BIOLOGICAL OVERVIEW

The elongation potential of RNA polymerase II has been proposed to be controlled by negative transcription elongation factors and positive transcription elongation factors. All initiated RNA polymerase II molecules enter abortive elongation in which only short transcripts are generated due to the function of negative transcription elongation factors. Upon the action of positive factors such as positive transcription elongation factor b (P-TEFb, a dimer of Cyclin dependent kinase 9 and Cyclin T), the RNA polymerase II molecules overcome the promoter-proximal pausing and premature termination, and as a consequence, enter productive elongation. After the initial switch into a productive mode, the efficiency of elongation can be further increased by other factors to generate long transcripts (Peng, 1998b and references therein).

Phosphorylation of the carboxyl-terminal domain (CTD) of the largest subunit of RNA polymerase II is a regulatory event in transcription. The unphosphorylated RNA polymerase II (IIA) has been found in preinitiation complexes and in early elongation complexes in vitro, whereas the hyperphosphorylated polymerase II (IIO) has been observed in productive elongation. The phosphorylation state of the CTD is controlled by the action of kinases and phosphatases. Drosophila P-TEFb has been identified as a kinase with subunits of 124 and 43 kDa that can phosphorylate the CTD of RNA polymerase II (Marshall, 1996). The kinase activity of P-TEFb is very sensitive to the purine analog DRB (Marshall, 1996). Consistently, the transition from abortive to productive elongation can be inhibited by DRB in vitro and in vivo. The 43-kDa subunit of Drosophila P-TEFb has been cloned and found to be similar to Cdc2, a cyclin-dependent kinase (CDK) (Zhu, 1997; Peng, 1998b and references therein).

Most known CDKs play a role in regulating the cell cycle, but some CDKs have been found to be involved in other cellular events. A CDK can be activated by binding of a cyclin and phosphorylation of a conserved threonine residue in the T-loop of the catalytic subunit. Conversely, a CDK/cyclin complex can be inactivated by phosphorylation of a threonine residue and a tyrosine residue near the ATP binding site or by binding of a family of small proteins termed CKIs. The catalytic subunits are well conserved in the CDK family, but the cyclins are not conserved except for the cyclin box that is predicted as a helix-rich structure. Drosophila P-TEFb has been shown to be composed of a CDK/cyclin pair (Peng, 1998b and references therein).

The Drosophila P-TEFb possesses a CTD kinase activity that is very sensitive to DRB (Marshall, 1996). To examine the kinase activity of recombinate Drosophila P-TEFb, Drosophila RNA polymerase II was incubated with increasing amounts of either native P-TEFb or recombinant P-TEFb in the presence of ³²P labelled ATP for 5 min. The reactions were analyzed by 6%-15% SDS-PAGE. The protein gel was silver-stained and then subjected to autoradiography. As expected, increasing the amount of native Drosophila P-TEFb increases the fraction of the hyperphosphorylated IIO form of the large subunit of RNA polymerase II as well as the total amount of label incorporated into the IIO form. The activity of recombinant P-TEFb was indistinguishable from that of native Drosophila P-TEFb. The sensitivity of recombinant P-TEFb to DRB was compared with that of native P-TEFb using a similar assay. The radioactivity incorporated into the large subunit of RNA polymerase II was quantitated, normalized to the no DRB point (100%), and plotted. The 50% inhibition point of recombinant P-TEFb (0.5 µM) is similar to that of P-TEFb (Peng, 1998a).

The function of recombinant P-TEFb in elongation control was analysed. To do this it was first necessary to generate a P-TEFb-dependent transcription system. P-TEFb allows the generation of long DRB-sensitive transcripts (Marshall, 1996), and in a human transcription system depletion of P-TEFb eliminates DRB-sensitive transcription (Zhu, 1997). To deplete Drosophila P-TEFb from Kc cell nuclear extract (KcN), antibodies to Drosophila P-TEFb were produced. A glutathione S-transferase fusion protein containing the carboxyl-terminal half of the cyclin T subunit was used to immunize rabbits. A Western blot indicated that the resulting antiserum recognized both the native and recombinant P-TEFb. Moreover, the antibodies in the antiserum were able to deplete the cyclin subunit of Drosophila P-TEFb (Peng, 1998a).

The depleted KcN was characterized and then used to compare the activity of recombinant P-TEFb to the native factor. The starting extract as well as a preimmune depleted extract both exhibit DRB-sensitive runoff transcripts indicative of the function of P-TEFb. Depletion of P-TEFb decreases these runoff transcripts. As was found in the human transcription system (Zhu, 1997), depletion of P-TEFb does not deplete other factors needed to generate DRB-sensitive transcription. Addition of increasing amounts of native P-TEFb to the depleted extract restores the appearance of runoff transcripts to levels higher than that observed with the low level of endogenous P-TEFb in the starting extract. As expected, the runoff transcripts stimulated by P-TEFb are sensitive to DRB. Incomplete inhibition by DRB is likely due to the saturating amount of P-TEFb being added. The recombinant P-TEFb gives identical results to native Drosophila P-TEFb, demonstrating that it is fully functional in transcription (Peng, 1998a).

To begin to elucidate the role of the cyclin subunit and to determine the requirement of the kinase activity, recombinant proteins containing various mutations were generated. The proteins were produced in baculovirus-infected Sf9 cells and then purified. Recombinant proteins include a kinase knockout, CDK9 alone, and two CDK9/cyclin pairs in which portions of the carboxyl-terminal region of the cyclin subunit were deleted. In each case, the CDK9 subunit was His tagged, and as long as the cyclin subunit contained an intact cyclin box, it associated strongly with the kinase subunit. This was also found with another construct that contained only amino acids 1-348 of the cyclin subunit (Peng, 1998a).

The recombinant proteins were examined for their ability to phosphorylate the CTD of RNA polymerase II as well as their ability to function during transcription. CDK9 alone has no activity in either assay, demonstrating that, indeed, its activity is cyclin dependent. The kinase knockout mutant also has no activity in either assay, demonstrating that the function of P-TEFb is through its ability to carry out phosphorylation. Deletion of the unique carboxyl-terminal domain of cyclin T causes a dramatic reduction in both activities measured. The two cyclin T truncation mutants gave only 10%-20% of the kinase activity seen with the intact protein, and this reduction was quantitatively mirrored in the transcription assay. An unusual feature of the phosphorylation of RNA polymerase II by the two truncation mutants is that the shift from IIA to IIO does not occur as readily. Drosophila P-TEFb has been shown to preferentially phosphorylate a CTD that has already been phosphorylated (Marshall, 1996), and at low levels of the kinase this leads to the hyperphosphorylation of a subset of the polymerases. Evidently, the carboxyl-terminal region of cyclin T is responsible for this preferential phosphorylation of the partially phosphorylated polymerase molecules (Peng, 1998a).

To better compare the transcriptional activity of the two cyclin truncation mutants with intact DmP-TEFb, a wider range of kinase concentrations was titrated. It was especially interesting to determine if higher levels of the truncation mutants could compensate for their lower kinase activity or if they were in some other way defective. Intact P-TEFb functioned as expected, giving rise to a high level of runoff transcripts. Comparison of the amount of runoff at intermediate levels of intact P-TEFb (4x to 16x) indicated that addition of 2-fold more kinase has a greater than 2-fold effect. A different result was obtained with the two truncation mutants. Neither showed the effect seen at intermediate levels of intact P-TEFb. Even though very high levels were used, the effect of both truncation mutants saturates before reaching the high level of runoff seen with the intact protein. It is concluded that the carboxyl-terminal region of cyclin T is important for the function of P-TEFb in transcription (Peng, 1998a).

P-TEFb must interact at least briefly with the polymerase to carry out phosphorylation, and this transient interaction may be enhanced by the positive charge of the carboxyl-terminal region of cyclin T. This could explain why P-TEFb preferentially phosphorylates a CTD that was already partially phosphorylated (Marshall, 1996). When the CTD is partially phosphorylated, it contains more negative charge and might interact more strongly with the positively charged region of cyclin T. Removal of this region decreases the kinase activity (Peng, 1998a).

The kinase activity of TFIIH has been implicated in CTD phosphorylation and in elongation control. Therefore, it was of interest to determine if the CDK-activating kinase associated with TFIIH could affect the CTD kinase activity of P-TEFb. Kinase reactions were set up with RNA polymerase II as substrate, limiting amounts of P-TEFb or lower levels of Drosophila TFIIH. As expected, CTD phosphorylation by P-TEFb but not TFIIH was sensitive to DRB. In reactions containing both kinases, the signals were apparently the sum of the two activities. The phosphorylation of cyclin T and CDK9 was also unaffected by the presence of TFIIH. This suggests that neither kinase has a dramatic effect on the other (Peng, 1998a).

These results suggest that Drosophila P-TEFb is not part of a larger complex and does not stably associate with RNA polymerase II or the elongation complex. Depletion of P-TEFb even under low salt conditions does not remove other required factors because addition of pure DmP-TEFb is able to restore DRB-sensitive transcription. It is possible that other required factors are in excess over the P-TEFb. This however is thought not to be the case because when the kinase knockout mutant was added in more than 10-fold excess over the wild type kinase, no reduction in kinase activity or DRB-sensitive transcription was observed. Consistent with this, no more than one chromatographic form of P-TEFb has been detected (Peng, 1998a).

The timing of phosphorylation of the polymerase in an early elongation complex is important because of the termination activity of factor 2 (Xie, 1996). Essentially, there is a functional competition between factor 2 and P-TEFb. During a short period of time after initiation, P-TEFb must phosphorylate the CTD to an appropriate extent to cause the transition into productive elongation. Otherwise factor 2 will cause premature termination. This may be why the cyclin mutants lacking the carboxyl-terminal region are able to generate only low levels of runoff transcripts. The reduced ability of the cyclin mutants to recognize partially phosphorylated RNA polymerase II in an early elongation complex would lead to a reduced rate of phosphorylation and, therefore, ultimately a reduced number of hyperphosphorylated polymerases that generate runoff (Peng, 1998a).

As a cyclin-dependent kinase, P-TEFb is likely to be activated by phosphorylation of the catalytic subunit. The TFIIH-associated kinase is a CDK-activating kinase termed CDK7 (see Drosophila Cdk7) that has been shown to activate CDC2, CDK2, and CDK4. TFIIH has also been shown to function in transcription elongation. Although it is possible that TFIIH plays a role in activating P-TEFb, the results presented here do not support this hypothesis. Using pure RNA polymerase II as a substrate, TFIIH has no effect on the kinase activity of P-TEFb. It is possible that in the context of transcription with other factors involved, TFIIH might play a role in activating P-TEFb. It is also possible that P-TEFb purified from eucaryotic cells is already activated. Further work in vitro and in vivo will be required to understand how P-TEFb activity is regulated (Peng, 1998a).

The super elongation complex drives neural stem cell fate commitment

Asymmetric stem cell division establishes an initial difference between a stem cell and its differentiating sibling, critical for maintaining homeostasis and preventing carcinogenesis. Yet the mechanisms that consolidate and lock in such initial fate bias remain obscure. This study used Drosophila neuroblasts to demonstrate that the super elongation complex (SEC) acts as an intrinsic amplifier to drive cell fate commitment. SEC is highly expressed in neuroblasts, where it promotes self-renewal by physically associating with Notch transcription activation complex and enhancing HES (hairy and E(spl)) transcription. HES in turn upregulates SEC activity, forming an unexpected self-reinforcing feedback loop with SEC. SEC inactivation leads to neuroblast loss, whereas its forced activation results in neural progenitor dedifferentiation and tumorigenesis. These studies unveil an SEC-mediated intracellular amplifier mechanism in ensuring robustness and precision in stem cell fate commitment and provide mechanistic explanation for the highly frequent association of SEC overactivation with human cancers (Liu, 2017).

Both normal development and tissue homeostasis rely on the remarkable capacity of stem cells to divide asymmetrically, simultaneously generating one identical stem cell and one differentiating progeny. Extensive studies have unveiled how extrinsic niche signals and intrinsic cell polarity cues ensure proper orientation of mitotic spindle and, hence, asymmetric division of stem cells. However, it remains unclear whether the initial fate bias, established by unequal exposure to niche signals or differential partitioning of cell fate determinants, can be immediately and automatically consolidated and stabilized into distinct and irreversible cell fate outcomes. In fact, in vivo timelapse imaging of the developing zebrafish hindbrain using the Notch activity reporter showed that, immediately after the asymmetric division of a radial glia progenitor, Notch activity is not noticeably biased in the paired daughter cells. Instead, the differential Notch activity in the pair of daughter cells only gradually increases afterward, over a time span of 3-8 hr, indicating the existence of a progressive and tightly regulated transition phase between the initial cell fate decision and the ultimate cell fate commitment. Stem cells and progenitors, especially the fast-cycling ones, face the daunting challenges of ensuring timely, precise, and robust cell fate determination in every cell cycle and are likely to achieve so through rapid amplification of the initial small fate bias upon their asymmetric division. In electronics, a device called an amplifier magnifies a small input signal to a large output signal until it reaches a desired level. Conceivably, a similar 'amplifier' mechanism could be employed in the stem cells or progenitors to accelerate the transition phase and drive cell fate commitment. Dysregulation of such an amplifier could cause an imbalance between self-renewal and differentiation, resulting in impaired tissue homeostasis. However, the regulatory modules governing the transition phase from stem cell fate decision to fate commitment, especially the identity and control of this putative 'amplifier,' remain largely unexplored (Liu, 2017).

Drosophila type II neural stem cells (NSCs), known as neuroblasts (NBs), provide an excellent model system for studying stem cell fate commitment. Firstly, distinct from type I NB lineages, type II NB lineages contain transit-amplifying cells called intermediate neural progenitors (INPs), similar to mammalian NSC lineages in both functional and molecular criteria, yet with much simpler anatomy and lineage composition. Each type II NB undergoes stereotypic, self-renewing divisions to produce immature INPs, which, upon maturation, undergo a few rounds of asymmetric, self-renewing divisions to give rise to ganglion mother cells (GMCs) that subsequently generate post-mitotic neurons or glia. Secondly, the identity of each cell type in the NB lineages can be unambiguously determined by a combination of cell fate makers as well as by their geological positions within the lineages. Thirdly, the molecular mechanisms underlying initial NB versus INP fate decision are well understood. Unidirectional Notch signaling is both necessary and sufficient to promote type II NB self-renewal. At each division, type II NBs asymmetrically segregate differentiation-promoting determinants, such as Notch antagonist Numb, into immature INPs. As a consequence, Notch pathway effector HES (hairy and E(spl)) genes, such as E(spl)mγ, are highly expressed in NBs but not in immature INPs. HES genes, encoding basic helix-loop-helix (bHLH) transcription factors, are crucial for promoting NB self-renewal. Importantly, numb mutant immature INPs fail to complete maturation but instead revert fate back into NBs and result in tumorigenesis, indicating that the asymmetric segregation of Numb protein is critical for establishing the initial fate bias between a type II NB and its sibling INP. However, whether such initial bias is sufficient to confer differential Notch activity and achieve definitive fate commitment is currently unclear. Lastly, type II NBs undergo fast cell divisions, dividing every 2 hr, placing them under huge pressure to timely yet precisely achieve differential fate outcomes. Therefore, within type II NB lineages, a regulatory module that drives robust cell fate commitment is likely to exist, plausibly with high activity (Liu, 2017).

Overactivation of Notch signaling leads to immature INP dedifferentiation and tumorigenesis, providing a sensitized background for identifying factors pivotal for NB or INP fate commitment. In such a genetic background a genome-wide RNAi-based screen was carried out for genes whose downregulation specifically suppresses the supernumerary NB phenotype induced by Notch overactivation, and subunits of the super elongation complex (SEC) were identified. The SEC is composed of the elongation factor ELL (eleven-nineteen lysine-rich leukemia) 1/2/3, the flexible scaffolding protein AFF (AF4/FMR2 family) 1/2/3/4, the ELL-associated factor EAF1/2, eleven-nineteen leukemia (ENL)/AF9, as well as the Pol II elongation factor P-TEFb consisting of cyclin T (CycT) and cyclin-dependent kinase 9 (CDK9). The screen identified all subunits of SEC except EAF and ENL/AF9, suggesting that SEC interplays with Notch signaling in promoting NB self renewal. The SEC subunits were originally identified as frequent translocation partners of MLL (mixed-lineage leukemia) in inducing leukemogenesis, and play key roles in c-Myc-dependent carcinogenesis and HIV viral DNA transcription. Previous studies demonstrated that SEC executes its functions by inducing rapid gene transcription, mainly through phosphorylating RNA polymerase II (Pol II) C-terminal domain and releasing it from promoter-proximal pausing (Liu, 2017).

This study shows that the SEC is specifically expressed in Drosophila NBs, where it acts as an amplifier to drive type II NB fate commitment. SEC exerts its function by physically associating with Notch transcription activation complex to stimulate dHES (Drosophila HES) transcription. dHES in turn promotes SEC expression/activity. Thus, driven by a self-reinforcing feedback loop between SEC and Notch signaling, an initial small bias of Notch activity between an NB and its sibling INP is rapidly amplified and consolidated into robust and irreversible fate commitment (Liu, 2017).

Is the establishment of an initial fate bias at the end of stem cell asymmetric division truly the end, or just the beginning of the end? The current findings revealed that a progressive and tightly controlled transition phase exists between the initial fate decision and the final definitive fate commitment. The results identified the evolutionarily conserved SEC as a crucial intrinsic amplifier, accelerating this previously overlooked fate transition phase and ensuring NSC fate commitment in Drosophila type II NB lineages. Inactivation of SEC prevents the self-reinforcing feedback loop between SEC and Notch signaling from running, resulting in NBs with ambiguous stem cell identity and ultimate fate loss. Conversely, ectopic overactivation of SEC initiates and sustains this positive feedback loop within progenitors, driving dedifferentiation and tumorigenesis. It is interesting to note that, as one of the most active P-TEFb-containing complexes in controlling rapid transcriptional induction in response to dynamic developmental or environmental cues, SEC is particularly suitable for being an amplifier in driving timely cell fate commitment. Since fast-cycling stem cells are under huge pressure to achieve robust fate determination in every cell cycle, it is not surprising that they employ SEC as a regulatory component to induce immediate activation of master fate-specifying genes that in turn form a self-amplifying loop with SEC to rapidly magnify the initial fate bias and ensure prompt fate commitment (Liu, 2017).

Such an intracellular amplifier mechanism revealed by these studies might complement the well-established intercellular lateral inhibition mechanism and represent a general, cell-autonomous paradigm to ensure robustness and precision in binary cell fate commitment. Lateral inhibition is a widely used mechanism underlying cell fate diversification, whereby unidirectional Notch signaling utilizes intercellular feedback loops to amplify an initial small difference between adjacent daughter cells, and eventually confers distinct cell fates. Lateral inhibition relies on intercellular interactions between adjacent cells. This study proposes a model whereby an intracellular amplifier mechanism may also diversify cell fates (Liu, 2017).

The intracellular amplifier and intercellular lateral inhibition mechanisms, both acting through feedback loops, are not mutually exclusive. Instead, they are complementary to each other and can be used concomitantly or sequentially to achieve differential fate outcomes in a timely, precise, and robust manner. An amplifier design often employs negative feedback to prevent excessive amplification. In this study the dHES-Earmuff/Brahma-SEC double-negative regulatory mechanism that this study has revealed in NBs might also operate in neural progenitors, where the Erm/Brm complex could serve as a crucial 'brake' to prevent the Notch-SEC-Notch self-reinforcing positive feedback loop from starting (Liu, 2017).

Notch signaling plays a conserved role during vertebrate embryonic neurogenesis in maintaining the undifferentiated status of NSCs. Intriguingly, expression of HES-1, a primary target of Notch pathway in mammalian neural development, oscillates every 2 hr. It has been proposed that oscillations in HES-1 expression drive fluctuations in gene expression, resulting in differential expressions between neighboring cells, which needs be further amplified to confer distinct cell fates. How such an amplification step is triggered and modulated remains elusive. Given that SEC is highly conserved in mammals, it is interesting to speculate that a similar amplifier mechanism is employed to ensure mammalian NSC fate commitment. Whether SEC interplays with Notch signaling to drive cell fate commitment in other stem cell lineages also warrants future investigation (Liu, 2017).

Despite extensive studies elucidating how SEC regulates transcription elongation, the in vivo function of SEC in normal development and physiology remains enigmatic. The current results indicate that SEC is highly expressed in Drosophila NSCs, where it is recruited by the Notch transcription activation complex to stimulate the transcription of dHES genes and promote self-renewing fate. Interestingly, the dHES genes in fly larval brain NB lineages are non-pausing genes, raising the possibility that SEC promotes the transcriptional activation of dHES in the absence of paused Pol II. Consistent with this view, recent studies have demonstrated that the rapid transcriptional induction of some nonpausing genes, such as Cyp26a1 in human embryonic stem cells and a subset of pre-cellular genes in early Drosophila embryos, depends on SEC activity and Pol II occupancy. The current findings that SEC physically and genetically interplays with the dCSL-NICD-MAM transcription activation complex to activate dHES transcription thus provide a unique physiological context for elucidating the detailed molecular mechanisms underlying transcriptional induction of non-pausing genes by SEC (Liu, 2017).

The upstream signals and molecular mechanisms controlling SEC activity in normal development or physiology are just unfolding. It has been previously shown that the activity of SEC could be regulated by modulating the kinase activity of CDK9, the catalytic subunit of SEC. The results unveil a new and unexpected mechanism underlying the control of SEC: the Notch-HES axis spatially restricts SEC activity within NSCs by cell-autonomously promoting the protein abundance of dAFF and dELL, two regulatory subunits of SEC. Consistently, overactivation of Notch signaling led to dedifferentiation of immature INPs, in which the expression levels of dAFF/dELL and, hence, the activity of SEC evidently increase (Liu, 2017).

Dysregulation of the SEC subunits is frequently associated with various human cancers including leukemia and glioblastoma. However, in most cases, whether SEC acts as a cancer driver or passenger is unclear. Furthermore, whether SEC subunits exert their oncogenic or tumor suppressive roles as a component of SEC or independent of SEC remains poorly understood. Intriguingly, the results show that overexpression of dELL and dAFF but not either alone induces a dramatic surge of dHES expression in immature INPs and causes progenitorderived tumor. These findings strongly suggest that, in NSC lineages, SEC drives tumorigenesis as an integral complex and exerts its oncogenic function in a dose-dependent manner. Supporting this view, the kinase activity of CDK9 is essential for dELL/dAFF-induced tumorigenesis. It will be interesting to investigate whether upregulation of dELL/dAFF abundance is sufficient to induce carcinogenesis in other biological contexts. Tne findings highlighting the self-reinforcing feedback loop between SEC and Notch signaling in driving tumorigenesis further suggest that CDK9 inhibitors could be pursued as an effective therapy for Notch overactivation-induced tumors (Liu, 2017).

Cyclin-dependent kinase 9 is required for the survival of adult Drosophila melanogaster glia

Neuronal and glial progenitor cells exist in the adult Drosophila brain. The primarily glial progenitor cells rely on a microRNA, mir-31a, to inhibit the expression of a predicted E3 ubiquitin ligase, CG16947. Erroneous inheritance of CG16947 by the progeny when the neural progenitor cell divides leads to death of the progeny, however how CG16947 achieves glial cell death is unknown. This study has identified the interacting partner of CG16947 to be Cdk9. Reduction of cdk9 expression in glia causes glial loss; highlighting the importance of cdk9 in mediating the survival of glia. Further, glial loss observed in mir-31a mutants was prevented with adult-specific expression of cdk9 in glia. Biochemical evidence is provided that the binding of CG16947 to Cdk9 causes its degradation. Taken together, this data shows that cdk9 plays a role in the survival of adult glia in the Drosophila brain. Thus, a fine balance exists between mir-31a and CG16947 expression in the progenitor cells that in turn regulates the levels of cdk9 in the progeny. This serves to allow the progenitor cells to regulate the number of glia in the adult brain (Foo, 2017).

CG16947 is expressed in the progenitor cells and mir-31a serves to limit its expression in the progenitor cells. When this interaction between the microRNA and the CG16947 transcript is abolished, CG16947 is aberrantly translated at higher levels, leading to increased inheritance by the progeny (Foo, 2017).

This paper shows that depletion of cdk9 alone in adult glia can reduce the number of glia in the adult brain. Thus, cdk9 is necessary for adult glial survival. Additionally, this study shows that in the condition where there are fewer glia in the brain; as in the mir-31a mutants, the adult-specific overexpression of cdk9 in glia prevents glial loss. This shows that it is loss of cdk9 in adult glia that is the likely reason for the glial loss observed in mir-31a mutants (Foo, 2017).

Using co-immunoprecipitation, evidence is provided that CG16947 binds to and interacts directly with cdk9 leading to its degradation. Taken together, the genetic results and the biochemistry suggest that the subsequent ubiquitination and degradation of cdk9 due to excessive inheritance of CG16947 by the progeny in mir-31a mutants, is the likely cause for glial cell death in these mutants. This study does not exclude the possibility that cdk9 is degraded by other mechanisms, but do show that cdk9 is necessary for glial cell survival. Given the role of cdk9 in controlling RNA-polymerase II-mediated transcription, it is possible that cdk9 works by facilitating the transcription of pro-survival genes (Foo, 2017).

As an increase in glial number was not observed in the brain when cdk9 was overexpressed in adult glia, it suggests that cdk9 is not sufficient to generate more glial cells unless in aberrant conditions where glia number is reduced, as is in the case of the mir-31a mutants (Foo, 2017).

Knocking down of CG16947 in progenitor cells did not lead to an increase in glia numbers in the brain. Thus, it suggests that glial numbers are very tightly regulated in the adult brain. The rational being that since high levels of CG16947 inherited by glia from the progenitor cells is detrimental to the survival of glia then, in the wildtype setting, if more glia than necessary are made, then knocking down of CG16947 in progenitor cells, would lead to increased survival of glia and more glia observed. But as more glia is not observed in either setting, when CG16947 is knocked down or cdk9 is overexpressed, it suggests that the number of glia that are made is very tightly regulated and no more than necessary glia are made by the progenitors in the adult central brain of flies (Foo, 2017).

In mammals, there are several examples of the number of glia in the mature brain being tightly regulated by contact with a target. For instance, mammalian astrocytes are thought to be matched to blood vessels, mediated by the limited secretion of trophic factors from the vasculature. A similar reliance of oligodendrocytes on their target axons for survival has also been observed. This study shows that the number of glia can be in part, regulated from the progenitor cell itself, through controlled expression of CG16947, rather than reliance on the target of the glial cell. These results do not exclude the possibility that glial number in the Drosophila adult brain is regulated by another mechanism (Foo, 2017).

The mechanism by which mir-31a expression in the progenitor cell is regulated and in turn CG16947 expression is regulated, has yet to be uncovered, however, this study provides evidence that misregulation of this fine balance between mir-31a with CG16947 in the progenitor cells and subsequent excessive degradation of cdk9 in the progeny can alter the number of glia in the adult Drosophila brain. Thus providing evidence for a mechanism by which the progenitor cell itself can regulate the number of progeny that are generated. This study does not exclude the possibility that cdk9 can be regulated in a mir-31a-CG16947-independent manner, but illustrates one way in which cdk9 expression can be regulated in glia (Foo, 2017).

TFIID Enables RNA Polymerase II Promoter-Proximal Pausing

RNA polymerase II (RNAPII) transcription is governed by the pre-initiation complex (PIC), which contains TFIIA, TFIIB, TFIID, TFIIE, TFIIF, TFIIH, RNAPII, and Mediator. After initiation, RNAPII enzymes pause after transcribing less than 100 bases; precisely how RNAPII pausing is enforced and regulated remains unclear. To address specific mechanistic questions, human RNAPII promoter-proximal pausing was reconstituted in vitro, entirely with purified factors (no extracts). As expected, NELF and DSIF increased pausing, and P-TEFb promoted pause release. Unexpectedly, the PIC alone was sufficient to reconstitute pausing, suggesting RNAPII pausing is an inherent PIC function. In agreement, pausing was lost upon replacement of the TFIID complex with TATA-binding protein (TBP), and PRO-seq experiments revealed widespread disruption of RNAPII pausing upon acute depletion (t = 60 min) of TFIID subunits in human or Drosophila cells. These results establish a TFIID requirement for RNAPII pausing and suggest pause regulatory factors may function directly or indirectly through TFIID (Fant, 2020).

RNA polymerase II (RNAPII) transcribes all protein-coding and many non-coding RNAs in the human genome. RNAPII transcription initiation occurs within the pre-initiation complex (PIC), which contains TFIIA, TFIIB, TFIID, TFIIE, TFIIF, TFIIH, RNAPII, and Mediator. After initiation, RNAPII enzymes typically pause after transcribing 20-80 bases, and paused polymerases represent a common regulatory intermediate. Accordingly, paused RNAPII has been implicated in enhancer function, development and homeostasis, and diseases ranging from cancer to viral pathogenesis. Precisely how RNAPII promoter-proximal pausing is enforced and regulated remains unclear; however, protein complexes, such as NELF and DSIF, increase pausing, whereas the activity of CDK9 (P-TEFb complex) correlates with pause release (Fant, 2020).

Although much has been learned about RNAPII promoter-proximal pausing and its regulation, the underlying molecular mechanisms remain enigmatic. One reason for this is the complexity of the human RNAPII transcription machinery, which includes the ∼4.0 MDa PIC and many additional regulatory factors. Another underlying reason is that much current understanding derives from cell-based assays, which are indispensable but cannot reliably address mechanistic questions. For instance, factor knockdowns or knockouts cause unintended secondary effects and the factors and biochemicals present at each gene in a population of cells cannot possibly be defined. In vitro assays can overcome such limitations, but these have typically involved nuclear extracts, which contain a similarly undefined mix of proteins, nucleic acids, and biochemicals. To circumvent these issues, this study sought to reconstitute RNAPII promoter-proximal pausing entirely from purified human factors (no extracts). Success with this task enabled addressing some basic mechanistic questions and opens the door for future studies to better define the contribution of specific factors in RNAPII promoter-proximal pause regulation (Fant, 2020).

Structural data indicate that TFIID lobe C subunits TAF1 (see Drosophila Taf250) and TAF2 bind promoter DNA downstream of the TSS (Louder, 2016; Patel, 2018). Past studies revealed that insertion of 10-bp DNA at the +15 site relative to the TSS disrupted RNAPII pausing at the HSP70 gene in Drosophila S2 cells (Kwak, 2013). This led to a 'complex interaction' model for pausing, in which a promoter-bound factor(s) establishes an interaction (directly or indirectly) with the paused RNAPII complex. In agreement with this model, a TFIID requirement was observed for RNAPII promoter-proximal pausing in vitro, which is further supported by PRO-seq data in TAF-depleted human and Drosophila S2 cells. Additional evidence for TFIID-dependent regulation of RNAPII pausing derives from correlations among paused genes and DNA sequence elements bound by TFIID. Defects in TFIID function are linked to numerous diseases, including cancer and neurodegenerative disorders. Its requirement for RNAPII promoter-proximal pause regulation may underlie these and other biological functions (Fant, 2020).

Biochemical reconstitution of RNAPII promoter-proximal pausing provides a level of mechanistic control that is simply not possible with cell-based assays; consequently, it was discovered that RNAPII pausing is an inherent property of the human PIC and that TFIID is a key PIC factor that establishes pausing. The results also reveal NELF, DSIF, and P-TEFb as auxiliary factors that, although not required for pausing, enable robust regulation of this common transcriptional intermediate state. Time course experiments indicated that polymerases in the paused region remained active and generated elongated transcripts over time. Experiments with P-TEFb showed enhanced release of paused intermediates, providing further evidence that polymerases in the paused region were active and competent for elongation. However, some transcripts remained in the pause region after the 10-min reactions, even with added P-TEFb. This result is also consistent with current models that invoke alternative outcomes for promoter-proximal paused RNAPII, including premature termination, arrest, or a more stable paused intermediate. Addressing the mechanisms and factors that regulate these distinct outcomes could be explored in future studies (Fant, 2020).

Despite its advantages, the reconstituted in vitro transcription assay does not match the complexity of regulatory inputs that converge upon active promoters in a living cell. To test the TFIID requirement for promoter-proximal pausing in cells, it was possible to rapidly deplete TFIID lobe C subunits TAF1 and TAF2 using Trim-Away, and genome-wide changes in nascent transcription were assessed with PRO-seq. Consistent with the in vitro data, global transcription increased at protein-coding genes upon TAF1/2 knockdown, with evidence for enhanced pause release. PRO-seq reads increased at 5' ends and downstream of promoter-proximal pause sites at thousands of genes in TAF1/2-depleted cells. These data are consistent with increased pause release and increased re-initiation, two processes that are coupled in metazoan cells. Unexpectedly, however, increased pause release did not yield similar genome-wide increases in gene body reads. Instead, the PRO-seq data revealed a sharp reduction in reads downstream of promoter-proximal pause sites, at around +300 from the TSS in both human and Drosophila cells. These results implicate additional regulatory mechanisms, downstream of the pause site, that may terminate or arrest RNAPII. Although future studies are needed to identify the factors involved, it is noted that the Integrator complex was recently shown to cleave nascent transcripts downstream of pause sites at hundreds of genes in Drosophila cells (Tatomer, 2019). Because promoter-proximal pausing helps ensure proper capping of transcripts at their 5' ends, downstream regulatory mechanisms may become important when RNAPII promoter-proximal pausing is disrupted (Fant, 2020).

A TFIID requirement for RNAPII promoter-proximal pausing implies that other pause regulatory factors may function directly or indirectly through TFIID. Although additional mechanistic aspects remain to be addressed, it is notable that pause regulatory factors, including P-TEFb and MYC, interact (directly or indirectly) with TFIID; moreover, TFIID is conformationally flexible and likely undergoes structural reorganization during RNAPII transcription initiation and pause release. Such structural transitions may contribute to TFIID-dependent regulation of RNAPII pausing. Whereas nucleosomes likely affect promoter-proximal pausing, they are not required, based upon our results and data in Drosophila and mammalian systems. TFIID possesses multiple domains that bind chromatin marks associated with transcriptionally active loci, including H3K4me3, which suggests TFIID function is regulated in part through epigenetic mechanisms. Future studies should help establish whether specific chromatin marks contribute to TFIID-dependent regulation of RNAPII pausing, potentially by affecting TFIID promoter occupancy or by impacting TFIID structure and function (Fant, 2020).

PROTEIN STRUCTURE

Amino Acids - 1097

Structural Domains

A cDNA encoding the small subunit of the Drosophila factor has been cloned. The two subunits of purified P-TEFb were separated by gel electrophoresis, and the small subunit was excised and subjected to protein sequencing. The peptide sequence information was used in the cloning of full-length cDNA. The deduced amino acid sequence identified the small subunit of Drosophila P-TEFb as a member of the Cdc2-like cyclin dependent kinase family with >40% identity to Schizosaccharomyces pombe Cdc2. A search of the protein database revealed a human protein, PITALRE, that exhibits 72% identity and 83% similarity to the Drosophila protein. The high level of sequence similarity indicates that PITALRE is a potential homolog of the small subunit of Drosophila P-TEFb and therefore may be a component of human P-TEFb. Two kinases from Saccharomyces cerevisiae, SGV1 and CTK1, each share 43% identity with PITALRE and the small subunit of Drosophila P-TEFb. Although sequence similarity does not allow the prediction of a potential yeast homolog, CTK1 has recently been demonstrated to increase the elongation efficiency of RNA polymerase II (Zhu, 1997 and references therein).

date revised: 2 September 2022

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.