Mechanistic Analysis of CCP1 in Generating ΔC2 α-Tubulin in Mammalian Cells and Photoreceptor Neurons

An important post-translational modification (PTM) of α-tubulin is the removal of amino acids from its C-terminus. Removal of the C-terminal tyrosine residue yields detyrosinated α-tubulin, and subsequent removal of the penultimate glutamate residue produces ΔC2-α-tubulin. These PTMs alter the ability of the α-tubulin C-terminal tail to interact with effector proteins and are thereby thought to change microtubule dynamics, stability, and organization. The peptidase(s) that produces ΔC2-α-tubulin in a physiological context remains unclear. Here, we take advantage of the observation that ΔC2-α-tubulin accumulates to high levels in cells lacking tubulin tyrosine ligase (TTL) to screen for cytosolic carboxypeptidases (CCPs) that generate ΔC2-α-tubulin. We identify CCP1 as the sole peptidase that produces ΔC2-α-tubulin in TTLΔ HeLa cells. Interestingly, we find that the levels of ΔC2-α-tubulin are only modestly reduced in photoreceptors of ccp1−/− mice, indicating that other peptidases act synergistically with CCP1 to produce ΔC2-α-tubulin in post-mitotic cells. Moreover, the production of ΔC2-α-tubulin appears to be under tight spatial control in the photoreceptor cilium: ΔC2-α-tubulin persists in the connecting cilium of ccp1−/− but is depleted in the distal portion of the photoreceptor. This work establishes the groundwork to pinpoint the function of ΔC2-α-tubulin in proliferating and post-mitotic mammalian cells.


Introduction
Microtubules are important for a vast array of cellular processes, including intracellular transport, segregating chromosomes during cell division, and formation of motile and primary cilia [1]. The functional plasticity of microtubules derives in part from their posttranslational modifications (PTMs), which include phosphorylation, acetylation, methylation, glutamylation, and glycylation [2,3]. These PTMs occur through covalent modification by "writer" enzymes, but the tyrosinated (Y) C-terminal tail (CTT) of α-tubulin can also undergo enzymatic shortening by specific exo-peptidases to sequentially generate detyrosinated (∆1 or ∆Y), ∆C2 (∆2), and ∆C3 (∆3) α-tubulin [4][5][6]. Tubulin PTMs alter the structural and chemical properties of α,β-tubulin and can thus, in principle, impact the ability of microtubules to engage non-motile microtubule-associated proteins (MAPs) and molecular motor proteins, i.e., kinesins and dynein. Consistent with this idea, dynein-dynactin lands preferentially on Y-microtubules via a CAP-Gly protein motif in p150 Glued [7]. Conversely, kinesin-1 and CENP-E preferentially engage and translocate on ∆Y-microtubules [8,9]. More recent work has demonstrated that microtubules assembled from recombinant Y, ∆Y, ∆C2, or ∆C3 α-tubulin exhibit similar dynamic properties in the absence of MAPs, suggesting that PTM-specific microtubule dynamics are generated by effector ("reader") proteins. This idea is supported by the knowledge that Y-microtubules are more effective at recruiting CLIP-170 and EB1 to microtubule plus ends, leading to preferential regulation of Y-microtubule dynamics [10].

Quantseq
Total RNA prepared from HeLa or TTL∆ cell lines was submitted to the University of Michigan Advanced Genomics core for QuantSeq 3 mRNA sequencing (https://doi. org/10.1038/nmeth.f.376, accessed on 1 December 2022). FASTQ files from the 100 bp sequencing run were first processed with the umi2index script supplied by Lexogen to strip the UMI from the reads. Cutadapt v1.15 was used to remove adapter resulting in 28,26,26, and 29 million reads for downstream analysis for samples AP-1, 2, 3, and 4, respectively. STAR v2.7.2b (https://doi.org/10.1093/bioinformatics/bts635, accessed on 1 December 2022) was used to align reads to the GRCh38 reference (command provided below). The Lexogen provided, collapse_UMI_bam script was then used to remove PCR duplicates.
Features were counted using featureCounts from the Rsubread Bioconductor package (https://doi.org/10.1093/nar/gkz114, accessed on 1 December 2022) with respect to features defined by the GTF file supplied with the GRCh38 reference. Normalized counts per million were generated using edgeR (https://doi.org/10.1093/bioinformatics/btp616, accessed on 1 December 2022) and the trimmed mean of M-values (TMM) method. Values were log2 transformed.
Isolated mouse outer segments: Immunostaining protocol originally described in [35]. To prepare isolated outer segments, retinas were dissected and collected in a microcentrifuge tube containing 125 µL of Mouse Ringers per retina. The retinas were vortexed on HIGH for 1 min and large debris pelleted using a benchtop spinner for 5 s. The supernatant containing isolated outer segments was then plated onto 13 mm poly-L-lysine glass coverslips (354085; Corning, Glendale, AZ, USA) before fixation in 4% paraformaldehyde in PBS at room temperature for 5 min. Plated outer segments were then rinsed with PBS and permeabilized for 5 min by incubating in 0.02% SDS diluted in PBS. Coverslips were rinsed, and primary antibodies were diluted in 2% donkey serum (Thermo Fisher Scientific, Cat# NC0629457) and 0.5% Triton X-100 in PBS and incubated for 2 h at room temperature. Primary antibodies used are rabbit monoclonal anti-∆C2 α-tubulin RM447, 0.1 µg/mL; rabbit monoclonal anti-detyrosinated α-tubulin RM444, 0.1 µg/mL; mouse monoclonal anti-α-tubulin DM1A (Sigma-Aldrich, Cat# T9026), 1:2000; mouse monoclonal anti-polyglutamylation GT335 (AdipoGen, San Diego, CA, USA, Cat# AG-20B-0020), 1:2000. Coverslips were rinsed and incubated with donkey secondary antibodies listed above for 1 h at room temperature before rinsing and mounting using Prolong Gold.

Image Analysis
Images from mouse retinal sections or isolated outer segments were acquired using a Zeiss Observer 7 inverted microscope equipped with a 63× oil-immersion objective (1.40 NA), LSM 800 confocal scanhead outfitted with an Airyscan super resolution detector controlled by Zen 5.0 software (Carl Zeiss Microscopy, Oberkochen, Germany). Manipulation of images was limited to adjusting the brightness level, image size, rotation, and cropping using FIJI (ImageJ, https://imagej.net/Fiji, accessed on 1 September 2017). All phenotypes were measured using images taken from at least 3 independent retinas.
Measuring fluorescence intensity of axoneme in isolated outer segments: For each antibody condition, the axoneme was stained with either α-tubulin or GT335 antibodies and outer segment membranes co-stained with WGA lectin conjugated to Alexa594 (W11262, Thermo Fisher Scientific). For each mouse, 10-15 isolated outer segments were imaged with an Airyscan detector on an LSM 800 microscope (Carl Zeiss Microscopy). Resolution was standardized to 0.04 µm/pixel by setting the axial zoom to 3X, and 0.35 µm Z-stacks were taken to collect the entire thickness of the outer segment, generally 0.9-1.3 µm in depth. A pre-determined laser power was set to the ccp1 +/+ levels for each antibody condition. Using FIJI software, images were stacked, and a 15-pixel segmented line was drawn along the axoneme to acquire the fluorescence intensity measurements for each channel. The intensity measurements of each antibody were then aligned to the start of the connecting cilium, based on predetermined WGA parameters, and plotted using Prism 9 software.

HeLa Cells Lacking TTL Contain High Levels of ∆C2-α-Tubulin
We began by examining the levels of ∆Y-α-tubulin and ∆C2-α-tubulin in a panel of commonly used cell lines with PTM-specific rabbit monoclonal antibodies that we generated in collaboration with RevMAb Biosciences (clones RM444 and RM447 for ∆Yand ∆C2-α-tubulin, respectively; Figure S1). By immunoblotting, the levels of ∆Y-α-tubulin were found to be low in HeLa, DLD-1, HEK293T, U2OS, and CHL-1 cells when compared to bovine brain tubulin ( Figure 1A). ∆C2-α-tubulin was undetectable in all of these cell lines under the conditions of our experiment. To determine if the enzymatic machinery that produces ∆C2-α-tubulin is present in HeLa cells, we used an inducible CRISPR/Cas9 system [36] to edit the TTL open reading frame (ORF) since deletion of this enzyme leads to widespread production of ∆C2-α-tubulin in mice [20]. We screened for potential TTL∆ clones by immunoblotting with ∆Y-α-tubulin antibodies, and sequencing of one TTL∆ clonal isolate revealed three distinct insertions/deletions (INDELs; Figure 1B). Each INDEL presumably corresponds to one TTL allele, consistent with the literature, suggesting that various HeLa strains are hypertriploid [37,38]. All three INDELs produce frameshift mutations that result in early stop codons, and we confirmed that no TTL protein is present in cell lysates of this TTL∆ clone ( Figure 1C).
We proceeded by staining the parental HeLa and TTL∆ cell lines with antibodies against α-tubulin (DM1A), ∆Y-α-tubulin, and ∆C2-α-tubulin. Consistent with immunoblotting ( Figure 1A), ∆Y-α-tubulin levels were very low in HeLa cells, although short segments of ∆Y-MTs were present in the cytoplasm of interphase cells ( Figure 1D). ∆C2-α-tubulin was undetectable ( Figure 1D), but we note that our antibody stains puncta in the nuclei of a subset of interphase cells, which likely results from cross-reactivity with a protein that is not tubulin. In contrast to the parental cell line, TTL∆ cells harbored very high levels of both ∆Y-α-tubulin and ∆C2-α-tubulin ( Figure 1D), a result that was recapitulated by immunoblotting of lysates prepared from these cell lines ( Figure 1E). Levels of ∆Y-α-tubulin and ∆C2-α-tubulin were reduced following re-introduction of mouse TTL into TTL∆ cells, indicating that high levels of ∆Y-α-tubulin and ∆C2-α-tubulin are a specific consequence of TTL deletion ( Figure S2). We obtained two additional TTL∆ clones and confirmed that the results were consistent among the three cell lines ( Figure S3). By the reduction of Y-tubulin on immunoblots, we estimate that roughly half of the α-tubulin present in TTL∆ cells is either detyrosinated or in the ∆C2 form.

C2-α-Tubulin Levels Do Not Increase When Microtubules Are Artificially Stabilized
Several explanations could account for the increase in ∆C2-α-tubulin that is present in cells that lack TTL. The enzyme responsible for this PTM could be present but maintained in an inactive state, e.g., through an inhibitory mechanism(s). Alternatively, the peptidase could be induced at the transcriptional or post-transcriptional level upon removal of TTL. Since the six CCP genes are not differentially expressed in HeLa versus TTL∆ cells ( Figure  S4), this did not seem a likely possibility. Lastly, the catalytic activity of the peptidase may be too slow to keep pace with the tyrosination/detyrosination cycle. To investigate this, we treated cells with the microtubule stabilizing drug taxol, a perturbation well known to increase microtubule detyrosination [28]. This effect reflects the fact that VASH1-SVBP more efficiently detyrosinates microtubules, whereas TTL tyrosinates tubulin in its unpolymerized form [21,39,40]. Thus, we hypothesized that taxol treatment of cells would provide the ∆C2-α-tubulin peptidase with ample substrate (∆Y-α-tubulin) and time to generate ∆C2-α-tubulin.  We treated cell lines that harbor varying levels of ∆Y-α-tubulin (HeLa, DLD-1, CHL-1, and TTL∆) with 10 µM taxol for 12 h and then processed them for immunofluorescence ( Figure 2A) or immunoblotting ( Figure 2B). As expected, taxol treatment caused HeLa, DLD-1, and CHL-1 cells to express high levels of ∆Y-α-tubulin and arrest in mitosis. Surprisingly, however, taxol treatment did not cause an increase in ∆C2-α-tubulin levels in these cell lines. The presence of both ∆Y-α-tubulin and ∆C2-α-tubulin was only observed in taxoltreated TTL∆ cells (Figure 2A). To validate this result biochemically, we prepared cell lysates from DMSO-or taxol-treated HeLa, DLD-1, CHL-1, and TTL∆ cells and subjected them to immunoblotting analysis using antibodies against α-tubulin (DM1A), ∆Y-α-tubulin, and ∆C2-α-tubulin. Mirroring observations made at the cellular level, we observed that taxol treatment increased the levels of ∆Y-α-tubulin but not ∆C2-α-tubulin in HeLa, DLD-1, and CHL-1 cells. TTL∆ cells exhibited high levels of both ∆Y-α-tubulin and ∆C2-α-tubulin regardless of drug treatment. Collectively, our results are inconsistent with the idea that ∆C2-α-tubulin levels are low because the enzyme responsible is too slow to keep pace with the tyrosination/detyrosination cycle.

Depletion or Gene Deletion of CCP1 Blunts the Production of ΔC2-α-Tubulin in TTLΔ Cells
To identify the peptidase that generates ΔC2-α-tubulin in TTLΔ cells, we first conducted a focused siRNA screen targeting the six CCPs (CCP1-6) encoded by the human genome. HeLa or TTLΔ cells were subjected to two rounds of siRNA transfections, fixed five days following the start of the experiment, and stained with antibodies against αtubulin (DM1A) and ΔC2-α-tubulin. Depletion of CCP1 uniquely caused a reduction in the number of cells that were positive for ΔC2-α-tubulin ( Figures 3A). By immunoblotting, we confirmed that CCP1 protein levels were reduced in siRNA-transfected cells, although depletion was incomplete. Immunoblotting also demonstrated that ΔC2-α-tubulin levels, but not ΔY-α-tubulin, dropped by 78% relative to control siRNA-transfected cells ( Figure 3B). In parallel, we used QuantSeq to analyze the expression levels of CCP1-6 in HeLa and TTLΔ cells and found that CCP1 and CCP5 were the only two CCPs that are expressed at detectable levels ( Figure S4). Since CCP5 is largely thought to remove the E

Depletion or Gene Deletion of CCP1 Blunts the Production of ∆C2-α-Tubulin in TTL∆ Cells
To identify the peptidase that generates ∆C2-α-tubulin in TTL∆ cells, we first conducted a focused siRNA screen targeting the six CCPs (CCP1-6) encoded by the human genome. HeLa or TTL∆ cells were subjected to two rounds of siRNA transfections, fixed five days following the start of the experiment, and stained with antibodies against αtubulin (DM1A) and ∆C2-α-tubulin. Depletion of CCP1 uniquely caused a reduction in the number of cells that were positive for ∆C2-α-tubulin ( Figure 3A). By immunoblotting, we confirmed that CCP1 protein levels were reduced in siRNA-transfected cells, although depletion was incomplete. Immunoblotting also demonstrated that ∆C2-α-tubulin levels, but not ∆Y-α-tubulin, dropped by 78% relative to control siRNA-transfected cells ( Figure 3B). In parallel, we used QuantSeq to analyze the expression levels of CCP1-6 in HeLa and TTL∆ cells and found that CCP1 and CCP5 were the only two CCPs that are expressed at detectable levels ( Figure S4). Since CCP5 is largely thought to remove the E residue from mono-glutamylated α-tubulin [6], it is reasonable that CCP1 is solely responsible for generating ∆C2-α-tubulin in TTL∆ HeLa cells.
To rule out the possibility that other enzymes generate ∆C2-α-tubulin in CCP1depleted TTL∆ HeLa cells, we again used the CRISPR/Cas9 system [31] to disrupt the CCP1 gene in TTL∆ cells. We screened for potential TTL∆ CCP1∆ clones by immunoblotting with ∆C2-α-tubulin antibodies, and sequencing of one CCP1∆ clonal isolate revealed three distinct insertions/deletions (INDELs; Figure 3C). All three INDELs produce frameshift mutations that result in early stop codons in CCP1 isoform 1 (UniProt, Q9UPW5-1). For the isoform 2 that lacks internal 120-nt (UniProt, Q9UPW5-2; NCBI Reference Sequence, NM_015239.3), one of the INDELs produced a 16-nt deletion across intron 10 and exon 11, which most likely results in skipping of exon 11 (58-nt) and subsequent frameshift and an early stop codon. We confirmed that no CCP1 protein is present in cell lysates of this TTL∆ CCP1∆ clone ( Figure 3D). TTL∆ CCP1∆ cells did not exhibit gross defects in cell proliferation (data not shown).

CCP1 Requires ∆Y-α-Tubulin as a Substrate to Generate ∆C2-α-Tubulin
While levels of total Y-and ∆Y-α-tubulin were not affected in TTL∆ CCP1∆ cells compared to TTL∆ cells ( Figure 3D), ∆C2-tubulin in the TTL∆ line seemed to represent only a small portion of total α-tubulin. One possible explanation is that removal of residues from the α-CTT occurs in a stepwise fashion, i.e., ∆Y-α-tubulin is generated first, followed by production of ∆C2-tubulin. If this were the case, CCP1 overexpression should generate ∆C2-α-tubulin more efficiently in cells that contain higher levels of ∆Y-α-tubulin. To test this, we overexpressed EGFP-tagged CCP1 or a catalytically dead version of the enzyme (H880S/E883Q, hereafter HS/EQ, [6]) in HeLa, HEK293T, or CHL-1 cells; HeLa cells do not harbor significant levels of ∆Y-α-tubulin, whereas ∆Y-α-tubulin is readily detectable in HEK293T and CHL-1 cells ( Figure 1A). HeLa and HEK293T cells were efficiently transfected with our constructs, and we could thus test the effect of CCP1 overexpression on ∆C2-αtubulin levels in these cell lines by both immunostaining and immunoblotting. In HeLa cells, ∆C2-α-tubulin was undetectable in both immunoblots and immunostained cells regardless of whether they were transfected with wild-type EGFP-CCP1 or the HS/EQ CCP1 mutant. In contrast, overexpression of wild-type CCP1, but not the HS/EQ mutant, led to an increase in the amount of ∆C2-α-tubulin in HEK293T cells ( Figures 4A,B and S7). ∆C2α-tubulin decorated perinuclear portions of microtubules, which presumably represent minus ends that are concentrated at the centrosome(s) ( Figure 4B). Unfortunately, CHL-1 cells were not efficiently transfected, precluding a test of whether CCP1 overexpression can increase ∆C2-α-tubulin levels by immunoblotting. However, immunostaining of CHL-1 cells revealed that ∆C2-microtubules were abundant in interphase cells transfected with wild-type CCP1 but not the HS/EQ mutant ( Figures 4B and S7).    To conclusively demonstrate that CCP1 uses ∆Y-microtubules as a substrate to generate ∆C2-α-tubulin, we transfected HeLa cells that inducibly express VASH1-SVBP with wildtype CCP1 or the HE/SQ mutant. VASH1-SVBP expression was induced by the addition of doxycycline for 16 h and then transfected with EGFP-CCP1 or EGFP-CCP1 HS/EQ . Cells were processed for immunofluorescence or harvested for immunoblot analysis 24 h following transfection. These experiments clearly demonstrated that overexpression of wild-type CCP1, but not the HE/SQ mutant, caused an increase in the levels of ∆C2-α-tubulin only in cells that were induced to express VASH1-SVBP (Figures 5 and S8). We conclude that proteolytic trimming of the α-CTT occurs in a stepwise manner: VASH1-SVBP first removes the C-terminal Y residue, producing ∆Y-α-tubulin that is then used as a substrate for CCP1 to generate ∆C2-α-tubulin.

CCP1 Converts Only a Subset of Photoreceptor Axoneme Microtubules to ∆C2-α-Tubulin
To investigate whether CCP1 can generate ∆C2-α-tubulin in an in vivo context, we looked at photoreceptor cells in the mouse retina that contain a specialized light-sensitive primary cilium, called the outer segment. The outer segment contains all the structural features of a primary cilium, including a microtubule axoneme emanating from the basal body and extended transition zone referred to as the connecting cilium but has been modified to house hundreds of flattened membrane vesicles for efficient light capture. A mouse model with mutation in CCP1 was shown to slowly lose photoreceptor cells during the first year of life in addition to rapid loss of cerebellar Purkinje cells [41]. Due to the ongoing photoreceptor degeneration present in the ccp1 knockout mice [42,43], we analyzed axoneme tubulin PTM modifications at postnatal day 15, a timepoint when the ciliary outer segment is formed but ongoing degeneration has not begun. Fresh mouse retinal crosssections were stained with antibodies against ∆C2-α-tubulin and centrin-1 to label the connecting cilium and counterstained with Alexa-594-conjugated WGA that labels outer segment membranes ( Figure 6A). Confocal imaging of the outer segment showed reduced levels of ∆C2-α-tubulin in the ccp1 −/− mice compared to littermate controls. To quantify tubulin staining intensity, we isolated outer segments from ccp1 −/− and ccp1 +/+ mice at P15, plated onto coverslips, and stained outer segments with WGA-594 and antibodies against α-tubulin (DM1A) and ∆C2-α-tubulin. Airyscan images were acquired of individual outer segments so that intensity measurements could be collected along the axoneme. Axonemal intensity measurements across outer segments were aligned to the beginning of the connecting cilium based on a predetermined threshold of WGA signal. While ccp1 +/+ mouse outer segments contain an even distribution of ∆C2-α-tubulin across the axoneme, the intensity of ∆C2-α-tubulin staining is reduced outside the connecting cilium in ccp1 −/− outer segments ( Figure 6B). In contrast, ∆Y-α-tubulin staining of the photoreceptor axoneme did not change in isolated ccp1 −/− outer segments ( Figure S9). A~20% reduction in ∆C2-αtubulin was also observed by immunoblot analysis from retina lysates of ccp1 +/+ and ccp1 −/− mice ( Figure 6C). Together, our data suggest that CCP1 is used to generate axonemal ∆C2α-tubulin in photoreceptor outer segments outside of the connecting cilium. In addition, our data indicate that a second peptidase is capable of generating ∆C2-α-tubulin within the connecting cilium, which presumably comprises a large fraction of ∆C2-α-tubulin in retinal tissue.

CCP6 Is a Candidate for a Second Peptidase That Generates ∆C2-α-Tubulin
To identify other CCPs that can generate ∆C2-α-tubulin, we cloned full-length openreading frames of CCP1-6 into pEGFP-C1 and assessed the ability of these constructs to produce ∆C2-α-tubulin when transfected into HeLa TTL∆ CCP1∆ cells. This cell line is ideal to screen for enzymes that generate ∆C2-α-tubulin because this PTM is absent when CCP1 is deleted in HeLa cells and the precursor to ∆C2-α-tubulin, ∆Y-α-tubulin, is abundant ( Figure 3E). Immunostaining of transfected cells with antibodies against ∆C2-α-tubulin showed that CCP1 and CCP6 were the most effective at generating ∆C2-α-tubulin in TTL∆ CCP1∆ cells and that CCP3 produced a small amount of ∆C2-α-tubulin (Figure 7).

CCP6 Is a Candidate for a Second Peptidase that Generates ΔC2-α-Tubulin
To identify other CCPs that can generate ΔC2-α-tubulin, we cloned full-length open reading frames of CCP1-6 into pEGFP-C1 and assessed the ability of these constructs t produce ΔC2-α-tubulin when transfected into HeLa TTLΔ CCP1Δ cells. This cell line i  Figure 3E). Immunostaining of transfected cells with antibodies against ΔC2α-tubulin showed that CCP1 and CCP6 were the most effective at generating ΔC2-α-tubulin in TTLΔ CCP1Δ cells and that CCP3 produced a small amount of ΔC2-α-tubulin (Figure 7). Figure 7. CCP6 is a candidate for a second peptidase that generates ΔC2-α-tubulin. Immunofluorescence of TTL∆ CCP1∆ cells transiently expressing EGFP-tagged CCPs. In merged images, total α-tubulin (DM1A staining) is shown in yellow, ∆C2-α-tubulin in magenta, EGFP in green, and DNA (DAPI staining) in blue. Scale bars, 10 µm.

Discussion
In contrast to many tubulin PTMs, such as poly-glutamylation, poly-glycylation, and acetylation, the function(s) of ΔC2-α-tubulin is unknown. ΔC2-α-tubulin is enriched in highly stable microtubules, such as those in neurons, the primary cilium, and centrosomes [5], but whether ΔC2-α-tubulin emerges as a consequence of or promotes microtubule stability has not been determined. A major reason for this knowledge gap is that ΔC2-αtubulin-generating enzymes-the CCPs-have dual functions in post-translationally Figure 7. CCP6 is a candidate for a second peptidase that generates ∆C2-α-tubulin. Immunofluorescence of TTL∆ CCP1∆ cells transiently expressing EGFP-tagged CCPs. In merged images, total α-tubulin (DM1A staining) is shown in yellow, ∆C2-α-tubulin in magenta, EGFP in green, and DNA (DAPI staining) in blue. Scale bars, 10 µm.

Discussion
In contrast to many tubulin PTMs, such as poly-glutamylation, poly-glycylation, and acetylation, the function(s) of ∆C2-α-tubulin is unknown. ∆C2-α-tubulin is enriched in highly stable microtubules, such as those in neurons, the primary cilium, and centrosomes [5], but whether ∆C2-α-tubulin emerges as a consequence of or promotes microtubule stability has not been determined. A major reason for this knowledge gap is that ∆C2-α-tubulin-generating enzymes-the CCPs-have dual functions in post-translationally modifying microtubules, complicating analysis. CCPs remove glutamate residues that are appended onto the C-terminal tail of α-tubulin, i.e., poly-glutamate chains, as well as those that are encoded at the -2 and -3 positions by the α-tubulin genes [6,23]. Most work on CCPs has focused on their role in regulating poly-glutamate chains, as this activity is required to prevent the degeneration of multiple neuronal sub-types including Purkinje cell [27], motor [26], and retinal neurons [44]. A second reason is that we do not yet know the enzymes that generate ∆C2-α-tubulin in a physiological context. CCP1, a widely expressed CCP1 ( Figure S5 and [23]), has been shown to generate ∆C2-α-tubulin when overexpressed in cultured cells [6,23] and to be required for the formation of ∆C2-α-tubulin in HEK293T cells [24]. However, only skeletal muscle is depleted of ∆C2-α-tubulin in tissues of ccp1 −/− mice [6], indicating that other enzymes cooperate with CCP1 to generate ∆C2-α-tubulin in differentiated cells.
In this work, we used knowledge that deletion of TTL increases the levels of ∆C2-αtubulin [20] to screen for CCPs that generate ∆C2-α-tubulin in HeLa cells. Previous work to study CCPs employed HEK293T cells, which contain only low levels of ∆Y-α-tubulin ( Figure 1A). In contrast, ∆Y-α-tubulin and ∆C2-α-tubulin comprise~half of the α-tubulin pool in TTL∆ cells ( Figure 1C). Since ∆Y-α-tubulin is a prerequisite for the formation of ∆C2-α-tubulin (this work), TTL∆ cells are a robust model system to study the enzymes that generate the ∆C2, and presumably ∆C3 [4], form of α-tubulin. A second advantage of using TTL∆ cells to study the biogenesis of ∆C2-α-tubulin is that the ∆C2-α-tubulin-generating reaction appears to be gated by TTL through a mechanism that is unclear. Specifically, treatment of TTL+ cells with taxol for 24 h increases the levels of ∆Y-α-tubulin but not ∆C2-α-tubulin. The reason for this is not clear. CCP1 may not be able to compete with other α-tubulin C-terminal tail-binding proteins or may be kept inactive through an unknown regulatory mechanism(s). Future work is needed to test these possibilities.
TTL∆ cells did not exhibit gross proliferation defects or abnormal progression through cell division (data not shown), which is consistent with the idea that TTL∆ cells adapt to heightened levels of ∆Y-α-tubulin [45] and the ability of ttl −/− mice to undergo normal embryonic development [20]. In HeLa cells lacking TTL, CCP1 is the sole carboxypeptidase that generates ∆C2-α-tubulin, which we demonstrated unambiguously by showing that ∆C2-α-tubulin is absent in HeLa cells lacking both TTL and CCP1. Like TTL∆ cells, cells lacking both TTL and CCP1 did not exhibit defects in cell proliferation, suggesting that ∆C2-α-tubulin is not required for viability of HeLa cells. Of the five remaining CCPs, only CCP5 is expressed at detectable levels in HeLa cells. CCP5 is unique among the CCPs in that it (1) removes the E residue from mono-glutamylated α-tubulin and (2) does not generate ∆C2-α-tubulin when overexpressed [6]. It is therefore reasonable that ∆C2-αtubulin cannot be produced in HeLa cells lacking CCP1, and an interesting direction for future work will be to survey the levels of ∆C2-α-tubulin in various tissues from ttl −/− ccp1 −/− mice.
Our work in photoreceptors is consistent with the notion that other CCPs can generate ∆C2-α-tubulin in differentiated tissues [6]. Indeed, the levels of ∆C2-α-tubulin are only decreased by~20% in retinal extracts prepared from ccp1 −/− mice ( Figure 6C). Unexpectedly, however, we observed that ∆C2-α-tubulin is specifically depleted from a region of the photoreceptor cilium of ccp1 −/− mice that is distal to the connecting cilium ( Figure 6B), whereas the fluorescence intensity of ∆C2-α-tubulin in the connecting cilium is comparable to wild-type photoreceptors. This finding suggests that the activities of at least two CCPs are directed towards different regions of the outer segment: CCP1 targets the outer segment distal to the connecting cilium, whereas another CCP(s) generates ∆C2-α-tubulin within the connecting cilium. Previous work showed that CCP1-4 and CCP6 can generate ∆C2α-tubulin when overexpressed in HEK293T cells [6,23]. However, the low levels of ∆Y-αtubulin in HEK293T cells, coupled with the use of truncated CCP2 and CCP3 constructs in earlier work [23], motivated us to reinspect the activities of the CCPs in TTL∆ CCP1∆ cells. Although we cannot rule out the possibility that co-factors, e.g., regulatory binding partners, are required for CCPs to be active in HeLa cells, our data strongly suggest that CCP6 and CCP1 are best positioned to generate ∆C2-α-tubulin in mammalian cells. Interestingly, CCP6 is not expressed at significant levels in rod photoreceptors ( Figure S10) prior to P15, the time at which we processed our specimens for analysis. Therefore, it is currently unclear which CCP is responsible for generating ∆C2-α-tubulin in the outer segment, and this outstanding issue will be a focus for future work.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/biom13020357/s1, Figure S1: Specificity of a-CTT antibodies used in this study; Figure S2; The levels of ∆Yand ∆C2-α-tubulin decrease in TTL∆ cells transiently expressing GFP-mTTL; Figure S3: Characterization of additional TTL∆ clones; Figure S4: CCP1 and CCP5 are the only expressed cytosolic carboxypeptidases in the parental HeLa and TTL∆ lines; Figure S5: Expression levels of CCP genes in various tissues; Figure S6: Immunofluorescence staining of TTL∆ CCP1∆ cells undergoing cytokinesis; Figure S7: Transient expression of CCP1 mutant in HeLa, HK293T, and CHL-1 cells; Figure S8: Immunofluorescence images of ∆Y-α-tubulin in HeLa cells expressing VASH1-SVBP; Figure S9: ∆Y-α-tubulin staining is not changed in the photoreceptor axoneme of ccp1 −/− mouse retinas; Figure S10: Expression levels of CCPs in photoreceptor neurons at the indicated days of post-natal development; Table S1: Oligonucleotide DNA primers used in this study. Institutional Review Board Statement: All mice were handled following protocols approved by the Institutional Animal Care and Use Committees at the University of Michigan (registry number A3114-01). The University of Michigan is an AAALAC accredited organization. As there are no known gender-specific differences related to vertebrate photoreceptor development and/or function, male and female mice were randomly allocated to experimental groups. All mice were housed in a 12/12 h light/dark cycle with free access to food and water.