Impaired nucleotide excision repair upon macrophage differentiation is corrected by E1 ubiquitin-activating enzyme

Abstract

Global nucleotide excision repair is greatly attenuated in terminally differentiated mammalian cells. We observed this phenomenon in human neurons and in macrophages, noting that the transcription-coupled repair pathway remains functional and that there is no significant reduction in levels of excision repair enzymes. We have discovered that ubiquitin-activating enzyme E1 complements the repair deficiency in macrophage extracts, and although there is no reduction in the concentration of E1 upon differentiation, our results indicate a reduction in phosphorylation of E1. In preliminary studies, we have identified the basal transcription factor TFIIH as the potential target for ubiquitination. We suggest that this unusual type of regulation at the level of the E1 enzyme is likely to affect numerous cellular processes and may represent a strategy to coordinate multiple phenotypic changes upon differentiation by using E1 as a “master switch.”

Nucleotide excision repair (NER) is the most versatile DNA repair system; it handles a wide variety of lesions, from UV-induced pyrimidine dimers to bulky chemical adducts, to intrastrand DNA cross-links and even bound protein fragments. The early steps in NER can be divided into two subpathways: global genomic repair (GGR) that recognizes and removes lesions throughout the genome, and transcription-coupled repair (TCR) that preferentially repairs the transcribed strand of active genes, most likely by using translocating RNA polymerase as a lesion sensor.

The efficiency of NER is known to be affected by cellular differentiation (reviewed in ref. 1). In particular the GGR subpathway is greatly attenuated in human neurons, either primary (1) or derived from NT2 neuroteratoma cells (2), whereas TCR is still active in both cases. In the terminally differentiated cells, however, the nontranscribed strand in active genes is also efficiently repaired, a phenomenon that we have termed differentiation-associated repair (DAR) and which we have reasoned is required to maintain lesion-free templates for TCR.

It seems likely that attenuation of NER at the global genome level, and its reliance on TCR and DAR to maintain the essential active genes, is common in terminally differentiated cells. Indeed, studies with other differentiated cell systems have yielded similar results. Among numerous examples, rat skeletal myocytes exhibited a decrease in repair synthesis in the course of differentiation (3, 4), whereas TCR was less affected (4). Mouse 3T3 cells displayed attenuated UV-induced repair synthesis after differentiation into adipocytes (5). The repair of UV-induced lesions was reduced in differentiated keratinocytes, compared with that in cells from the basal layers of human skin (6). Human neuroblastomas repaired benzo[a]pyrene diol-epoxide adducts less rapidly after differentiation (7), and differentiated mouse neuroblastomas exhibited similar deficiencies in the repair of UV-induced lesions (8).

What is the mechanism by which GGR is attenuated in terminally differentiated cells? To answer this question two human leukemia cell lines were used to generate terminally differentiated macrophages. HL60 is a promyelocytic leukemia, which can be differentiated into macrophages by application of phorbol ester (TPA). THP1 is a monocytic leukemia, farther down the monocyte/macrophage differentiation pathway, which can also be differentiated into macrophages with TPA. The advantage of these cell systems for study over preneurons is that they can be cultured more efficiently to obtain large numbers of cells for analysis. As with neurons, GGR was reduced in macrophages compared with that in the naïve precursor cells (Fig. 1). The GGR deficiency was confirmed in vitro with extracts from macrophages and a complementation assay was used to learn which factor(s) might be involved. The in vitro assay has revealed that the E1 ubiquitin-activating enzyme can complement the GGR deficiency.

Fig. 1.

In vivo repair of UV-induced lesions. HL60 and THP1 cells, naïve (white bars) or differentiated with TPA for 16 h (gray bars) or 48 h (black bars), were irradiated with a dose (10 J/m²) of 254-nm UV light. Cells were harvested either immediately or 24 h later, and DNA was purified, blotted onto a nitrocellulose membrane, and probed with antibodies specific for CPDs (Left) or (6-4)PPs (Right). Results are expressed as percentage of the lesions measured at T0 that remained 24 h after irradiation. Error bars are the SEM of two to four experiments. ∗, P < 0.05; ∗∗, P < 0.01 (Student's t test). More detailed time-course experiments can be found in refs. 9 and 21.

Results and Discussion

NER Is Impaired at the Global Genomic Level upon Macrophage Differentiation.

Following upon our earlier studies with differentiating neurons, we examined the efficiency of NER in naïve versus terminally differentiated human leukemia cells by measuring the repair of the two main UV-induced lesions: CPDs and (6-4)pyrimidine–pyrimidone photoproducts [(6-4)PPs]. We found that CPDs were proficiently repaired in naïve HL60 cells, but less efficiently in naïve THP1 cells (Fig. 1). Macrophage-like cells differentiated from either cell line were almost completely deficient in repair of CPDs. By contrast, the repair of (6-4)PPs was proficient in both cell lines, before and after differentiation, although more detailed time-course experiments (9) revealed that it was slightly slower in both cell lines after differentiation for 48 h.

The difference between the repair efficiencies for these lesions probably reflects the fact the (6-4)PPs are a better substrate for NER than are CPDs (10), most likely because they cause a greater distortion of the double helix structure, but also because of their preferential locations in the linker regions of nucleosomes (11). This is reminiscent of our previous observations in fetal human neurons, which, when kept in culture for several months, lost the ability to efficiently repair CPDs, but still dealt efficiently with (6-4)PPs (1). By contrast, when the human neuroteratoma cell line NT2 was differentiated into hNT neurons, repair was markedly reduced for CPDs, (6-4)PPs (2) and benzo[a]pyrene diol-epoxide adducts (12). This could represent a unique characteristic of the NT2/hNT system, but it may also indicate that various degrees of differentiation result in a progressive decrease in NER capability. CPDs, being weak substrates, would be the first to be affected, whereas (6-4)PPs would still attract the remaining functional NER enzymes.

Repair of UV-Induced Lesions in Vitro.

Cell extracts were prepared according to Manley (13) and incubated with UV-irradiated plasmid and³²P-labeled dCTP, using a nonirradiated plasmid as a control. Repair of UV-induced lesions, mostly cyclobutane-pyrimidine dimers (CPDs) with a small proportion of (6-4)pyrimidine-pyrimidone photoproducts [(6-4)PPs], results in incorporation of the radiolabel into repair patches, which can be detected by autoradiography after electrophoresis in agarose gels (14). Fig. 2A and B, demonstrates that naïve HL60 extracts (filled circles) carry out repair synthesis in the UV-irradiated plasmid much more efficiently than do extracts from HL60 cells differentiated into macrophages with TPA for 48 h (open circles).

Fig. 2.

In vitro repair of UV-induced lesions and cisplatin cross-links. (A and B) Two plasmids of different sizes, irradiated or not with 254-nm UV light, were incubated with “Manley” extract from naïve (filled symbols) or differentiated (open symbols) HL60 cells, in the presence of [³²P]dCTP. The damaged (circles) and undamaged (triangles) plasmids were then resolved by agarose-gel electrophoresis, and the amount of radioactive precursor incorporated into the DNA was quantified with a PhosphorImager. (C and D) A plasmid carrying a single cisplatin GTG intrastrand cross-link close to a ³²P label was incubated with Manley extracts from HL60 (circles) or THP1 (squares) cells, either naïve (filled symbols) or differentiated with TPA for 48 h (open symbols). Excision products spanning the lesion, of 26–35 nucleotides in length, were resolved by acrylamide-gel electrophoresis and quantified with a PhosphorImager. Similar experiments were performed with 4–12 distinct preparations of cell extracts.

This was important to establish because it ruled out the possibility that the decrease in NER observed in vivo could be directly due to a different, perhaps more compact, chromatin structure in terminally differentiated cells. However, this assay yielded a rather substantial background, estimated by the incorporation of ³²P-dCTP into the nonirradiated plasmid (triangles), probably because of nonspecific nicking activities in the extracts. Although low in comparison with the signal, this background rendered the approach inadequate to further dissect the mechanisms of NER attenuation in macrophages.

Excision of Cisplatin Intrastrand Cross-Links in Vitro.

We therefore turned to a more defined system in which a plasmid bearing a single DNA lesion close to a³²P label in the substrate DNA was incubated with cell extract. Excision of the lesion by NER released ³²P-labeled oligonucleotides of 26–35 nucleotides, which could be quantified on sequencing gels (15).

Because CPDs are poor substrates in this assay (16), we selected a different lesion: a cisplatin intrastrand cross-link between the two Gs in a GTG sequence context. Fig. 2 C and D demonstrate that these cross-links are efficiently recognized and excised by both naïve HL60 (filled circles) and naïve THP1 (filled squares) cell extracts, although somewhat more efficiently by the HL60 extract. Extracts from macrophages differentiated from either cell line (open symbols) consistently displayed a much lower excision activity, correlating with the phenotype we had observed for CPDs in vivo and with UV-irradiated plasmids in vitro.

Because the assay measures the excision of the lesion rather than the filling of the resulting gap, we concluded that the decrease in NER activity is due to events in early steps of the NER pathway, related to either lesion recognition or incision.

Complementation with Xeroderma Pigmentosum (XP) Extracts.

To determine whether terminally differentiated macrophages were deficient in an enzyme involved in the early steps of the NER pathway, the cisplatin excision assay was used to perform in vitro complementation tests with extracts prepared from lymphoblast or fibroblast cell lines established from XP patients, who lack one of the seven required NER enzymes. A given XP extract was mixed in various proportions with macrophage extract, keeping the total amount of protein constant. If the macrophages were missing the same enzyme activity as the XP cell line, mixing the extracts should have no effect. If the two extracts were lacking different proteins, mixing them should restore the full set of enzymes required for the excision reaction.

Fig. 3 demonstrates that macrophage extracts are complemented by XP extracts from all known complementation groups: XP-A, XP-B, XP-C, XP-D, XP-F, and XP-G. XP-E extracts are a special case, because XPE is not strictly required for the excision of cisplatin adducts in vitro, although it somewhat speeds up the reaction (17). It can be seen that the excision activity of an XP-E cell extract by itself is substantial, and still slightly improved by addition of XP-B (data not shown) or macrophage extracts; this rules out XPE as a candidate for a missing NER factor. From these results, we conclude that the NER deficiency of macrophages is not because of the absence or inactivity of any “XP” enzyme.

Fig. 3.

In vitro complementation of differentiated macrophage extracts with XP extracts. Extracts from differentiated HL60 cells and NER-deficient extracts from the seven complementation groups of XP were mixed in various proportions and incubated with a plasmid containing a single cisplatin cross-link as described for Fig. 2C. In each panel, the amount of XP extract decreases from left to right as the amount of macrophage extract increases, keeping the total amount of proteins constant. A mix resulting in proficient excision of the cisplatin lesion indicates that differentiated HL60 cells do not lack the corresponding XP enzyme. Because XP-E extracts have an intrinsic activity, they were assayed with and without macrophage extract. N, naïve HL60 extract, included for comparison.

Fractionation of HeLa Cell Extract to Isolate a Complementing Factor.

To identify a factor required for NER, potentially missing or deficient in macrophage extracts, an NER-proficient HeLa cell extract was fractionated and the various fractions were tested for their ability to complement cisplatin excision in macrophage extracts. Preliminary tests with hydrophobicity columns established that an activity that could restore NER in macrophage extracts eluted in only a few fractions, and we verified that these fractions did not contain any NER activity by themselves (data not shown). Similar results were obtained with ion-exchange columns, either anionic or cationic. This strongly suggested to us that macrophages were indeed lacking an activity or a factor required for NER, rather than expressing some hypothetical NER inhibitor.

Fig. 4A summarizes our purification strategy: Manley extracts were prepared from a 20-liter culture of HeLa cells, and fractionated on a DEAE-Sepharose anion-exchange column (Fig. 4B); a complementing activity eluted between 125 and 175 mM NaCl at pH 6.3. Fractions 20–25 were further purified on a butyl-Sepharose hydrophobic column (Fig. 4C). The complementing activity eluted between 480 and 380 mM ammonium sulfate. Fractions 12–17 were further purified by size exclusion on a Sephacryl S-200 gel filtration column, the first 28 fractions of which were analyzed by SDS/PAGE followed with silver staining (Fig. 4D). The NER-complementing activity was detected in fractions 10–15, and copurifying with a protein doublet of 115 and 120 kDa.

Fig. 4.

Purification of an activity complementing differentiated macrophage extracts. (A) Overall purification scheme. (B) Elution of 15 ml of Manley extract on a DEAE-Sepharose column in a 0–400 mM sodium chloride gradient, pH 6.3 (diagonal line). Solid line, protein concentration; bars, cisplatin excision activity of differentiated THP1 extracts complemented with a portion of the relevant fractions (the background activity in differentiated THP1 extracts was subtracted to better visualize the active fractions). (C) Further purification of the active fractions on a butyl-Sepharose column in a 640–0 mM ammonium sulfate gradient (diagonal line). The activity of the resulting fractions was measured as above (bars). (D) Further purification of the active fractions with a sephacryl S-200 gel filtration column, analyzed by SDS/PAGE and silver staining. Lanes denote: D, DEAE fraction #21; B, butyl fraction #13; 9–14, S-200 fractions number 9 through 14, which contained the bulk of the complementing activity.

Fractions 8–15 were pooled and run on a preparative gel. The major, 120-kDa, band was excised for analysis by Edman sequencing, which was unsuccessful, probably because of blocking of the N terminus of the protein. The entire purification was then repeated, both bands were excised, and analysis by mass spectroscopy identified only one protein, with a high score of confidence, the ubiquitin-activating enzyme E1. Hence, the minor band is probably a truncated form of E1, as it is known that this enzyme appears as a doublet in protein gels (18).

The unique E1 enzyme carries out the first step in the ubiquitination sequence: it binds the C terminus of ubiquitin to one of its own cysteines through a thioester bond. This “activated” ubiquitin is then transferred to one of many ubiquitin-conjugating E2 enzymes, where it is again linked via a thioester bond. Finally, one of many ubiquitin E3 ligases transfers ubiquitin from the E2 carrier to the target protein, forming an isopeptide bond with the side chain of a lysine. Further molecules of ubiquitin can then be linked to a lysine in the ubiquitin molecule itself, possibly with the help of an E4 chain-conjugation factor, resulting in the formation of chains of ubiquitin.

Confirmation of E1 Requirement for Proficient NER.

To verify that E1 was indeed responsible for the NER-complementing activity, we supplemented NER-deficient macrophage extracts with various amounts of rabbit E1 enzyme (Fig. 5 A and B, triangles), or of recombinant his-tagged human E1 produced in Escherichia coli (Fig. 5, circles). E1 from either organism restored the excision activity of macrophage extracts in a dose-dependent manner.

Fig. 5.

Confirmation of the involvement of E1. (A and B) Various amounts of ubiquitin-activating enzyme E1 purified from rabbit (triangles) or recombinant His-tagged human E1 (circles) were mixed with differentiated THP1 extract and assayed for cisplatin excision. The activity of a repair-proficient HeLa extract mixed with E1 buffer is shown for comparison (diamond). (C and D) Cisplatin excision was measured with various amounts of extracts from ts85 cells, a thermolabile E1 mutant, grown at 33°C and exposed (open triangles) or not (filled triangles) at 39°C for 16 h. The parental cell line FM3A was similarly exposed to 39°C (squares) to exclude a nonspecific effect of the temperature. Average of three experiments is shown. Error bars are SD. (E) Complementation of differentiated THP1 extract with purified TFIIH. Average of two experiments is shown. Error bars are SEM.

To further confirm the involvement of E1 in activating the NER pathway, we used a mouse mammary cell line, ts85, containing a thermolabile E1 enzyme (Fig. 5 C and D). To exclude a possible toxic effect in the absence of active E1, which might indirectly affect NER, we kept cells at the restrictive temperature of 39°C for only 16 h. Under these conditions, ts85 cells remain fully viable (>95% trypan blue exclusion) and indistinguishable from the parental cell line grown under identical conditions. Extracts from cells grown at the permissive temperature (33°C, filled triangles) were as efficient as those from the parental cell line FM3A (grown at 39°C, open squares) at excising cisplatin GTG cross-links. By contrast, extracts from ts85 cells that were kept at 39°C for 16 h (open triangles) displayed a ∼3-fold reduction in their NER activity, as judged from the difference in slopes of the linear regressions. We also verified that this NER deficiency could be complemented by addition of recombinant human E1 (data not shown).

A previous indication that E1 might be involved with NER was reported by Ikehata et al. (19), who observed that a temporary exposure to the restrictive temperature rendered ts85 cells highly sensitive to UV light, and resulted in an increased rate of mutagenesis. These results are in complete concordance with our findings that E1 is required for activation of the NER pathway. A recent paper by Wang and coworkers (20) has confirmed these findings in a mouse embryo fibroblast cell line also carrying a temperature-sensitive allele of E1. These authors observed a marked deficiency in NER at the global genome level, but at the gene-specific level a persistent strand bias in favor of the transcribed strand indicated that TCR was probably not affected. This is also what we have observed in macrophages, which retain proficient repair of CPDs in transcribed genes despite their low GGR (9, 21). The results we report here, however, reveal another important element of information: that there must be some difference in the amount, the subcellular localization, or the activity of E1 between naïve and differentiated cells, resulting in a deficient NER phenotype.

E1 Phosphorylation in Differentiated Versus Naïve Cells.

It has been reported that the expression of E1 is higher during the active cell cycle and lower in quiescent cells and in terminally differentiated tissues like muscle (22). This corresponds well to the reduced NER activity that we and others have observed in terminally differentiated tissues (reviewed in ref. 1). Thus, there was a possibility that the general pattern of NER attenuation in terminally differentiated cells might result from low levels of E1 in postmitotic cells. This was considered somewhat unlikely, as E1 is thought to be necessary for cell survival, and indeed our Western blot analyses did not detect any difference in E1 levels between naïve and differentiated cells (Fig. 6A).

Fig. 6.

Differences in E1 phosphorylation. (A) Western blot of various batches of HL60 and THP1 extracts, naïve or differentiated, probed with an anti-E1 polyclonal antibody. (B) One microgram of human recombinant E1 was preincubated for 30 min with or without 0.6 mU of potato acid phosphatase before being mixed with THP1-differentiated cell extract and assayed for cisplatin excision. Together: E1 was preincubated alone, and the phosphatase was added together with the extract. (C) E1 was immunoprecipitated from naïve (Upper) or differentiated (Lower) THP1 cells, submitted to 2D gel electrophoresis and analyzed by Western blotting with antiphosphoserine antibodies. Numbers 0 through 4 indicate the number of phosphates on E1. (D) Quantification of the above, together with a similar experiment. Results are displayed as changes between naïve and differentiated extracts in the contribution of the various phosphorylated forms to the total amount of E1.

Another possibility was that E1 could be segregated into a subnuclear compartment in which it could not associate with the E2 factor used by NER. Immunofluorescence studies revealed that the subnuclear distribution of E1 was often strikingly uneven in naïve cells, concentrated at the periphery of the nucleus, whereas in differentiated cells it was diffuse and partly cytoplasmic (data not shown). This suggested to us that E1 can exist in different recognizable forms according to the differentiation status of the cells. However, because E1 was found throughout the nucleus in the differentiated cells, it is unlikely that an E2 enzyme would be unable to access it.

It is known that E1 is phosphorylated in at least four distinct patterns in a cell-cycle dependent manner (23), notably by Cdc2 (24), and it has been suggested that phosphorylation may affect its subcellular localization or modulate its activity by affecting interactions between E1 and a particular E2 enzyme (23). It is thus possible that differences in E1 phosphorylation in differentiated cells could modulate its interaction with an E2 enzyme used by NER. Indeed, preincubating 1 μg of recombinant human E1 (produced in E. coli but properly phosphorylated, data not shown) with 0.6 milliunits of potato acid phosphatase for 30 min at 30°C strongly reduced its ability to complement the THP1 differentiated extract (Fig. 6B). Potato acid phosphatase by itself influenced the excision reaction at higher doses but not at the concentration we selected, as shown in the last bar of Fig. 6B, for which E1 and phosphatase were added simultaneously to the excision reaction.

The next question, of course, was whether there are indeed differences in phosphorylation of E1 between naïve and differentiated cells. We first appraised the global level of E1 phosphorylation by metabolic labeling of naïve and differentiated cells with [³²P]orthophosphate. E1 was immunoprecipitated, quantified by Western blotting, and the E1 band was then quantified for ³²P. The results (data not shown) indicated only a 25% reduction of E1 phosphorylation in differentiated cells, when compared with that in naïve cells. However, E1 is known to be phosphorylated on at least four distinct serines (25), so if only one of these sites were dephosphorylated, a 25% reduction would not be surprising.

To better visualize the various phosphorylation forms of E1, we immunoprecipitated E1 from naïve and differentiated THP1 cells, and analyzed it together with recombinant E1 produced in bacteria (data not shown) by 2D gel electrophoresis, followed by Western blotting with an anti-phosphoserine antibody (Fig. 6C). A 20% decrease in overall phosphorylation was observed upon differentiation (paired t test, P < 0.05), which correlates reasonably well with the value measured by metabolic labeling. Because the anti-phosphoserine antibody should not recognize the “0 phosphate” spot, it should be noted that its quantification by Western blot underestimates the actual amount of protein by at least two fold, judging from Ponceau S staining. Comparison of the distributions of the various phosphorylated forms of E1 in naïve and differentiated extracts clearly demonstrated a shift toward less phosphorylated forms upon differentiation (Fig. 6D), suggesting a reduction in the phosphorylation of at least one site on E1 in the differentiated cells.

E1 is known to be phosphorylated on at least four sites (25), all of which are serines (26). Ser-4 has been identified as a target for Cdc2, both in vitro and in vivo during G₂ phase (24, 25). Another site, the precise location of which is not known, is phosphorylated by Cdc2 in vitro, and by a related kinase active in G₁/S in vivo (24). Ser-835 was also found to be weakly phosphorylated by Cdc2 in vitro, but such a phosphorylated form was not observed in vivo (24). Finally, a fragment of E1 phosphorylated on Ser-46 (which is not part of a Cdc2 consensus site) was identified in a “shotgun” screening for phosphorylated proteins from HeLa cells (27). We do not yet know which phosphorylation site(s) is/are affected by differentiation, but the above experiments demonstrate that appropriate phosphorylation of E1 is required for it to stimulate NER.

How Does Ubiquitination Control NER?

A major question that remains to be answered is how ubiquitination affects NER activity. Ubiquitination is best known for its ability to tag a target protein for degradation by the 26 S proteasome (28). One could thus imagine that an inhibitor of NER (or a dominant inactive form of an NER enzyme) must be degraded before NER can proceed. However, proteasome targeting occurs only when at least four ubiquitin moieties are chained onto the target protein. Furthermore, because ubiquitin contains seven lysines, there can be several different chain types. Some, such as Lys-63 chains, do not trigger degradation, but rather participate in regulating the activity of the target protein (28), as does mono-ubiquitination. Examples include histones (29), several transcription factors (30), and the FANCD2 protein, which is thought to activate repair of double-strand breaks (31). It is thus an attractive hypothesis to suggest that an NER enzyme might be activated by ubiquitination.

Immunoprecipitation of ubiquitinated proteins followed with Western blots (data not shown) indicated that two NER factors are ubiquitinated in the cell lines we used: the TFIIH complex, and XPC. Addition of rat TFIIH to extracts from differentiated THP1 cells significantly increased their cisplatin excision efficiency (Fig. 5E), whereas addition of XPC/HR23B did not improve it (data not shown). We thus believe that the TFIIH complex might contain the NER enzyme(s) activated by ubiquitination for efficient GGR; 2D gel electrophoresis of purified human TFIIH revealed that two subunits were ubiquitinated (data not shown). This could explain why our purification procedure isolated E1 but not TFIIH: the high salt conditions used with some columns would have dissociated the TFIIH complex.

By analogy, these results allow us to answer a previously puzzling question: why would our in vitro assay, performed in whole-cell extracts, require an additional factor other than those required in a cell-free system, as demonstrated by the group of Rick Wood (17)? A possible answer is that the NER factors used in their reconstituted reaction were purified from cell extracts in which activation by ubiquitination had already occurred. Similarly, we were able to complement extracts from all XP groups with macrophage extract (Fig. 3), very likely because the latter contains nonubiquitinated TFIIH, which becomes activated by ubiquitination because of the presence of functional E1 in the XP extracts.

Our results imply that ubiquitination is necessary for TFIIH to perform its role in GGR, but dispensable for TCR (and for initiation of RNA PolII transcription). Such a differential regulation mechanism is not surprising: there are many known mutations in the XPB and XPD subunits of TFIIH, some of which impair only TCR whereas others affect both TCR and GGR (32). This suggests that the contributions of TFIIH to these two NER subpathways are functionally distinct, and may thus be independently regulated.

E1 as a Master Switch During Differentiation?

Even though activation of proteins by ubiquitination is well established, the involvement of E1 in this regulatory mechanism is rather surprising. Until now it was thought that regulation of ubiquitination occurs as a specific pair of E2 and E3 enzymes targets a particular protein. By contrast, there is only one E1 enzyme and its repression is likely to affect the ubiquitination of many distinct proteins. Interestingly, Salvat and coworkers (33) have analyzed the effects of a >90% reduction in E1 in cells carrying a temperature-sensitive mutant of E1 and found that different pathways are affected in widely different manners. This may be due to the different E2 enzymes used by distinct pathways: those using an E2 with a high affinity for E1 would be little affected, whereas those depending on an E2 of lower affinity would be more severely attenuated.

Similarly, our current model (Fig. 7) postulates that E1 is under-phosphorylated in differentiated cells like macrophages, which reduces its interaction with the E2 enzyme responsible for the activation by ubiquitination of an NER factor, most likely TFIIH. As a result, the activity of TFIIH in the GGR subpathway of NER is impaired, whereas its activities in transcription and transcription-coupled repair, apparently not regulated by ubiquitination, remain unaffected. The model also predicts that any other pathway relying on the same E2, or on an E2 sensitive to the same phosphorylation site on E1, will be affected by differentiation. Because there are only a few dozen E2s in the human genome, versus hundreds of E3s and ubiquitinated targets, it is likely that there are many such pathways. One can even speculate that differentiating cells may use E1 as a “master switch” to trigger with a single dephosphorylation event many of the changes in phenotype required during terminal differentiation.

Fig. 7.

Model of NER control by E1 phosphorylation. In naïve cells, E1 is phosphorylated (P) on four sites, leading to ubiquitination (Ub) of an NER enzyme, most likely TFIIH, among others proteins (X, Y, Z). E1 phosphorylation is decreased in differentiated cells, which impairs its association with the E2 enzyme required to ubiquitinate TFIIH (E2.z). A lack in TFIIH ubiquitination decreases its activity in global genomic repair (GGR), but not in transcription-coupled repair (TCR), or in mediating RNA PolII transcription (Tx). The model predicts that ubiquitination of other proteins (e.g., Z) may also be deficient if it relies on the same E2, or on an E2 affected by the same phosphorylation site in E1.

Methods

See supporting information, which is published on the PNAS web site, for detailed procedures. Briefly, repair synthesis assays were performed as described in ref. 17, and cisplatin excision assays were performed as described in ref. 15. For complementation tests, extract from differentiated THP1 cells was mixed in various proportions with extract from XP cells, keeping the amount of proteins constant. For complementation with E1, various amounts of E1 purified from rabbits (Calbiochem EMD Biosciences, San Diego, CA), or recombinant human E1 produced in E. coli (Biomol, Plymouth, PA), were added to extract from differentiated THP1 cells. For E1 purification, Manley extract (13) was prepared from 20 liters of HeLa S3 cells and successively purified on a DEAE Sepharose column, a butyl-Sepharose column, and a sephacryl S-200 column. Aliquots from each fraction from a column were added to extract from differentiated THP1 cells to identify fractions containing a complementing activity, which were then pooled and loaded onto the next column. The final product was excised from a preparative acrylamide gel and identified by mass spectrometry. Immunoprecipitations and Western blotting followed standard procedures, 2D gels were performed with 24-cm Immobilin dry strips, pH 4–7 (Amersham, Piscataway, NJ), and precast 4–15% SDS/PAGE gradient gels (Bio-Rad, Hercules, CA).

Abbreviations

(6-4)PPs

(6-4)pyrimidine–pyrimidone photoproducts

CPDs

cyclobutane pyrimidine dimers

NER

nucleotide excision repair

GGR

global genomic repair

TCR

transcription-coupled repair

XP

xeroderma pigmentosum.