Distinct mutations in the secreted extracellular matrix protein, fibulin-3 (F3), have been associated with a number of ocular diseases ranging from primary open angle glaucoma to cuticular age-related macular degeneration to a rare macular dystrophy, Malattia Leventinese (ML). The R345W F3 mutation that causes ML leads to F3 misfolding, inefficient secretion and accumulation at higher intracellular steady state levels in cultured cells. Herein, we determined whether fifteen other clinically-identified F3 mutations also led to similar levels of misfolding and secretion defects, which might provide insight into their potential pathogenicity. Surprisingly, we found that only a single F3 variant, L451F, presented with a significant secretion defect (69.5 ± 2.4% of wild-type (WT) F3 levels) and a corresponding increase in intracellular levels (226.8 ± 25.4% of WT F3 levels). Upon follow-up studies, when this conserved residue (L451) was mutated to a charged (Asp or Arg) or bulky (Pro, Trp, Tyr) residue, F3 secretion was also compromised, indicating the importance of small side chains (Leu, Ala, or Gly) at this residue. To uncover potential inherent F3 instability not easily observed under typical culture conditions, we genetically eliminated the sole stabilizing N-linked glycosylation site (N249) from select clinically-identified F3 mutants. This removal exacerbated R345W and L451F secretion defects (19.8 ± 3.0% and 12.4 ± 1.2% of WT F3 levels, respectively), but also revealed a previously undiscovered secretion defect in another C-terminal variant, Y397H (42.0 ± 10.1% of WT F3 levels). Yet, glycan removal did not change the relative secretion of the N-terminal mutants tested (D49A, R140W, I220F). These results highlight the uniqueness and molecular similarities between the R345W and L451F variants and also suggest that previously identified disease-associated mutations (e.g., R140W) are indistinguishable from WT with respect to secretion, hinting that they may lead to disease by an alternative mechanism.

Fibulin-3 (F3 or EFEMP1) is a disulfide-rich, secreted glycoprotein necessary for maintaining extracellular matrix (ECM) and connective tissue integrity. Two studies have identified distinct autosomal recessive F3 mutations in individuals with Marfan Syndrome-like phenotypes. Herein we characterized how one of these mutations, c.163T>C; p.Cys55Arg (C55R), disrupts F3 secretion, quaternary structure, and function by forming unique extracellular disulfide-linked homodimers. Dual cysteine mutants suggest that the C55R-induced disulfide species forms because of new availability of Cys70 on adjacent F3 monomers. Surprisingly, mutation of single cysteines located near C55 (i.e., C29, C42, C48, C61, C70, C159, and C171) also produced similar extracellular disulfide-linked dimers, suggesting that this is not a phenomenon isolated to the C55R mutant. To assess C55R functionality, F3 knockout (KO) retinal pigmented epithelial (RPE) were generated, followed by reintroduction of wild-type (WT) or C55R F3. F3 KO cells produced lower levels of the ECM remodeling enzyme, matrix metalloproteinase 2 and reduced formation of collagen VI ECM filaments, both of which were partially rescued by WT F3 overexpression. However, C55R F3 was unable to compensate for these same ECM-related defects. Our results highlight the unique behavior of particular cysteine mutations in F3 and uncover potential routes to restore C55R F3 loss-of-function.

Destabilizing domains (DDs) are an attractive strategy allowing for positive post-transcriptional small molecule-regulatable control of a fusion protein’s abundance. However, in many instances, the currently available DDs suffer from higher-than-desirable basal levels of the fusion protein. Accordingly, we redesigned the E. coli dihydrofolate reductase (ecDHFR) DD by introducing a library of ∼1200 random ecDHFR mutants fused to YFP into CHO cells. Following successive rounds of fluorescence-activated cell sorting, we identified six new ecDHFR DD clones with significantly enhanced proteasomal turnover in the absence of a stabilizing ligand, trimethoprim (TMP). One of these clones, designated as “C12”, contained four unique missense mutations (W74R/T113S/E120D/Q146L) and demonstrated a significant 2.9-fold reduction in basal levels compared to the conventional ecDHFR DD (i.e., R12Y/G67S/Y100I). This domain was similarly responsive to TMP with respect to dose response and maximal stabilization, indicating an overall enhanced dynamic range. Interestingly, both computational and wet-lab experiments identified the W74R and T113S mutations of C12 as the main contributors toward its basal destabilization. However, the combination of all the C12 mutations was required to maintain both its enhanced degradation and TMP stabilization. We further demonstrate the utility of C12 by fusing it to IκBα and Nrf2, two stress-responsive proteins that have previously been challenging to regulate. In both instances, C12 significantly enhanced the basal turnover of these proteins and improved the dynamic range of regulation post stabilizer addition. These advantageous features of the C12 ecDHFR DD variant highlight its potential for replacing the conventional N-terminal ecDHFR DD and improving the use of DDs overall, not only as a chemical biology tool but for gene therapy avenues as well.

With the increasing use of molecular genetics approaches for determination of potential diseasecausing mutations, it is becoming more important to be able to interpret and act upon the provided results. As an example of such an instance, nearly 300 mutations have been identified in the myocilin (MYOC) gene, which is the most commonly mutated gene causing primary open angle glaucoma. Yet a lack of sufficient information exists for many of these variants, hindering their definitive classification. While the function of MYOC is unclear, biochemically, the vast majority of glaucoma-causing MYOC mutations result in protein non-secretion and intracellular insoluble aggregate formation in cultured cells. Previously we generated a Gaussia luciferase-based MYOC fusion protein to sensitively track secretion of the protein. Herein we applied this same assay to fourteen clinically-derived MYOC variants with varying degrees of predicted pathogenicity and compared the luciferase secretion results with the better established MYOC assay of western blotting. Eight of the variants (G12R, V53A, T204T, P254L, T325T, D380H, D395_E396insDP, and P481S) had not been biochemically assessed previously. Of these, P254L and D395_E396insDP demonstrated significant secretion defects from human embryonic kidney (HEK-293A) cells reminiscent of glaucoma-causing mutations. Overall, we found that the luciferase assay results agreed with western blotting for thirteen of the fourteen variants (93%), suggesting a strong concordance. These results suggest that the Gaussia luciferase assay may be used as a complementary or standalone assay for quickly assessing MYOC variant behavior, and anticipate that these results will be useful in MYOC variant curation and reclassification.

No abstract

Fibulin-3 (F3 or EFEMP1) is a disulfide-rich, secreted glycoprotein necessary for maintaining extracellular matrix (ECM) and connective tissue integrity. Two studies have identified distinct autosomal recessive F3 mutations in individuals with Marfan Syndrome-like phenotypes. Herein we characterized how one of these mutations, c.163T>C; p.Cys55Arg (C55R), disrupts F3 secretion, quaternary structure, and function by forming unique extracellular disulfide-linked homodimers. Dual cysteine mutants suggest that the C55R-induced disulfide species forms because of new availability of Cys70 on adjacent F3 monomers. Surprisingly, mutation of single cysteines located near C55 (i.e., C29, C42, C48, C61, C70, C159, and C171) also produced similar extracellular disulfide-linked dimers, suggesting that this is not a phenomenon isolated to the C55R mutant. To assess C55R functionality, F3 knockout (KO) retinal pigmented epithelial (RPE) were generated, followed by reintroduction of wild-type (WT) or C55R F3. F3 KO cells produced lower levels of the ECM remodeling enzyme, matrix metalloproteinase 2 and reduced formation of collagen VI ECM filaments, both of which were partially rescued by WT F3 overexpression. However, C55R F3 was unable to compensate for these same ECM-related defects. Our results highlight the unique behavior of particular cysteine mutations in F3 and uncover potential routes to restore C55R F3 loss-of-function.

With the increasing use of molecular genetics approaches for determination of potential disease-causing mutations, it is becoming more important to be able to interpret and act upon the provided results. As an example of such an instance, nearly 300 mutations have been identified in the myocilin (MYOC) gene, which is the most commonly mutated gene causing primary open angle glaucoma. Yet a lack of sufficient information exists for many of these variants, hindering their definitive classification. While the function of MYOC is unclear, biochemically, the vast majority of glaucoma-causing MYOC mutations result in protein non-secretion and intracellular insoluble aggregate formation in cultured cells. Previously we generated a Gaussia luciferase-based MYOC fusion protein to sensitively track secretion of the protein. Herein we applied this same assay to fourteen clinically-derived MYOC variants with varying degrees of predicted pathogenicity and compared the luciferase secretion results with the better established MYOC assay of western blotting. Eight of the variants (G12R, V53A, T204T, P254L, T325T, D380H, D395_E396insDP, and P481S) had not been biochemically assessed previously. Of these, P254L and D395_E396insDP demonstrated significant secretion defects from human embryonic kidney (HEK-293A) cells reminiscent of glaucoma-causing mutations. Overall, we found that the luciferase assay results agreed with western blotting for thirteen of the fourteen variants (93%), suggesting a strong concordance. These results suggest that the Gaussia luciferase assay may be used as a complementary or standalone assay for quickly assessing MYOC variant behavior, and anticipate that these results will be useful in MYOC variant curation and reclassification.

Destabilizing domains (DDs) are an attractive strategy allowing for positive post-transcriptional small molecule-regulatable control of a fusion protein’s abundance. Yet in many instances, the currently available DDs suffer from higher-than-desirable basal levels of the fusion protein. Accordingly, we redesigned the E. coli dihydrofolate reductase (ecDHFR) DD by introducing a library of ~1200 random ecDHFR mutants fused to YFP into CHO cells. Following successive rounds of FACS sorting, we identified six new ecDHFR DD clones with significantly enhanced proteasomal turnover in the absence of a stabilizing ligand, trimethoprim (TMP). One of these clones, designated as ‘C12’, contained four unique missense mutations (W74R/T113S/E120D/Q146L) and demonstrated a significant 2.9-fold reduction in basal levels compared to the conventional ecDHFR DD YFP. This domain was similarly responsive to TMP with respect to dose-response and maximal stabilization, indicating an overall enhanced dynamic range. Interestingly, both computational and wet-lab experiments identified the W74R and T113S mutations of C12 as the main contributors towards its basal destabilization. Yet, the combination of all the C12 mutations were required to maintain both its enhanced degradation and TMP stabilization. We further demonstrate the utility of C12 by fusing it to IκBα and Nrf2, two stress-responsive proteins that have previously been challenging to regulate. In both instances, C12 significantly enhanced the basal turnover of these proteins and improved the dynamic range of regulation post stabilizer addition. These advantageous features of the C12 ecDHFR DD variant highlight its potential for replacing the conventional N-terminal ecDHFR DD, and overall improving the use of destabilizing domains, not only as a chemical biology tool, but for gene therapy avenues as well.