Plans within plans: post-transcriptional regulation governs macrophage responses

Abstract

Macrophages are sentinels and first responders of the innate immune system. By sensing danger signals, they initiate and amplify inflammatory and regenerative cascades to control appropriate responses to pathogens and tissue damage. Transcriptional activation of macrophage gene expression has been studied extensively, but macrophage responses also rely on regulation of mRNAs following transcription. In this review, we discuss mechanisms of post-transcriptional regulation that alter macrophage gene expression programs in profound and sometimes surprising ways. We explore how these control nodes are layered to form complex and dynamic circuits, discuss their role in disease, and conclude by outlining opportunities for future study of post-transcriptional regulation in macrophages.

Macrophages cannot run on transcription alone

Macrophages are innate immune cells of the myeloid lineage that act as the first line of defense against pathogens and key regulators of tissue homeostasis [1]. Human macrophages either exist in their mature form within tissues or are differentiated from circulating monocytes in response to injury or infection. Facing a complex milieu of pathogen- and host-derived signals, macrophages possess a remarkable capacity to initiate diverse gene expression networks based on the molecular cues detected (e.g. pathogens, nutrient availability, cytokine production, and tissue damage markers). To maintain healthy tissues, macrophages need to activate the right program at the right time. When macrophages produce the wrong signals, disease can be triggered or worsened, as is the case in chronic diabetic wounds and solid tumors [2,3].

Stimulation of classical macrophage gene programs such as the JAK-STAT (see Glossary) and NF-κB pathways have been intensely studied, giving rise to detailed models of the transcription factor cascades instigating de novo transcription of inflammatory and antimicrobial genes [4]. Global proteomic and transcriptomic methods have enabled the mapping of kinases and chromatin regulators that enable specificity downstream of signal pathways that often converge on only a handful of regulatory molecules[5,6]. However, transcription is only the beginning of gene expression. Before an mRNA leaves the nucleus, it will undergo capping, polyadenylation, and commonly pre-mRNA splicing. Once exported to the cytoplasm, an mRNA will spend anywhere from a few minutes to many hours being translated into protein before it is degraded. Each of these steps in the mRNA life cycle are tightly regulated and collectively comprise “post-transcriptional regulation” [7].

This review focuses on what is currently known about the intersection of macrophage biology with post-transcriptional regulation of gene expression. We aim to integrate the canonical literature with up-to-date direct evidence of post-transcriptional regulation in macrophages. Tracing the mRNA life cycle, we discuss how post-transcriptional regulation governs core macrophage pathways, highlighting the importance of layered regulation at multiple post-transcriptional nodes (Table 1). We close with a discussion of opportunities for using a post-transcriptional lens to make further discoveries in macrophage biology.

Table 1. Macrophage functions and pathways regulated by post-transcriptional regulation.

Post-transcriptional Process	Pathway or Function	Refs
Alternative Splicing	TLRs, MyD88, MD2, Type I Interferon Response, IL-6, Mtb Infection	[13,14,16–18,59,63,64,70,71,74]
Alternative Cleavage	TLRs, RIG-I, TNF-α, SARS CoV-2 Antiviral Response	[21–24, 0]
mTOR Regulation of Translation Initiation and Elongation	Phagocytosis, Chemotaxis, Type II Interferon Response, Immunosupression	[26–28]
Inhibition of Translation Initiation by the Integrated Stress Response	Inflammasome Activation, PRR Sensing, Immunosupression	[31–33]
Regulation of mRNA stability	IL-6, TNF-α, OAS, Autoimmunity	[44,47,54,58,82]

Macrophages employ post-transcriptional regulation to control newly synthesized mRNAs

To fight infection and damage without risking chronic inflammation and autoimmunity, immunological signals must act like switches with firm ‘on’ and ‘off’ states. Historically, this has been conceptualized at the transcriptional level, whereby pro-inflammatory gene transcription is activated in the presence of pathogen or damage associated molecular patterns (PAMPs or DAMPs) [8]. To resolve inflammation, another round of transcription—this time of anti-inflammatory genes—is initiated. Beyond these waves of transcription, post-transcriptional control is increasingly recognized as indispensable for proper execution of gene expression reprogramming in immune cells [9], particularly at the levels of alternative pre-mRNA splicing and polyadenylation.

Splicing encodes transcript fate during mRNA synthesis

The processing of newly synthesized transcripts begins as soon as they exit RNA polymerase II (RNAPII) [10,11]. Co-transcriptional splicing allows cells to coordinate intron removal with transcription and chromatin remodeling; post-transcriptional splicing allows for staged processing before the mRNA is exported [12]. Macrophages regulate splicing to encode the fate of specific transcripts and express diverse protein products in important pathways.

Alternative splicing is widespread in macrophages and can be regulated by their differentiation and activation states. For example, a study examining genome-wide splicing patterns found that macrophage polarization (into either an “M1” inflammatory or “M2” wound-healing state) is associated with thousands of differential alternative splicing events, particularly global increases in long transcript isoforms [13]. Through comparing alternative splicing with differential expression data, the authors identified MBNL1 as a key regulator of alternative splicing during macrophage differentiation [13]. Similarly, stimulation of macrophages with lipopolysaccharide (LPS) has been shown to induce a significant increase in alternative splicing events globally (particularly exon skipping and intron retention), disproportionally impacting genes involved in innate immune and metabolic pathways [14]. Some of the pathways enriched for alternative splicing during LPS treatment include bacterial invasion of epithelial cells, chemokine signaling, and Fc gamma R-mediated phagocytosis, indicating that many facets of the rapid and dynamic macrophage response hinge on regulation at the level of splicing. Notably, while numerous splicing alterations are observed in this context, the proportion of these that have biologically relevant consequences is unclear. More mechanistic studies focused on specific splicing changes are necessary to complete our understanding of which splicing events are truly meaningful during macrophage activation.

One well-studied alternative splicing event that tunes macrophage activity occurs in Myeloid differentiation primary response protein 88 (MyD88), a factor important for inflammatory signal transduction by pattern recognition receptors (PRRs). MyD88 acts as an adaptor protein linking the IL-1 receptor and members of the TLR family to IL-1R-associated kinases, bridging the initial receptor detection of PAMPS, DAMPs, or cytokines with a signaling cascade that will activate downstream transcription of cytokines and chemokines to mount a response [15]. Two isoforms of MyD88 - ‘long’ and ‘short’ - are produced through alternative splicing. The long isoform, MyD88-L, promotes downstream inflammatory signaling after TLR activation whereas the short isoform, MyD88-S, inhibits signaling [16]. The choice of isoform has been attributed to regulation by two RNA-binding proteins (RBPs), SRSF1 and hnRNP-U. The splicing factor SRSF1 binds MyD88 transcripts to promote inclusion of exon 2, increasing production of MyD88-L. While the mechanism of hnRNP-U regulation of MyD88-S usage has not been formally tested, it has been suggested that hnRNP-U binds the 3’ end of intron 1 in the MyD88 transcript, resulting in an inhibition of exon 2 inclusion. MyD88-S dominates if there is a lack of SRSF1 binding or the presence of hnRNP-U [17]. How splicing factors are themselves ‘programmed’ to promote the generation of one MyD88 isoform over another during the course of macrophage activation remains an important outstanding question.

Controlling the balance of MyD88 isoforms has emerged as an attractive therapeutic target for modulating hyperinflammation. An agonist of Liver X Receptors (LXRs) was recently shown to down-regulate production of splicing factor 3A subunit 1 (SF3A1), which also regulates MyD88 splicing [18]. Decreased SF3A1 results in increased removal of exon 2, thereby elevating MyD88-S abundance and terminating downstream NF-κB signaling to inhibit inflammation [18]. Despite potential for anti-inflammatory drugs designed to target this mechanism, expression of MyD88-S has also been shown to suppress macrophage responses to heat-inactivated S. aureus. This suggests that elevated MyD88-S could result in an inability to control infection [19], and therefore future studies should follow up on the potential of increased infection risk as an unintended consequence of LXR agonist treatment.

Transcription termination and polyadenylation define the 3’ end of mRNA

Isoform diversity can also be controlled by the process of 3` end formation. Once a cleavage signal is transcribed by RNAPII, the cleavage and polyadenylation specificity factor (CPSF) recognizes the hexanucleotide polyadenylation signal (generally, AAUAAA) in the transcript, and CstF binds the UG-rich downstream sequence element. The interaction between CPSF and CstF determines the site where the endonuclease CPSF73 cleaves the pre-mRNA, and polyA polymerase is then recruited by CPSF to transcribe the poly(A) tail, thus defining the 3’ untranslated region (UTR) of the mRNA [20]. The resulting differences in 3` UTRs and the associated poly(A) tails can be quite large, as are their impacts on gene function.

Recent studies have begun to uncover genome-wide alterations in 3’ end processing in macrophages under different conditions. The use of tail-end displacement sequencing (TED-seq) to measure poly(A) tail length has shown that LPS stimulation of THP-1 macrophages results in global changes in poly(A) tail length, with poly(A) lengthening occurring in many transcripts encoding proteins involved in immune function and post-transcriptional regulation. This is presumed to enhance transcript stability [21]. While the authors of this study hypothesize that increases in poly(A) tail length are a consequence of readenylation of pre-existing transcripts, metabolic labeling to distinguish newly synthesized RNA would be necessary to support this conclusion.

The distribution of 3’ UTR length can also be modulated in macrophages. For example, exposing macrophages to vesicular stomatitis virus induces them to shorten the 3` UTR of their transcripts by an average of 150 nt after 8 hours [22]. Many impacted transcripts encode pattern-recognition receptors (PRRs) such as TLR7, TLR8, or are involved in antiviral signaling such as the RIG-I-like receptors DDX56 and IFIH1. This work demonstrated that shortening of the 3’ UTR can influence protein production through impacts on mRNA abundance and translation efficiency, but the specific impacts were gene dependent. Other work has shown that the abundance of the 3’ processing factor CPSF6 decreases during viral infection, resulting in shortening of 3’ UTRs, particularly those involved in viral infection, autophagy, signal transduction, metabolism, and cell proliferation [23]. Transcripts in these ontological groups have increased stability and translation efficiency, thereby increasing protein production and promoting activation of the type-I IFN response.

Additional clues to how macrophages control gene expression through the 3` UTR can be found in studies of inflammatory signaling in other cell lines. In fibroblasts, an alternative element in the 3` UTR of Tumor Necrosis Factor (TNF)-α can couple its post-transcriptional splicing to activation of protein kinase R (PKR), a kinase involved in antiviral defense in many cell types, including macrophages [24]. By dictating the nature of mature mRNAs that persist in macrophages, splicing and 3` end formation can help these cells up- and down-regulate inflammatory responses. Taken together, these studies suggest that macrophages may differentially regulate 3’end processing in response to distinct pathogen signals, perhaps to establish an appropriate antibacterial vs. antiviral proteome.

Cytoplasmic RNA regulation determines transcript translation and lifetime

Cytoplasmic processes control mRNA expression through both global and targeted mechanisms. Global pathways include metabolic sensing via mechanistic target of rapamycin (mTOR) and the integrated stress response (ISR). Targeted regulation is achieved through mRNA structure and sequence recognition. Together, these processes determine transcript translation efficiency and lifetime.

Macrophages modulate global translation rates through mTOR and ISR signaling

All cells regulate their metabolism through mTOR. Classically, mTOR controls cellular proliferation by upregulating biosynthetic pathways to make more ribosomes, increase translation efficiency, and generally turn up the ‘cellular thermostat’ [25]. Macrophages use mTOR to translationally upregulate cellular processes essential to mounting innate immune responses [26]. For example, in macrophages inhibition of mTOR by rapamycin reduces expression of ROCK1 (Rho associated kinase 1), a kinase synthesized downstream of mTOR signaling that is involved in several key cellular processes such as cell migration and cytoskeletal dynamics. This ultimately leads to downstream defects in phagocytosis and chemotaxis [26].

Regulation of mTOR produces effects that are specific to macrophages and studies that exploit this have revealed specific functional consequences resulting from the modulation of global translation rates. IFN-γ treatment leads to mTORC1 inhibition, depressing general translation but permitting translation of IFN-γ stimulated transcripts [27]. Macrophages lacking the mTORC1 inhibitor Tsc1 are unresponsive to IL-4 and fail to adopt a regenerative state [28]. Taking these findings together, we speculate that mTOR may be important for translational reprogramming when macrophages encounter a new stimulus.

Translation is globally downregulated during cell stress by the ISR, wherein one of four related kinases phosphorylate eIF2α leading to widespread translational shutdown. The four core ISR kinases are PKR, PERK, HRI, and GCN2, with each responding to a specific set of homeostatic disruptors including nutrient deprivation and oxidative stress [29,30]. Genetic deletion of the four ISR kinases leads to a variety of defects in macrophage activation. PKR deletion prevents inflammasome activation and HMGB1 release in macrophages stimulated by pathogenic bacteria [31]. PERK deletion blocks the immunosuppressive activation of macrophages, with little effect on their response to LPS and IFN-γ [32]. Loss of HRI reduces production of CXCL1 and IL-6 and generally disrupts pattern recognition molecule related signaling in BMDMs and other cell types [33]. In a study of tumor-associated myeloid cells, GCN2 was required for murine macrophages to adopt the anti-inflammatory, pro-tumor proliferation phenotype frequently seen in tumors [34].

New pharmacological tools targeting the ISR, ISRIB and 2Bact, are becoming increasingly available. Both prevent translation inhibition through the ISR by preventing phosphorylated eIF2α from inactivating eIF2β [35]. Misfolded proteins accumulating in the endoplasmic reticulum (ER) are a potent trigger of the ISR [36]. Treatment of macrophages with ISRIB during ER stress changes cytokine expression, implying that the ISR tunes the inflammatory output of macrophages [37]. The notion that the ISR has signaling roles outside of acute stressors is intriguing, and how cells transiently ‘pulse’ translation inhibition to change gene programs will be an exciting avenue of future study. Therapeutics targeting specific ISR kinases are being developed to treat cancer, neurodegeneration, and other conditions, offering a prelude to new tools that could be deployed in macrophages [38,39].

Macrophage gene programs are switched off through mRNA decay

The lifetime of an mRNA is tightly regulated. Decay can be initiated by 5’ decapping, 3’ deadenylation, or endonucleolytic cleavage [40]. Coordinated decay of transcripts encoding potent inflammatory cytokines is crucial to ensure these signals are expressed when, and only when, they are needed [41]. The timing of transcriptional activation and transcript decay rate are highly variable between inflammatory genes, allowing each gene product to exert its effects at precisely the right time and for the appropriate duration. This fine-tuned regulation ensures that macrophages mount a response robust enough to eliminate the detected threat without causing excessive inflammation. Thus, RNA decay plays a major role during macrophage inflammation and its resolution.

Global transcriptomic analyses in monocytic cell lines and primary macrophages have revealed that RNA stability and decay are highly regulated in the context of macrophage activation. Physiological hypoxia, for example, induces a shift toward transcript instability in undifferentiated THP-1 cells, with many nuclear-encoded mitochondrial mRNAs and transcripts involved in immunity being destabilized [42]. In primary macrophages, global RNA decay rates nearly double when stimulated with hypoxia, LPS, and IFN-γ, with pathways involved in iron metabolism and immune signaling enriched for destabilized transcripts [43]. In both studies, the signaling profile of the cells was altered, with numerous cytokines and cytokine receptors being differentially expressed. However, the precise consequences in vivo remain unknown. Consistent with a role for regulated mRNA decay in tuning immune outcomes, viral infection has been shown to decrease the average 3’ UTR length of macrophage mRNAs, resulting in global transcript instability [22].

Control of immune-related transcript stability has been linked to nucleases and RBP binding. IL-6 and TNF-α are two examples of canonical inflammatory cytokines controlled by RBPs. IL-6 expression is attenuated by the zinc finger nuclease Regnase-1, which degrades the IL-6 transcript in response to activation of the NF-κB pathway [44]. The result is a tight pulse of IL-6 production. Macrophages lacking Regnase-1 induce an augmented response when stimulated with LPS, and Regnase-1 knockout mice fail to survive past twelve weeks due to a spectrum of hyperinflammatory phenotypes [45,46]. TNF-α is regulated in a similar manner. The Roquin RNA-binding proteins have been shown to dampen TNF-α production and the broader inflammatory response in macrophages. Rather than direct action by a nuclease, TNF-α transcripts are bound by RBPs Roquin-1 and Roquin-2, which subsequently recruit the Ccr4-Not deadenylase complex, ultimately leading to TNF-α transcript degradation [47].

Targeted transcript decay is also achieved through micro RNAs (miRNA). As they are reviewed elsewhere [48,49], we briefly note that miRNAs are used by macrophages to control post-transcriptional steps of gene expression in trans. Specifically, BMDMs have been shown to deliver miRNAs targeting the NF-κB and TNF-α signaling pathways in exosomes to suppress inflammatory gene expression in neighboring cells [50]. In response to inflammatory stimuli, it has also been shown that macrophages can prevent spurious cytokine production by upregulating factors that facilitate the translocation of miRNA processing machinery to the nucleus, thereby increasing production of miRNAs that inhibit MyD88 expression [51].

By clearing existing mRNAs and allowing for new transcripts to dominate the transcriptome, RNA decay may also be used to control the timing with which macrophages can shift between different states, using global or transcript directed mechanisms. The post-transcriptional mechanisms employed by macrophages to transition between different gene expression programs in various stages of inflammation and regeneration remain an important gap in knowledge.

Post-transcriptional control creates multilayered gene circuits

Gene circuits and switch-like behaviors arise from the enchainment of RNA processing steps like splicing and decay. We next discuss the importance of a wide view of how post-transcriptional mechanisms affect macrophage gene programs because they frequently combine to form intricate regulatory circuits.

Nuclear post-transcriptional processing exerts effects through downstream events

RNA isoform determination can have pronounced impacts on post-transcriptional regulation. This commonly occurs via exon inclusion events that introduce premature termination codons (PTCs), which are sensed by UPF proteins and target transcripts for degradation via nonsense mediated decay (NMD) (Figure 1A) [52,53].

Figure 1. Post-transcriptional mechanisms create layered gene circuits.

An mRNAs expression is regulated throughout its life cycle, where upstream processing can exert effects on downstream post-transcriptional mechanisms. A) Some isoforms produced by alternative splicing include the premature stop codons that trigger NMD. Regulated production of the NMD isoform leads to elimination of the target transcript in spite of transcription. B) Regulated polyA tail length exerts downstream effects on translation efficiency, amplifying or inhibiting production of the protein product for a given transcript.

In murine macrophages, the oligoadenylate synthetase 1g (Oas1g) gene is produced in two isoforms: one that is translated into a key antiviral protein, and another that is a target for NMD [54]. Following stimulation with a double-stranded RNA mimic Poly I:C, the NMD target splice variant comes to dominate, thereby preventing OAS1g protein production and leading to attenuation of the antiviral response. Complete removal of the responsible splice site leads to an increase in apoptosis suggesting this negative feedback is essential to modulating the antiviral response in macrophages. Notably, humans possess only one OAS1 gene, whereas the murine genome has eight paralogs, which limits our ability to extrapolate these observations made in murine macrophages to humans. However, the human OAS1 gene has also been shown to be post-transcriptionally regulated through generation of an NMD target, as well as through the production of six different productive isoforms with varying oligoadenylate synthetase activity. These observations highlight a role for alternative splicing and NMD in modulating macrophage responses to pathogens by tightly controlling the quantity of translation-competent mRNAs, and thereby the production of pro-inflammatory proteins.

Polyadenylation also plays a major role in determining downstream translation in macrophages, as longer poly(A) tails can enhance the efficiency of polysome formation (Figure 1B) [55]. For example, TNF-α transcripts have truncated poly(A) tails in unstimulated macrophages, but upon treatment with LPS, deadenylases are inhibited and the tail is allowed to grow through the activity of cytoplasmic adenylases [56]. In this way, TNF-α transcripts are recruited to the ribosome more efficiently, and expression increases.

Tristetraprolin (TTP) is an immunoregulatory RBP that destabilizes transcripts, including the TNF-α mRNA, in a similar manner to Regnase and roquin proteins [57]. Recent work in mouse macrophage cell lines has demonstrated that TTP must first bind pre-mRNAs in the nucleus to license it to degrade targets [58]. Once exported, TTP recruits transcripts to a degradation complex for degradation in the cytoplasm, linking nuclear RNA processing with cytoplasmic decay.

Post-transcriptional regulation controls PRR signaling at every layer

One way that post-transcriptional processes attenuate macrophage inflammatory responses is through the generation of negative feedback loops. TLR signaling is a prime example of such negative feedback (Figure 2). All members of the TLR family are expressed as multiple splice variants, some of which are synthesized in response to TLR activation and serve as negative regulators of downstream signaling [9]. For example, LPS-stimulation promotes the inclusion of an additional exon in murine TLR4, resulting in a soluble form of TLR4 that inhibits NF-kB activation to modulate responses [59]. Importantly, TLR4 alternative splicing has demonstrated clinical relevance. The sTLR4/sMD2-complex has been shown to inhibit pro-inflammatory cytokine secretion by competing with TLR4/MD-2 at the cell surface, thereby reducing LPS-induced lung injury in vivo [60]. High serum sTLR4 levels have also been associated with severe COVID-19 [61], and urine sTLR4 has shown utility as a biomarker to predict pediatric UTIs and subsequent pyelonephritis, and to track treatment response [62].

Figure 2. Post-transcriptional regulation contributes to negative feedback loop formation at multiple stages of the TLR4 signaling pathway.

The TLR4 signaling pathway consists of several targets for post-transcriptional regulation, several of which contribute to the formation of negative feedback loops that modulate pro-inflammatory responses. (i) Upon detecting LPS, TLR4 is activated and binds several adaptor molecules that facilitate propagation of signaling, in part through the NFκB pathway. (ii) NFκB signaling leads to the initiation of transcription of many cytokines and chemokines. (iii) In response to activation of TLR signaling, the adaptor protein MD2 undergoes alternative splicing and is produced in a shorter isoform (lacking exon 2 in human macrophages). (iv) The protein product of this short isoform competes with canonical MD2 for binding to TLR4 to inhibit signal transduction. (v) Similarly, after TLR activation, adaptor protein MyD88 undergoes alternative splicing and is produced in a short isoform (MyD88-S) lacking exon 2. MyD88 isoform usage is dictated by the balance of SRSF1 and hnRNP U, where SRSF1 binding to the MyD88 transcript promotes inclusion of exon 2, and hnRNP U binding promotes removal of exon 2. (vi) MyD88-S acts as a dominant-negative inhibitor of IL-1 and LPS-induced activation of NFκB. (vii) Finally, TLR4 itself is also alternatively spliced to include an additional exon. (viii) This results in production of a soluble form of TLR4 that inhibits NFκB activation. Thus, regulation at the level of post-transcriptional processing allows for tight control over multiple stages of the TLR4 signaling cascade, facilitating rapid returns to homeostasis.

In another example, alternative splicing of the TLR4 co-receptor MD2 also serves to downregulate inflammation. Upon TLR stimulation, production of a short MD2 isoform (lacking exon 2 in human macrophages, and a portion of exon 3 in murine macrophages) increases, and competes with the canonical isoform for binding to TLR4 [63,64]. Together, these make three separate targets in one pathway (MyD88 (discussed earlier), MD2, and the TLRs themselves) whose alternative splicing triggers negative feedback loops. Layered negative feedback likely acts as a failsafe to prevent runaway TLR signaling - and thus, runaway inflammation - in macrophages.

Post-transcriptional control provides an additional level of gating for genes subject to promiscuous transcription

Some macrophage gene circuits operate downstream of futile transcription. For example, while regulation of TNF-α at the level of transcription has been well established, it has also been shown to undergo post-transcriptional control at the levels of nuclear export, translation, and stability, sometimes overwriting transcriptional activity [65–67]. For example, under basal conditions, transcription of TNF-α can be initiated but mRNA or protein do not accumulate due to low processivity of the transcript and failure to associate with polysomes resulting from poly(A) tail truncation [24,56,68]. Recent work has also implicated the mitochondrial fission protein DRP1 in regulating TNF-α. Dynamin-related protein 1 (DRP1) promotes TNF-α protein production without influencing Tnfa transcript abundance, suggesting a mechanism of DRP1-dependent regulation of TNF-α at the level of translation or protein stability [69]. Expression of another potent inflammatory cytokine, IL-6 is gated by an RNA binding protein hnRNP M, whereby hnRNP M binds and suppresses the splicing of IL-6 transcripts, slowing IL-6 processing as a means to prevent spurious inflammatory responses [70]. In response to the activation of TLRs or other pattern recognition receptors, hnRNP-M is phosphorylated and no longer blocks intron removal, permitting maturation of the IL-6 mRNA.

RNA binding proteins can also influence macrophage gene programs without targeting inflammatory signaling molecules directly. Zinc finger RNA-binding protein (ZFR) suppresses expression of the NMD splice isoform of macroH2A1, a histone variant that attenuates the type I interferon (IFN) response [71]. ZFR itself has multiple isoforms, with shortened isoforms expressed in monocytes, and the full-length functional transcript expressed in macrophages. Thus, ZFR represents a two-step splicing gate to control the type I interferon response as monocytes differentiate into macrophages.

Alternatively, gating of transcriptional activity can be achieved by blocking nuclear export of mRNAs. For example, intron retention in signaling and phagocytic transcripts prevents their export in THP-1 cells; once fully spliced, the transcripts are exported and expressed [72]. A recent study performed RNA-seq of the chromatin-associated, nucleoplasmic, and cytoplasmic fractions of endotoxin-stimulated mouse BMDMs to uncover the kinetics of inflammatory transcript export [73]. This revealed that export rates for innate immune mRNAs are gene specific and correlated to cytoplasmic decay rate. Notably, the kinetic competition between export rate and cytoplasmic decay does not directly regulate the timing, but rather the magnitude of gene expression. The authors speculate that this regulation serves to ensure similar abundances of long- and short-lived mRNAs during successive waves of gene expression. These results emphasize how tuning every step of a transcript life cycle allows specific kinetics for its expression.

Post-transcriptional mechanisms present unique therapeutic opportunities in disease

The intracellular bacterial pathogen Mycobacterium tuberculosis (Mtb) survives phagocytosis by macrophages and replicates within endosomes, providing an opportunity to understand post-translational regulation in the context of the host-pathogen arms race. RNA-seq of human macrophages infected with Mtb has revealed shifts in splicing patterns, with genes involved in the immune response, metabolism, and autophagy undergoing infection-specific alternative splicing [74]. Specific splicing events are associated with virulent Mtb strain H37Rv, implying that Mtb virulence can tune splicing output. As further evidence of regulated splicing during Mtb infection, PBMCs of tuberculosis patients possess an increased ratio of alternatively spliced TLR1 transcripts compared to healthy controls [75]. Given the contribution of TLR1 in TLR2 dimerization (a key step in promoting bacterial clearance), this variant may alter TLR1 protein production and thereby impact Mtb pathogenesis. Notably, it is not clear whether these splicing changes are mediated by the macrophage or the pathogen. Intriguingly, Mtb and other intracellular bacterial pathogens encode and secrete proteins capable of manipulating post-transcriptional regulation, suggesting an evolutionary incentive for the pathogen to corrupt these mechanisms [76–79].

During viral infection, there is evidence for dysregulated polyadenylation contributing to poor patient outcomes. RNA-seq of PBMCs from COVID-19 patients shows alternative polyadenylation events, most of which involved a global shortening of transcript 3’ UTRs [80]. Patients with shorter 3’ UTRs were more likely to be admitted to the intensive care unit and were also found to spend more days on a ventilator compared to those with longer 3` UTRs. The mechanism and causality of this UTR shortening in myeloid cells from COVID-19 patients remains unknown. While PBMCs do consist of myeloid lineage cells, this pool also includes lymphocytes, thus monocyte/macrophage-specific 3’ UTR changes in the context of COVID-19 infection remain unclear. Additional studies will be needed to parse out the contribution of 3’UTR shortening in monocytes and macrophages to severe COVID-19 pathology and the antiviral response to SARS-CoV-2.

Defects in RNA decay in macrophages can lead to autoimmunity. Nonsense mutations in Roquin-1, a key regulator of immune homeostasis, lead to a spectrum of human autoimmune diseases including macrophage activation syndrome [81]. A mouse model of Roquin and Regnase-1 depletion reproduces a lupus-like phenotype, though the specific effects on macrophages from these mice were not investigated [82]. These observations argue that Roquin and Regnase-1 are important for turning off inflammatory gene expression and hint at other gates and ‘off switches’ governed by post-transcriptional mechanisms contributing to autoimmunity[83].

Fortunately, post-transcriptional mechanisms are also therapeutic opportunities. A recent study showed that an engineered poly(A) tail binds and inhibits extracellular cold-inducible RNA-binding protein (eCIRP) to attenuate inflammation in acute lung injury caused by gut ischemia/reperfusion [84]. Development of isoform-specific therapies is another area of active investigation. For example, platelet-derived growth factor receptor isoform α (PDGFRα) agonists have been shown to mediate damage caused by cardiac inflammation in mice (whereas agonists targeting the PDGFRβ isoform exacerbate inflammation) [85]. One anticipates there are many undiscovered treatments for chronic inflammation, autoimmunity, and infection that could leverage post-transcriptional mechanisms active in macrophages.

Concluding remarks and future perspectives

The last five years have marked significant achievements in understanding how post-transcriptional regulation governs macrophage gene expression. Yet, the dynamic and context-specific nature of macrophage responses, and the complexity of crosstalk between post-transcriptional processing and other layers of regulation are significant obstacles to move beyond observational studies. Additionally, technologies enabling the high-throughput detection of alternative processing events (such as splicing and polyadenylation) and mRNA fate (such as translation efficiency and decay) have only become available in the last ten to fifteen years.

Evident in the literature described here, we possess the foundation for discovering the mechanistic and causal basis of post-transcriptional regulation in macrophage biology. We now possess a veritable mountain of data obtained with short-read sequencing methods describing transcript processing, translation, and decay events. Many studies highlight pervasive alternative post-transcriptional processing events during macrophage activation, but relatively few go further to show that these changes have biological consequences. We consider sorting out high-magnitude and high-consequence post-transcriptional events from large profiling datasets to be an exciting opportunity for future work. Some mechanistic insight has come into view for thoroughly studied pathways such the TLRs and TNF-α expression. While one logical approach to parsing out biologically relevant changes from these datasets would be to identify changes of the greatest magnitude, it is important to note that even subtle transcript processing changes may have amplified downstream effects. Orthogonal approaches to identifying genes of interest should be deployed to inform the selection of alternatively processed hits for further study.

It is also important to consider that RNA processing factors do not always control the same targets in distinct biological contexts [70]. It is tempting to speculate that the same signaling cascades that activate innate immune transcription factors in macrophages can promote post-translational modification (PTM) of RNA binding proteins as well. In support of this idea, mRNA processing and splicing related proteins are among the most differentially phosphorylated in several global proteomics analyses of macrophages infected by different pathogens [86,87]. As mass spectrometry gets better at quantifying PTMs on specific peptides, it will be important to catalog how RNA processing factors are modified in macrophages responding to distinct cues, as this will inform mechanistic studies of how RNA processing proteins orchestrate one transcriptome/proteome vs. another in times of stress.

Major advances will require a change in approach by immunologists and RNA biologists alike. We propose two strategies for moving forward: 1) Those working to understand the biochemical mechanisms of RNA processing should account for the importance of kinetics in macrophage function. Recognizing variation in timing and magnitude of macrophage responses, and considering these parameters when selecting appropriate model systems, will help identify how macrophages leverage post-transcriptional regulation to maintain precise control over responses to stimuli. 2) Those studying macrophage immune responses should consider the epistatic relationship of RNA processing to transcriptional activation. Post-transcriptional events can modulate or even override transcriptional responses. Methods such as single-cell RNA-seq and ATAC-seq are continuously providing new data, but time points are typically limited and blind to the role of post-transcriptional effects.

Post-transcriptional regulation plays a key role in all gene expression pathways. We regard these mechanisms as particularly important to macrophage biology because of the diversity of biological functions these cells perform during inflammation and regeneration. Given the availability of new technologies and techniques for studying both RNA biology and macrophages, and the number of unanswered questions that remain regarding the role of RNA processing in regulating macrophage responses (see Outstanding questions), we are excited to see how the upcoming five years of research further our mechanistic understanding of how post-transcriptional regulation affects macrophages and other specific immune cells.

OUTSTANDING QUESTIONS BOX.

• How have bulk measurements and single timepoint studies limited our understanding of post-transcriptional regulation in macrophages; and can new technologies overcome this limitation?

• Which post-transcriptional regulation mechanisms are rate limiting of gene expression responses to specific stimuli? For example, we know the tipping point for where new TNF-α transcription overcomes inhibitory post-transcriptional mechanisms, but many signal pathways are less well understood.

• In what tissues and pathways do post-transcriptional mechanisms have specificity for macrophages? Many studies here show post-transcriptional regulation is important in macrophages but do not detail if the mechanism is an exception or the rule for all cells.

• Which mechanisms of post-transcriptional regulation hold the most therapeutic potential for treating chronic infection targeting macrophages and inflammatory pathologies driven by macrophage signaling?

HIGHLIGHTS BOX.

• As first responders in the innate immune system, macrophages use post-transcriptional regulation to facilitate rapid and fine-tuned responses to environmental conditions, particularly at the levels of pre-mRNA splicing, 3’ end formation, mRNA stability, and translational control.

• Novel sequencing technologies have uncovered global changes in RNA processing during macrophage activation.

• Negative feedback loops generated by alternative splicing, such as those in TLR signaling pathways, modulate macrophage inflammatory responses and facilitate rapid returns to homeostasis.

• Global regulation via mTOR and the integrated stress response and targeted control by RNA-binding proteins regulate mRNA translation efficiency and stability, ultimately determining the expression and lifetime of transcripts during macrophage activation.

• Post-transcriptional processes present attractive therapeutic opportunities for the treatment of inflammatory diseases.

SIGNFICANCE.

Macrophages are first responders of the innate immune system and play a key role in initiating and resolving inflammation, antimicrobial defenses, and tissue repair. While changes in transcription during macrophage activation have been extensively studied, post-transcriptional processing has recently gained appreciation as a key regulatory node facilitating the speed and magnitude of macrophage responses. Post-transcriptional mechanisms can override transcription, combine into complex layered regulatory circuits, and are unexplored avenues for therapeutic development.

Glossary

3’ untranslated region (UTR)

the portion of mRNA that is after the translation stop codon. This region of the transcript contains key regulatory elements such as AU-rich elements and microRNA target sites, and thus regulates gene expression through impacts on mRNA stability, localization, and translation

DAMPs (damage-associated molecular patterns)

small molecules released from damaged or dying cells, which then interact with pattern recognition receptors to initiate an immune response

Integrated stress response (ISR)

a complex signaling network involved in maintaining cellular homeostasis in response to stress. 4 kinases (PERK, GCN2, PKR, and HRI) are activated in response to stress and phosphorylate eukaryotic translation initiation factor (eIF2), resulting in a global reduction in protein production. At the same time, the ISR activates a gene expression program aimed at combatting the stressor. If unable to successfully restore homeostasis, the ISR then triggers apoptosis

Interferons (IFNs)

a group of cytokines induced upon pathogen detection that are involved in mobilization of the immune response. These cytokines are divided into three classes (type I, type II, and type III). Type I interferons, including IFN-β, are produced in response to PRR activation and signal through the interferon-α/β receptor (IFNAR) to engage the JAK-STAT pathway

Janus activated kinase-signal transducer and activation of transcription (JAK-STAT) signaling pathway

A key signaling pathway that mediates numerous cell functions, including hematopoiesis, immune regulation, and tissue repair. JAK family proteins associate with and phosphorylate transmembrane receptors upon ligand binding, after which STAT proteins are recruited and phosphorylated. STAT proteins then translocate to the nucleus to induce transcription of cytokines and growth factors

Lipopolysaccharide (LPS)

an outer membrane component of gram-negative bacteria, also known as endotoxin. LPS is a potent activator of toll-like receptor 4 (TLR4)

Mechanistic target of rapamycin (mTOR)

a kinase that controls a myriad of cellular processes, including metabolism, immune responses, proliferation, migration, and autophagy. mTOR is also known to exert control over macrophage polarization into M1 or M2 phenotypes

Myeloid differentiation primary response protein 88 (MyD88)

an adaptor protein responsible for linking the IL-1 receptor (IL-1R) or TLRs to members of the IL-1R-associated kinase (IRAK) family of kinases

NF-κB

A transcription factor involved in regulating many facets of both innate and adaptive immune responses. Activation of NF-kB takes place in response to a diverse array of stimuli, including ligand binding to pattern-recognition receptors, to induce transcription of cytokine and chemokines

Nonsense-mediated decay (NMD)

the process by which aberrant mRNA transcripts are degraded, particularly those with premature stop codons. NMD also facilitates changes in transcriptomes (and therefore proteomes) in response to altered conditions such as pathogen detection

PAMPs (pathogen-associated molecular patterns)

small molecular motifs derived from pathogens that engage host pattern recognition receptors, triggering a host response

Pattern Recognition Receptors (PRRs)

a class of receptors that recognize and initiate responses to microbial components and damaged or dying cells. One key group of PRRs is the toll-like receptor (TLR) family, which are expressed on sentinels such as macrophages and serve as a critical link between innate and adaptive immune responses

Polyadenylation

An essential step in pre-mRNA 3’-end processing, in which ~200–250 adenosine monophosphates are added to the end of the transcript by poly(A) polymerase. Poly(A) tails are present on essentially all eukaryotic mRNAs, and are involved in transcript stability, export to the nucleus, and translation efficiency. Through the use of different polyadenylation sites, transcripts can be alternatively polyadenylated to generate isoforms with varying 3’ UTRs

Pre-mRNA splicing

the process by which intronic sequences are removed from mRNA precursors by machinery called the spliceosome. Cells can leverage this process to generate multiple protein products from the same gene through alternative splicing, creating functional diversity across cell types and enabling cells to respond to a variety of stimuli by generating different proteins

RNA-binding proteins (RBPs)

proteins that bind RNA to facilitate various aspects of its function and processing. This includes splicing factors, which are RPBs that regulate the splicing of pre-mRNA

TNFα

a cytokine and key mediator of inflammatory responses, produced by activated macrophages as well as natural killer (NK) and T cells. Through activation of the TNFR1 and TNFR2 receptors, TNFα can amplify inflammatory signaling through activation of NF-kB and MAPKs and can also induce programmed cell death via activation of caspase-8 and MLKL. Aberrant TNFα production is associated with several autoimmune diseases and thus has been widely recognized as an attractive therapeutic target for modulating inflammatory responses