitochondrial ClpX hexamer at which substrates are engaged before translocation through the central pore, suggesting that it could contribute to substrate recruitment (Fig. 1, B and C). In the yeast ClpX homolog (hereafter referred to by its standard name, Mcx1), the MI is predicted by AlphaFold3 with low-to-moderate confidence to form a helix-loop-helix structure (Fig. 1B). An α helix provided by the small element N-terminal to the AAA+ domain is modeled to bundle with the MI, although with low confidence (PAE > 12 Å). The MI of the human ClpX homolog (hereafter referred to by its standard name, CLPX) is predicted with very low confidence as a mixture of helix and coil; although depicted as packing against the N-domain of CLPX, there is no confidence in this conformation (PAE > 25 Å) (Figs. 1C and S3, A and B). To investigate the function of the mitochondrial insertion (MI), we first tested whether it contributed to activation of S. cerevisiae ALAS. We initially tested this by monitoring the steady-state levels of the product of ALAS, 5-aminolevulinic acid (ALA), in S. cerevisiae. ALA levels are reduced in cells lacking Mcx1, due to reduced ALAS activity (5). We generated a strain of S. cerevisiae in which we replaced the MI with a 3xGly-Ser linker at the genomic locus of Mcx1 (mcx1ΔMI, 3xGly-Ser replacing residues 81-127 of Mcx1) and then compared ALA levels in this strain to wild-type and mcx1Δ yeast. We found that truncation of the MI caused a reduction in cellular ALA equivalent to that observed in mcx1Δ cells (Fig. 2A). We observed no reduction in Mcx1 protein levels when the MI was deleted (Fig. S1A), demonstrating that the reduction in ALA we observed in mcx1ΔMI cells must derive from loss of Mcx1 activity. The mitochondrial insertion is critical to support heme biosynthesis through activation of ALAS.A, Steady state ALA levels in S. cerevisiae cell extracts, quantified using modified Ehrlich’s reagent (n = 4, error bars indicate SD). ∗p = 0.015 for mcx1ΔMI, ∗∗p = 0.012 for Δmcx1 using Welch’s t test for comparison with wild type. B, ATP hydrolysis by 0.5 μM Mcx1 hexamer plotted as a function of ATP concentration and fit with an allosteric sigmoidal equation (see Experimental Procedures) (n = 2, all data points displayed). C, ATP hydrolysis by Mcx1 with 2 mM ATP, plotted as a function of Mcx1 hexamer concentration and fit with an allosteric-sigmoidal equation (n = 2, all data points displayed). D, Mcx1-stimulated PLP binding to ALAS monitored by fluorescence (ex. 434 nm., em. 515 nm). Data are fit with the Michaelis-Menten equation (n = 3, all data points displayed). E and F, Kinetic parameters extracted from curve fits in panels B and C (E) and panel F (D) with standard error of the fit. All n values represent independent biological replicates. Having established that Mcx1 requires the MI to stimulate ALA production in vivo, we sought to determine the biochemical mechanism underlying this requirement. We first tested basal, substrate-independent activities of Mcx1: ATP hydrolysis and oligomerization, using recombinantly expressed and purified Mcx1ΔMI. We first monitored ATPase activity using an NADH-coupled assay. We found no change in the KM for ATP compared to Mcx1, but observed a twofold reduction in kcat, indicating that the MI does not contribute to ATP binding but that its loss reduces the maximal catalytic rate (Fig. 2, B and E). Because the ATPase active site of ClpX homologs resides at the subunit interface of the hexameric assembly, with important and conserved contacts with ATP provided from both sides of the interface, ClpX homologs must oligomerize to form a functional ATPase active site (19, 20). We therefore considered whether the reduced kcat for ATP we observed for Mcx1ΔMI might derive from impaired oligomerization. To assess oligomerization of Mcx1, we monitored ATPase activity across a concentration range of Mcx1 or Mcx1ΔMI. We observed equivalent K1/2 values for Mcx1 and Mcx1ΔMI (Fig. 2, C and E). Mutation of the conserved Walker A motif of Mcx1 abolished ATPase activity, confirming the specificity of this assay (Fig. S1B). We also analyzed Mcx1 and Mcx1ΔMI by native gel and observed similar amounts of a species whose migration was consistent with a hexamer, although under these conditions the majority of both proteins formed an apparent dimer (Fig. S1C). From these data, we infer that the MI is not important for Mcx1 oligomerization (likely hexamerization) but supports its maximal ATP hydrolysis rate. Although the MI is not predicted to contact the active site of Mcx1 (Fig. 1B), it might affect the dynamics of the large subdomain of the AAA+ domain to which it connects, in a manner that alters catalytic rate. We next tested how the MI contributes to Mcx1 activation of ALAS in vitro. Mcx1 activates ALAS for catalysis by partially unfolding the enzyme, accelerating the rate of incorporation of the cofactor of ALAS, pyridoxal phosphate (PLP) (12). We monitored Mcx1-stimulated PLP incorporation into ALAS over a range of ALAS concentrations, detecting ALAS-PLP by the fluorescence of the pyridoxyllysine bond PLP forms at the active site. We observed that truncation of the MI caused a four-fold reduction in kcat and a seven-fold increase in KM for stimulation of PLP binding by Mcx1, causing a large decrease in catalytic efficiency (Fig. 2, D and F). From the substantial increase in KM, we infer that the MI is important for recruitment of ALAS. The reduced kcat for ALAS activation could derive from the reduced ATPase activity we observe for Mcx1ΔMI, or it could indicate that the MI contributes to how Mcx1 engages ALAS for unfolding. To isolate features of the MI that are important for its function and determine if the effect of MI truncation on ATPase rate and substrate interactions are separable, we examined its sequence conservation and its predicted structure for coherent features. Although the MI diverges widely in sequence and approximately twofold in length across eukaryotes (Fig. 1D), it exhibits some sequence conservation across shorter evolutionary distances (Fig. 3A). Comparing the MI of Mcx1 with other ClpX homologs in ascomycota, two conserved features stand out (Fig. 3A). First, the loop linking two predicted alpha helices is highly conserved, with Phe101 showing nearly complete conservation among ascomycota. Arg84, which maps to the base of the predicted helix-loop-helix (Fig. 3B), also shows broad conservation. To test how these features contribute to Mcx1 ATPase activity and activation of ALAS, we analyzed two Mcx1 variants: Mcx1loop, in which all residues in the apical loop (Glu98-Ser105) are mutated to alanine, and Mcx1R84A. The kcat and KM values of Mcx1loop and Mcx1R84A, as well as their migration on a native gel, were equivalent to that of the wild-type enzyme, in contrast to the decreased kcat we observed for Mcx1ΔMI (Figs. 3, C and E and S1, C and D). Both Mcx1loop and Mcx1R84A, however, were impaired in their ability to activate ALAS. Mcx1loop exhibited a fourfold decrease in kcat and fivefold increase in KM for ALAS, similar to the impairment in ALAS activation observed for Mcx1ΔMI. Mcx1R84A exhibited a smaller decrease in kcat and increase in KM values for ALAS, causing a three-fold change in the specificity constant (0.16 min−1 μM−1 for Mcx1; 0.055 min−1 μM−1 for Mcx1R84A) (Fig. 3, D and E). These mutations thus uncouple the effects of the MI on ALAS activation from its effect on ATPase activity. This uncoupling demonstrates that the requirement for the MI in activation of ALAS is not caused by general impairment of the ATPase of Mcx1, but instead that the MI contributes to both ALAS recognition (thus affecting the KM for ALAS) and engagement/unfolding by the AAA+ motor (thus affecting kcat for ALAS). The apical loop of the mitochondrial insertion of Mcx1 mediates both substrate recruitment and engagement for unfolding.A, a central subsection of the mitochondrial insertion, drawn from a global alignment of ClpX sequences among ascomycota. The green bar indicates the conserved Arg84, and the orange bar indicates the apical MI loop residues. Region displayed corresponds to residues 81 to 110 in S. cerevisiae Mcx1. B, alphaFold3 model of Mcx1, in which the MI is displayed in blue ribbon with the apical loop colored orange and Arg84 depicted as green spheres. The N-terminal element is displayed as a gray ribbon and the AAA domain as gray surface. C, ATP hydrolysis rates of Mcx1 and variants (0.5 μM Mcx1 hexamer, 2 mM ATP, n = 3, error bars indicate SD). ∗∗∗∗p < 0.0001 for difference from wild-type rate in unpaired t test; other differences were not significant. D, Mcx1-stimulated PLP binding to ALAS, fit with the Michaelis-Menten equation. Parameters extracted from the fit and standard error of the fit are listed in (E). All n values represent independent biological replicates. We next sought to determine how the MI contributes to the function of human CLPX. The MI emerges from the same location of the AAA+ domain in all mitochondrial ClpX homologs, but its sequence and predicted structure vary considerably between yeast and human homologs (Fig. 1, B–D). We prepared a variant of the human CLPX protein analogous to Mcx1ΔMI, CLPXΔΜΙ (with a 3xGly-Ser linker replacing residues 210-280 of CLPX). We first monitored ATP hydrolysis while varying ATP or CLPX concentration to extract kinetic parameters for ATP hydrolysis and CLPX oligomerization. In contrast to the effect of MI truncation on Mcx1 ATP hydrolysis, we observed that the kcat for ATP hydrolysis is mildly stimulated upon MI truncation (Fig. 4, A and C). Human CLPX without the MI exhibited a similar KM for ATP and K1/2 for oligomerization (Fig. 4, B and C). Therefore, as for the yeast ClpX homolog, we conclude that the MI of CLPX does not contribute to formation of an intersubunit ATPase active site. Another group reported that truncation of the CLPX MI, defined as residues 230 to 304, or mutation of E285 decreased ATPase activity, impaired oligomerization, and nearly abolished degradation of casein by CLPXP (17). Position 304 lies well within a strongly conserved region of the AAA+ core domain (ending beyond the region displayed in Fig. 1D). E285 (Fig. 1D, third from last residue) is in a region of strong structural and moderate sequence conservation. The defects reported from this truncation may thus arise from removal of this more conserved C-terminal region or structural strain imposed by direct fusion of the flanking regions. The human CLPX MI is not required for degradation of ALAS2 but tunes recognition of casein.A, ATP hydrolysis by 0.5 μM CLPX hexamer plotted as a function of ATP concentration and fit with an allosteric-sigmoidal equation (n = 3, error bars indicate SD). B, ATP hydrolysis by CLPX with 2 mM ATP, plotted as a function of CLPX hexamer concentration, fit with an allosteric-sigmoidal equation (n = 2, all data points shown). C, Kinetic parameters extracted from curve fits in (A) and (B); error indicates standard error of fit. D, in vitro degradation of 10 μM ALAS2 with 0.5 μM CLPX hexamer, 0.5 μM CLPP 14-mer, 3 μM POLDIP2, 10 μM hemin, and 2 mM ATP. Lines represent linear fits to data, with error bars representing SD; n = 3. Rate is expressed in terms of ALAS2 molecules degraded per CLPX hexamer per minute. CLPX-FLAG was used in this experiment to sufficiently distinguish CLPXΔMI and ALAS2 sizes for analysis by SDS-PAGE. E, FITC-casein degradation rates by 0.5 μM CLPXP, plotted as a function of FITC-casein concentration and fit with Michaelis-Menten equation (n = 2, all data points shown). F, kinetic parameters extracted from curve fits in (E) with standard error of fit; p = 0.006 for difference between CLPX and CLPXΔMI by extra-sum-of-squares F test. All n values represent independent biological replicates. The moderate increase we observe in the maximal ATP hydrolysis rate of human CLPX after MI truncation was unexpected in comparison with the decreased rate we observed for Mcx1ΔMI. One clear difference between CLPX and Mcx1 is that Mcx1 has lost the zinc-finger N-domain that most ClpX homologs, including human CLPX, contain (Fig. 1, A–C). We considered whether this domain, which projects from the same face of the ClpX hexamer as the MI, might alter the effect of MI truncation. To test this, we measured the maximal ATPase rate of CLPX lacking the N domain (CLPXΔN) or lacking the N domain and the MI (CLPXΔN,MI). We observed a similar increase in ATPase activity due to MI truncation in CLPX, regardless of the presence of the N-domain (Fig. S2A). The N-domain therefore does not underlie the difference in the effect of MI truncation on the ATPase activity of yeast Mcx1 and human CLPX or otherwise influence the ATPase activity of CLPX. AlphaFold3 predicts with moderate confidence that the N-terminal portion of the MI in Mcx1 forms a continuous α helix with the preceding element of the AAA+ domain, and predicts a low-confidence contact with the N-terminal element of Mcx1. The structure of the human MI region is predicted with lower confidence, with a break in the analogous helix between the MI and the AAA+ domain, and with no predicted interaction with the N-domain (Fig. 1, B and C). These differences in predicted structure additionally suggest that the allosteric communication between the MI and the AAA+ domain is different in these two homologs. In metazoans, mitochondrial ClpX degrades ALAS in response to heme levels by a negative feedback mechanism. This feedback degradation requires an additional adaptor protein that recognizes recruits heme-bound ALAS, POLDIP2 (16). We next asked whether the MI contributes to adaptor-mediated feedback degradation of ALAS by human CLPX. We monitored heme-induced degradation of human ALAS (using the erythroid isozyme, ALAS2) by CLPX or CLPXΔMI (with CLPP and POLDIP2) using a reconstituted system with purified proteins that we described previously (16). We observed that the rate of ALAS2 degradation was not decreased by MI truncation from CLPX, at either 10 μM (Fig. 4D) or 1 μM ALAS2 (Fig. S2B). The rate of degradation decreased similarly across this ten-fold range of ALAS2 concentration for both wild-type CLPX (0.27 ALAS2 min−1 CLPX6−1 at 10 μM to 0.038 ALAS2 min−1 CLPX6−1 at 1 μM) and MI-truncated CLPX (0.36 ALAS2 min−1 CLPX6−1 at 10 μM to 0.046 ALAS2 min−1 CLPX6−1 at 1 μM), indicating that the underlying kinetic parameters are similar for CLPX and CLPXΔMI and that the KM for ALAS2 is above 1 μM. The MI therefore is not required for the adaptor-mediated degradation of ALAS2 by CLPXP. We next tested whether the MI contributes to the interaction of POLDIP2 and CLPX. POLDIP2 binds to the N-domain of CLPX (21), which projects from the CLPX hexamer in close proximity to the MI. AlphaFold3 models of CLPX–POLDIP2 interaction weakly predict a possible additional contact between the MI and the POLDIP2 (Fig. S3, A and B). To test this possibility, we determined the affinity of POLDIP2 for CLPX and CLPXΔMI using fluorescence anisotropy. We observed an equivalent affinity of POLDIP2 for CLPX and CLPXΔMI (KD of ∼60 nM), indicating that the MI is not required for this interaction (Fig. S3C). Although no natural substrate of CLPXP other than ALAS2 is well-characterized in vitro, casein is a well-established model substrate of human CLPX (22, 23). Casein is the eponymous substrate of the Clp (caseinolytic protease) family of unfoldases, but only some members of this family can recognize it. For example, Escherichia coli and C. crescentus ClpX homologs are inactive or nearly inactive again casein (24, 25). To test if the MI contributes to some aspects of CLPX substrate recruitment, we monitored casein degradation by CLPX or CLPXΔMI with CLPP. We found that CLPXPΔMI degraded casein with a two-fold increase in KM and a smaller increase in kcat, indicating the MI can contribute to substrate recruitment by CLPX (Fig. 4, E and F) but that its loss does not globally impair protein translocation by CLPX as previously proposed (17). Discussion Here, we have determined that the MI, an element ubiquitously inserted in the AAA+ domain of mitochondrial ClpX homologs, contributes to both substrate recruitment and engagement for unfolding. Both yeast and human mitochondrial ClpX homologs formed an oligomer that was active for ATP hydrolysis without the MI, although MI truncation altered the kcat for ATP hydrolysis (a several-fold decrease for yeast Mcx1, a small increase for human CLPX). Additionally, the effect of the MI on substrate processing can be uncoupled from its effect on ATPase activity: mutation of a region of the MI had a similar effect on substrate processing as a full deletion but had no effect on ATPase activity. We therefore propose that the MI acts primarily to mediate substrate interactions and is not required to assemble a functional mitochondrial ClpX core. The altered ATPase rates we observe in MI-truncated ClpX homologs supports the idea that the MI communicates with the AAA+ motor domain. However, because we observe a similar substrate-processing defect in both Mcx1ΔΜΙ and a variant with a mutated MI that preserves wild-type levels of ATPase activity (Mcx1loop), we propose that the primary role of this communication is not to globally regulate ATP hydrolysis but rather to coordinate substrate engagement with motor activity. The MI is near-essential for the activation of ALAS by Mcx1. Deletion or mutation of the MI increases the KM for ALAS by an order of magnitude but also substantially decreases the kcat. Based on the position of the MI on the substrate-recruiting face of mitochondrial ClpX homologs, we had hypothesized a likely contribution to substrate affinity, similar to the previously observed contribution of the bacterial ClpX N-domain and adaptors with multiple bacterial substrates (26). We unexpectedly observed that the MI is also important for the rate of activation of ALAS by Mcx1. This finding implies that the MI contributes in some way to positioning ALAS for efficient engagement and unfolding by Mcx1, or in coordinating the action of the substrate-engaged hexamer. The support of heme biosynthesis by mitochondrial ClpX was recently demonstrated to be important for drug resistance in a fungal pathogen (27); the divergence of the MI between fungal ClpX homologs and human CLPX may allow this interaction to be selectively targeted. We find that the mitochondrial insertion more subtly affects several activities of human CLPX. Deletion of the mitochondrial insertion does not affect the affinity of CLPX for POLDIP2, the sole adaptor protein identified for CLPX. POLDIP2-mediated degradation of ALAS as well as degradation of the model substrate casein by CLPXP is slightly accelerated rather than being diminished by MI truncation, perhaps resulting from the slightly increased ATPase rate of CLPXΔΜΙ. We do observe, however, a modest increase in the KM for casein degradation by CLPXP with MI truncation. These data, together with the ubiquitous presence of the MI in all mitochondrial CLPX homologs to our knowledge, suggest that the MI may contribute to substrate recruitment and processing for a subset of CLPX substrates. Because a mechanism of selection for only a single physiological substrate of CLPX, ALAS, has yet been characterized, future investigation of a broader array of candidate CLPX substrates will be necessary to test this idea. Experimental procedures Yeast strain generation Yeast genes were modified at chromosomal loci using standard homologous recombination techniques. Strains were grown for experimental purposes in synthetic defined media (SD + CSM, 2% glucose) (Sunrise Science) at 30 °C with shaking at 220 RPM. ALA measurement Saturated overnight cultures were diluted to OD = 0.1 and grown at 30 °C. At approximately OD600 1.0 (range 0.91–1.06), 10 ml × 1 OD equivalent of culture was rapidly filtered, and cell-laden filters were immediately immersed in 0.7 ml ice-cold 7.5% trichloroacetic acid, 13.3 mM N-ethylmaleimide and incubated for 15 min. Extracts were withdrawn and centrifuged at 2000g for 10 min at 4 °C. The supernatant was mixed with one-third volume 8% acetylacetone in 2 M sodium acetate and incubated for 90 °C for 15 min. After cooling for 5 min at room temperature, the solution was mixed with an equal volume of modified Ehrlich’s reagent (28) and incubated at room temperature for 15 min. Absorbance at 552 nm and 650 nm was measured; ALA content is proportional to A552-A650. Western blotting 1 ml of cells at OD600 = 1.0 (range 0.91–1.06) or equivalent was harvested by centrifugation at 21,000g for 30 s. The cell pellet was resuspended in 200 μl yeast alkaline lysis buffer (0.1 M NaOH, 50 mM EDTA, 2% SDS, 2% β mercaptoethanol) and incubated for 10 min at 90 °C. 5 μl 4 M acetic acid was added, the lysate was vortexed for 30 s, and then mixed with 50 μl SDS loading buffer (250 mM TRIS-HCl pH 6.8, 50% glycerol, 0.05% bromophenol blue) and incubated for 10 min at 90 °C. Samples were separated by SDS-PAGE and transferred to PVFD membrane for 1 h at 100 V. Membrane was blocked with 3% milk in TBST and probed with either 1 μg/ml anti-FLAG (Sigma, F1804) or anti-Por1 (Thermo Fisher Scientific, 16G9E6BC4) (antibody specificity validated by manufacturers). Blots were then probed with 0.5 ng/ml DyLite 800-conjugated goat-anti-mouse antibody and scanned on a LICOR Odyssey Clx. Cloning of vectors for protein purification Vectors for bacterial expression and purification were generated by cloning the following protein sequences, lacking mitochondrial targeting sequences, and their related variants into pET28b or pET23a with an N-terminal His6-SUMO tag, using standard molecular biology procedures: Mcx1 (P38323, 4-520), Hem1 (P09950, 35-548), CLPX (O76031, 65-633), ALAS2 (P22557, 54-587), and POLDIP2 (Q9Y2S7, 52-368). A construct for expressing CLPP (Q16740, 57-277) was previously described (22). Protein expression and purification Constructs were co-transformed with pRIL into BL21(DE3) E. coli for expression. Cultures were grown in Lennox LB at 30 °C with shaking at 220 rpm, except for FLAG-tagged CLPX variants, which were grown in Terrific Broth (TB). When the desired OD600 was achieved (Mcx1, Hem1, CLPX, ALAS2, CLPP: 0.5, POLDIP2: 0.8), the cultures were brought to the desired temperature for expression (Mcx1, CLPX, and ALAS2: 16 °C, CLPP: 37 °C, POLDIP2: 18 °C), induced with IPTG (Mcx1, CLPX, Hem1, and ALAS2: 1 mM, CLPP: 0.5 mM, POLDIP2, and POLDIP2V211Acd: 0.4 mM), and expressed (Mcx1, CLPX, ALAS2, and POLDIP2: overnight; CLPP: 3 h). After expression, cells were harvested by centrifugation at 5000g for 10 min, washed once in lysis buffer by centrifugation, snap-frozen in liquid nitrogen, and stored at −80 °C until purification. For expression of POLDIP2V211Acd, cells were cotransformed with Mj-AcdRS-A9 (encoding an aminoacyl tRNA synthetase/tRNA pair for incorporating acridonylalanine) and pET28b-His6-SUMO-POLDIP2 with position V211 mutated to the amber stop codon (TAG) (29, 30). Acridonylalanine was supplemented to 0.5 mM at time of ITPG induction; expression otherwise proceeded as for wild-type POLDIP2. Cell pellets were lysed by two passes through an ice-cold microfluidizer chamber (LM20-30, Microfluidics) in the same base buffer (25 mM HEPES pH 8, 2 mM MgCl2, 20 mM Imidazole, 10% glycerol, 1 mM DTT, 0.5 mM PMSF) with varying pressure and salt concentrations (Mcx1, CLPX: 10 kpsi, 100 mM KCl and 400 mM NaCl; Hem1, CLPP, ALAS2: 18 kpsi, 100 mM KCl and 400 mM NaCl; POLDIP2, POLDIP2V211Acd: 18 kpsi, 500 mM KCl). Buffers for Hem1 and ALAS2 were additionally supplemented with 20 μM PLP. Buffers for FLAG-tagged CLPX variants were supplemented with 0.01% CHAPS. Proteins were captured with HisPur Ni-NTA resin (Thermo Scientific), washed with additional lysis buffer, and then eluted with lysis buffer supplemented to 250 mM imidazole. Prior to elution, FLAG-tagged CLPX variants were additionally washed as follows: Resin was washed at room temperature with 50 ml of Ni-NTA lysis buffer with 1 mM DTT, 0.005% CHAPS, and 5 mM ATP, followed with lysis buffer supplemented with 1 mM DTT and 5 mM ATP. A final wash was performed at 4 °C with lysis buffer supplemented with 1 mM DTT. The rest of the protein purification proceeded the same as for other ClpX variants above. Eluted proteins were pooled and incubated with enhanced SUMO protease (31) during overnight dialysis into their respective storage buffers. All proteins were prepared in the same base storage buffer (25 mM HEPES pH 7.6, 10% Glycerol, 1 mM DTT, 2 mM MgCl2) with additional salt (Mcx1, CLPX: 300 mM KCl, CLPP: 100 mM KCl, Hem1: 100 mM KCl, ALAS2: 150 mM KCl, POLDIP2: 500 mM KCl). Cleaved SUMO tags were removed by Ni-NTA capture. CLPX was additionally purified by flowing through Sepharose Q resin. For PLP-depleted yeast ALAS, 5 mM hydroxylamine was added to SUMO-subtracted protein and incubated overnight at 4 °C All proteins were purified by gel filtration (Mcx1, CLPX, CLPP, Hem, ALAS2: Superdex S200, POLDIP2: Superdex S75). Fractions corresponding to desired proteins were concentrated, aliquoted, and snap frozen in liquid nitrogen. ATPase assay Mcx1 or CLPX was mixed with ATP and NADH mix: (20 mM NADH, 150 mM phosphoenolpyruvate, 400 U/ml pyruvate kinase, 400 U/ml lactate dehydrogenase) in a final volume of 25 μl reaction buffer (25 mM HEPES pH 7.4, 150 mM KCl, 5 mM MgCl2, 10% glycerol, 1 mM DTT). ATPase activity was monitored by loss of absorbance at 340 nM in a clear, flat-bottom 384-well plate (Corning 3640). To monitor oligomerization, Mcx1 or CLPX concentration was varied and ATP concentration was 2 mM. To monitor kinetic parameters for ATP binding, ATP concentration was varied and measured with 0.5 μM Mcx1 or CLPX hexamer. ATP hydrolysis rates were calculated by dividing the raw absorbance (U/min) by the product of NADH extinction coefficient (6220 M−1 cm−1), path length (0.328 cm), and enzyme concentration. Experiments were conducted at 30 °C for Mcx1, and 37 °C for CLPX. Curves were fit with an allosteric-sigmoidal curve where y is the ATPase rate, x is the concentration of ClpX hexamer or ATP, and H is the Hill coefficient. PLP binding assay PLP binding to Hem1 was monitored by fluorescence (ex. 434 nm, em. 515 nm) in 25 mM HEPES pH 7.4, 150 mM KCl, 5 mM MgCl2, 10% glycerol with 2 mM ATP, 150 μM PLP, and an ATP regenerating system (5 mM creatine phosphate, 50 μg/ml creatine kinase). Fluorescence was monitored at 30 °C in a 384-well plate with 0.5 μM Mcx1 variants and Hem1 concentrations as indicated in individual experiments. Data were fit to the Michaelis-Menten equation to extract kcat and KM paremeters. In vitro protein degradation Degradation of ALAS2 by CLPXP was monitored in vitro by incubating ALAS2 at 10 or 1 μM with 0.5 μM CLPX-FLAG (hexamer), 0.5 μM CLPP (14-mer), 3 μM POLDIP2, and 10 μM hemin (Fisher Scientific AAA1116503) in 25 mM HEPES pH 7.6, 150 mM KCl, 5 mM MgCl2, 10% glycerol, 1 mM DTT with 2 mM ATP and an ATP regenerating system (5 mM creatine phosphate and 50 mg/ml creatine kinase) at 37 °C. Samples were withdrawn at times indicated and immediately mixed with Laemmli SDS sample buffer. Samples were heat-denatured, separated by SDS-PAGE, and stained with SYPRO Red. Gels were imaged on an Amersham Typhoon RGB scanner and quantified in ImageQuant using rolling-ball background subtraction. ALAS remaining at each time point was calculated based on intensity relative to the known starting value. Rates were extracted using linear fits to all replicates. To monitor casein degradation, 0.5 μM CLPXP was incubated with indicated concentrations of FITC-casein (Sigma C3777) in the same buffer as for ALAS2 degradation. Degradation was monitored by fluorescence increase (ex. 384 nm, em. 485 nm) in a 384-well plate. Fluorescence anisotropy In all, 50 nM POLDIP2V211Acd was incubated with increasing concentrations of CLPXE359A in 25 mM HEPES pH 7.4, 150 mM KCl, 5 mM MgCl2, 10% glycerol, and 1 mM DTT. Measurements were conducted at 37 °C in a Fluoromax-4 spectrofluorometer with slit width of 5 nm (ex. 390 nm, em. 444 nm). Anisotropy is reported as a function of CLPX6 concentration, and fit with a quadradic binding equation where P is the polarization value, a is the polarization amplitude, c is the polarization value without added protein, and x is the concentration of CLPX: Data availability All data are contained in the article and supporting information. Supporting information This article contains supporting information. Conflict of interest The authors declare that they have no conflicts of interest with the contents of this article. Acknowledgments We thank Niels Bradshaw, Timo Street, Joey Davis, and members of the Kardon lab for valuable discussions and/or critical feedback on the manuscript. We also thank Tommy Cottle for contribution of Acd-labeled POLDIP2, Ryan Mehl for the gift of pDule2-Mj-AcdRS-A9, and E. James Petersson for the gift of acryidonylalanine. Author contributions A. D. and J. K. conceptualization; A. D., R. B., and J. K. methodology; A. D., R. B., and J. K. investigation; A. D. and J. K. supervision; A. D., R. B., and J. K. writing–original draft; A. D. and J. K. writing–review & editing; J. K. funding acquisition. Funding and additional information This work was supported by National Institutes of Health grant R01GM151332 (J. R. K.). A. D. was supported by NIH T32GM135126. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Reviewed by members of the JBC Editorial Board. Edited by Ursula Jakob Footnotes Present address for Adam DeCosta: Department of Genetics, Harvard Medical School; Rebecca Barrick, MMMP Graduate Program, Washington University in St Louis. Supporting information Supplementary Material References Associated Data Supplementary Materials Supplementary Material Data Availability Statement All data are contained in the article and supporting information.