HIV‐1 ALLINIs

Although the discovery of INSTIs has profoundly transformed HIV therapy by expanding the therapeutic options for the treatment of AIDS, the continued emergence of drug‐resistant viral strains has represented a major clinical challenge. Resistance to both first‐ and second‐generation INSTIs has been reported, particularly in treatment‐experienced patients with multidrug‐resistant HIV strains [7, 8, 9, 10, 12, 76, 77]. These observations have spurred the development of new antiretroviral agents targeting unexplored steps of viral life cycle to overcome drug resistance. Accordingly, since the approval of RAL, considerable attention has been devoted to the identification of allosteric inhibitors of IN, with a primary focus on disrupting the interaction between IN and host factors [30, 42, 78]. In this context, most research has focused on targeting the interaction between HIV‐1 IN and the cellular cofactor LEDGF/p75 [79, 80, 81]. LEDGF/p75 binds IN through its C‐terminal IBD, which engages a V‐shaped pocket at the CCD–CCD dimer interface, a region distinct from the catalytic site. Crystallographic studies have revealed that this interface forms a well‐defined binding pocket stabilized by hydrophobic interactions and hydrogen bonds [38, 82]. Mutagenesis experiments have further confirmed the critical role of residues from both IN and LEDGF/p75 in mediating this interaction (Figure 5) [38, 45, 83, 84, 85, 86, 87]. Specifically, LEDGF/p75 interacts with E170 and H171 of one subunit of IN via bidentate H‐bonds mediated by D366, while the IBD residue I365 nestles into a hydrophobic pocket composed by L102, A128, and W132 from the other IN subunit (Figure 5). First‐generation ALLINIs were indeed pharmacologically designed to recapitulate these specific polar and nonpolar interactions, effectively mimicking the binding signatures of the natural cofactor [30, 38, 42, 45, 78, 79, 84, 88].

FIGURE 5: Crystal structure of dimeric CCD of HIV‐1 IN in complex with LEDGF/p75 IBD (PDB: 2B4J). Overall view of the CCD dimer of IN in complex with LEDGF/p75 IBD. On the right panel, a close view focused on CCD–IBD binding. The proteins are shown as cartoons, with IN CCD dimer chains colored in light green and light cyan and the LEDGF/p75 IBDs in light pink, while the catalytic triad residues are shown as yellow spheres.

Notably, several of these interface residues are central to the binding affinity and clinical profile of ALLINIs. Specifically, all ALLINIs retain a characteristic carboxylic group to mimic the bidentate interaction between LEDGF/p75 D366 and IN, alongside an aromatic side chain that occupies the W132/L102 cleft and recapitulates the hydrophobic role of I365 [20, 30, 42].

Among these residues, T125 is of clinical relevance as it is subject to natural polymorphism; the prevalence of alanine at this position (A125) in several viral clades can inherently reduce the baseline susceptibility to certain inhibitor scaffolds [89, 90].

Additionally, A128 plays a pivotal role in the emergence of viral resistance. Indeed, being frequently targeted by mutations, the replacement of this alanine with a bulkier or polar residue, such as threonine (A128T), introduces steric constraints that hinder inhibitor binding [30, 91].

Similarly, residue Y99 is often subject to mutations under drug pressure, usually compromising the stability of the inhibitors within the allosteric pocket [92].

Based on these structural insights, disruption of the LEDGF/p75–IN interface by small molecules emerged as a feasible strategy, particularly amenable to structure‐based drug design approaches owing to the availability of high‐resolution structural data [38, 42, 45, 82, 93].

The mechanism of action of ALLINIs has been extensively investigated and elucidated over the years. Strikingly, while early efforts aimed to identify compounds that directly inhibit the LEDGF/p75–IN interaction in vitro, only a subset of ALLINIs exhibited this activity. Rather, a shared and defining feature of this class is to act as molecular glues, inducing aberrant hyper‐multimerization of HIV‐1 IN and ultimately impairing viral infectivity.

Reflecting the evolving understanding of their biological behavior, these compounds have been described using various acronyms, including LEDGIN (LEDGF/p75–IN interaction site), NCINI (noncatalytic IN inhibitor), INLAI (IN–LEDGF/p75 allosteric inhibitor), and MINI (multimerization IN inhibitor). Importantly, compounds initially designed as disruptors of the LEDGF/p75–IN interaction were later shown to induce hyper‐multimerization of HIV‐1 IN and to primarily inhibit viral maturation and morphogenesis, rather than the integration step itself. In the following sections, we summarize the key advances in the discovery and development of ALLINIs, highlighting both mechanistic insights and the most promising candidates emerging for the treatment of HIV‐1.

Discovery and Development of Early ALLINIs

Two main strategies were pursued in the search for inhibitors targeting IN through mechanisms different from INSTIs: (i) a structure‐based drug design approach to identify protein–protein interaction inhibitors as disruptors of the LEDGF/p75–IN binding and (ii) high‐throughput screening to identify 3′‐P inhibitors. Following identification, the compounds were characterized to elucidate their multimodal mechanism of action and specific antiviral effects. In this framework, continuous efforts have led to the identification of potent inhibitors that demonstrated efficacy in in vitro models of infection. In addition, investigation of the chemical space through SAR studies allowed for the optimization of specific chemical features of early inhibitors, leading to the identification of additional modulators with improved antiviral potency, PK properties, and resistance profiles, thereby prompting their evaluation as innovative potential drug candidates.

Toward the Search for HIV‐1 IN–LEDGF/p75 Interaction Inhibitors: The Discovery of LEDGINs

Early ALLINI development aimed to block the HIV‐1 IN–LEDGF/p75 interaction using inhibitors that target the IN CCD dimer interface. While peptide derivatives demonstrated that targeting this protein–protein interaction was a feasible endeavor, more prominent efforts have focused on small‐molecule inhibitors owing to their superior potential as therapeutic agents [42, 45, 78, 94, 95, 96].

Building on emerging structural insights into the LEDGF/p75–IN interface [38, 82], structure‐based and virtual screening approaches ultimately led to the identification of the first quinoline‐based ALLINIs, providing early evidence that small molecules could effectively target this interface and exert antiviral effects [79, 97]. In a pioneering study, a large‐scale virtual screening campaign was carried out to derive a consensus pharmacophore model informed by the available crystal structures of the IN CCD [38, 79, 93, 98]. Among the identified hits, the ketimine derivative 6 [79] (Figure 6) was experimentally validated as an in vitro inhibitor of the IN–LEDGF/p75 interaction in the AlphaScreen‐based protein–protein binding assay (Table 1). Beyond eliminating the unstable secondary ketimine at the 3‐position, replacing the tetrazole with a carboxylic acid bioisostere led to the identification of the arylacetic acid 7 [79] (Figure 6) as a more potent inhibitor of LEDGF/p75–IN interaction with moderate antiviral activity and poor cytotoxicity (Table 1).

FIGURE 6: Chemical structures of ALLINIs 6–19.

Compounds | IN–LEDGF binding (IC 50 , µM) a | 3′‐P (IC 50 , µM) b | ST (IC 50 , µM) c | Antiviral efficacy (EC 50 , μM) d | CC 50 (MT‐4, μM) e | References
6 | 36%f | ND | ND | ND | ND | [79]
7 | 12.2 ± 3.4 | ND | ND | 41.9 ± 1.1 (MT‐4) | >150 | [79]
8 | 10.0 ± 0.4 | 3.9 ± 0.5 | 4.2 ± 0.6 | 12.2 ± 2.9 (SupT1) | 59.8 ± 0.5 | [79, 99]
9 | ND | 9.0 | ND | >40 (C8166) | ND | [100]
10 | 1.0 ± 0.1 | 2.3 ± 0.1 | 1.7 ± 0.1 | 5.8 ± 0.1 (SupT1) | >100 | [99, 101]
11 | 0.3 ± 0.02 | 0.18 ± 0.02 | 0.17 ± 0.03 | 0.63 ± 0.30 (HEK293T) | ND | [91]
12 | 0.215 | 0.106 ± 0.005 | 0.055 ± 0.003 | 0.039 ± 0.011 (MT‐4) | 76 ± 12 | [102]
13 | 0.019 | 0.151 ± 0.012 | 0.067 ± 0.004 | 0.018 ± 0.001 (MT‐4) | 42 ± 10 | [102]
14 | 0.228 | ND | ND | 0.055 ± 0.012 (MT‐4) | 48 ± 2.5 | [102]
15 | 0.58 ± 0.30 | ND | ND | 0.76 ± 0.08 (MT‐4) | 72.2 ± 5.15 | [103]
16 | 0.046 ± 0.012 | 0.739 | 0.573 | 0.069 ± 0.003 (MT‐4) | 96.0 ± 16.0 | [103]
17 | 0.050 ± 0.002 | ND | ND | 0.19 ± 0.06 (MT‐4) | 68 ± 8 | [104]
18 | 0.20 | ND | 0.17 | 0.54 (MT‐4) | >50 | [105]
19 | 0.096 ± 0.003 | ND | ND | 0.046 ± 0.016 (MT‐4) | >200 | [104]

Subsequent structure‐based hit‐to‐lead optimization efforts focused on recapitulating the key interactions established by the LEDGF/p75 IBD hotspot residues I365, D366, and L368 at the IN CCD dimer interface (Figure 5), ultimately leading to the identification of the 2‐(quinolin‐3‐yl)acetic acid derivative LEDGIN‐6 [79] (CX0516, 8, Figure 6). Notably, the compound demonstrated superior potency in inhibiting IN–LEDGF/p75 interaction and enhanced antiviral activity compared to derivative 7 (8: EC50 = 2.35 µM in MT‐4 cells) [79], while further studies indicated potency against 3′‐P and ST [88, 99] (Table 1). In addition to this, compound 8 exhibited antiviral activity against primary HIV‐1 isolates derived from human donors without cross‐resistance with established antiretroviral drugs, such as RAL, supporting a distinct mechanism of action. In contrast, a viral variant bearing the double mutation A128T/E170G in IN residues forming the LEDGF/p75‐binding pocket, previously selected for resistance to transdominant inhibition by overexpressed LEDGF/p75 IBD [85], was fully resistant to compound 8, while retaining susceptibility to other antiretroviral agents. Notably, the emergence of the A128T mutation under prolonged selective pressure with 8 strongly supported its function as an allosteric inhibitor that disrupts the LEDGF/p75–IN interaction, as this residue is positioned directly at the protein–protein interaction interface. The co‐crystal structure of derivative 8 in complex with the IN CCD (PDB: 3LPU) showed that the compound occupies the LEDGF/p75‐binding pocket, interacting with the IN residues E170 and H171 via its carboxyl group and reproducing key protein–protein contacts normally mediated by the IBD residue D366 (Figure 7). Furthermore, the phenyl moiety and the chlorine function of 8 nestle in the W132/L102 pocket, clearly mimicking the hydrophobic residues of IBD I365 and L368, respectively (Figure 5). Based on these structural and functional characteristics, compound 8 can be regarded as the prototype of the LEDGIN, representing the first class of ALLINIs specifically optimized to disrupt the LEDGF/p75–IN protein–protein interaction and to produce antiviral effects [79].

FIGURE 7: Crystal structure of derivative 8 in complex with LEDGF/p75 binding pocket of the IN CCD dimer (PDB: 3LPU). (A) Cartoon representation of the CCD IN dimer (PDB: 3LPU) in complex with compound 8 superimposed with the IBD‐IN core complex structure (PDB: 2B4J) shown in pink. (B) Co‐crystal structure of compound 8 bound to the IN CCD dimer shown as molecular surface. (C) Focus on compound 8–IN interactions. Individual IN subunits are depicted in light green and light cyan, compound 8 (white) and the key IN CCD residues (light green and light cyan) are represented as sticks, with nitrogen, oxygen, and chlorine atoms highlighted in blue, red, and green, respectively; the catalytic triad residues are shown as yellow spheres.

Initial Insights Into the Multimodal Mode of Action of ALLINIs

In the search for new antiviral agents, Boehringer Ingelheim reported a series of quinoline‐based inhibitors identified through a high‐throughput screen targeting IN‐catalyzed 3′‐P [106].

The initial hit, referred to as BI‐A [100] (9, Figure 6), inhibited IN 3′‐P activity in vitro with micromolar potency, but lacked antiviral activity [100] (Table 1). Subsequent optimization led to more potent analogs, including derivative BI‐1001 [99] (ALLINI‐1, 10, Figure 6), which exhibited enhanced 3′‐P inhibition and antiviral efficacy [97, 107].

To gain insights into their mechanism of action, derivative 8 was compared with compound 10 [99]. Both compounds inhibited the IN–LEDGF binding and the LEDGF‐independent catalytic activities of IN demonstrating antiviral efficacy in infected cells, with derivative 10 proving more active than compound 8 (Table 1).

X‐ray crystal structures of the compounds bound to the IN CCD dimer revealed nearly overlapping binding modes, indicating that the substituted benzene rings mainly form hydrophobic contacts with one IN monomer, while the carboxyl groups establish hydrogen bonds with residues E170 and H171 of the second monomer (Figure 5). Furthermore, compound 10 was found to establish an extra hydrogen bond between its methoxy group and residue T174, potentially explaining its greater potency compared with compound 8 in both LEDGF‐dependent and LEDGF‐independent assays. The strong similarity in binding modes and biological profiles of the two derivatives supported their classification as ALLINIs. Interestingly, further investigations demonstrated that both compounds promote aberrant IN multimerization by stabilizing IN–IN interactions. The resulting multimers are unable to form the SSC, leading to loss of catalytic activity and impaired downstream LEDGF binding. These findings highlighted a multimodal mechanism for ALLINIs, in which IN multimerization represents the primary event, while disruption of IN–LEDGF interaction and inhibition of catalytic activity are likely secondary consequences, supporting IN multimerization as a promising strategy for the development of ALLINIs [97, 99].

To investigate the structural and mechanistic properties of A128T IN resistance to ALLINIs, derivative 10 was examined alongside its tert‐butyl analog BI‐B2 [91] (ALLINI‐2, 11, Figure 6). Substituting the methoxy group with a tert‐butoxy moiety increased thinhibitory potency against 3′‐P and ST, as well as antiviral efficacy, by nearly 10‐fold. Furthermore, this modification tripled the compound's activity in inhibiting the IN–LEDGF/p75 interaction (Table 1). Conversely, the A128T substitution of IN conferred resistance to 11 compared to the wild‐type (WT) virus (11: EC50 = 12.2 µM); in addition, while the compound exhibited remarkably lower potency in inhibiting the 3′‐P and ST reactions, its impact on the IN–LEDGF/p75 interaction was less pronounced, resulting in only a twofold increase in IC50 value. In contrast, a more marked effect was highlighted on IN multimerization, and the A128T substitution was found to produce enhanced resistance to ALLINI‐induced aberrant multimerization of IN compared with WT IN, evidenced by a nearly 10‐fold increase in the IC50 values. To elucidate the structural basis of resistance, the crystal structures of 10 bound to either WT or A128T IN CCDs were compared. Interestingly, although the interaction networks between the inhibitor and the protein were largely preserved, the alanine‐to‐threonine substitution was found to affect the positioning of the quinoline core and the 4‐substituted phenyl ring of 10, likely accounting for the differential multimerization of WT and mutant IN [91].

In this framework, a series of 2‐(tert‐butoxy)‐2‐(4‐phenylquinolin‐3‐yl)acetic acids (tBPQAs), originally designed by Boehringer Ingelheim [107, 108] and subsequently licensed to Gilead Sciences, was explored [88, 102]. Compounds GS‐A, GS‐B, and GS‐C [102] (12–14, Figure 6) displayed remarkable antiviral efficacy in cellular assays, with EC50 values ranging from submicromolar to nanomolar concentrations (18 nM < EC50 < 55 nM, Table 1) in MT‐4 cells, with compound 13 emerging as the most potent derivative. In resistance selection experiments, compounds 12 and 13 promoted the emergence of mutations clustered at or near the IN dimer interface, such as A128T and T174I. Notably, the resistant viral strains showed cross‐resistance within the tBPQA series but retained susceptibility to RAL, supporting the conclusion that tBPQAs work through a mechanism distinct from that of classical INSTIs. Co‐crystal structures of the IN CCD in complex with compounds 12 and 13 (PDBs: 4E1M and 4E1N) revealed that these molecules bind at the CCD dimer interface, occupying the same pocket targeted by the LEDGF/p75 IBD. Notably, most of the residues identified as mutated in resistance selection studies are located within the inhibitor‐binding site. Specifically, A128 lies close to the quinoline scaffold, while T174 interacts with the carboxylate, tert‐butyl, and 4‐phenyl substituents of the ligands.

Investigations into their mode of action indicated that while the derivatives proved to inhibit HIV‐1 integration by elevating the level of 2‐LTR circles and decreasing the number of integration junctions in HIV‐1‐infected cells, the 3′‐P and ST steps were ruled out as targets of inhibition. Instead, the compounds proved to compete with LEDGF for binding to IN dimers, highlighting nanomolar potency consistent with their antiviral EC50 values. In addition, similar concentrations were found to promote IN dimer formation, suggesting that these specific interactions underlie the antiviral effects of tBPQAs. In addition to being one of the first reports of ALLINIs achieving antiviral efficacy in the nanomolar range, this study crucially provided innovative insights into the multimodal mode of action of ALLINIs [102].

In the search for new ALLINIs featuring alternative scaffolds, the quinoline moiety was modified into a thienopyridine core, and two analogs of derivative 8 were reported [103]. While derivative CX05045 [103] (15, Figure 6) retains the 2‐propyl group of 8, compound CX14442 [103] (16, Figure 6) features a 2‐tert‐butoxy moiety, a key hallmark of tBPQAs, to optimize the occupancy of the hydrophobic binding pocket and specifically mimic the IBD residue I365. Interestingly, both thienopyridines proved more active than the quinoline‐based analog, with derivative 16 exhibiting a 10‐fold increase in potency over 15 regarding both LEDGF/p75–IN interaction inhibition and antiviral efficacy [103] (Table 1).

To pinpoint the stage of the virus life cycle affected by derivative 16, time‐of‐addition (TOA) experiments were performed, obtaining a profile consistent with INSTIs which indicated interference with integration. However, unlike INSTIs, the compound displayed submicromolar activity against both the 3′‐P and ST reactions. Additional studies showed that compound 16 also influences the IN multimerization, by stabilizing the interaction between two IN monomers to likely limit the dynamic rearrangements required for viral DNA binding and thereby impair the intasome assembly. Interestingly, functional studies proved that cells treated with 16 produced noninfectious viral particles, highlighting a long‐term impact of ALLINIs on viral replication. Moreover, the compound displayed potential synergy without cross‐resistance with INSTIs, and no antagonistic effect was observed in combination with RAL [103].

To explore the contribution of LEDGF/p75 to the antiviral activity of ALLINIs, the potency of the 4‐(chroman‐6‐yl)quinoline ALLINI BI‐D [100, 108] (17, Figure 6) was explored in WT and Psip1 (i.e., the gene encoding for LEDGF/p75) KO mouse embryonic fibroblast cells [109]. Interestingly, the depletion of LEDGF/p75 from target cells significantly increased the antiviral activity of the compound, with the IC50 decreasing from 2.9 μM in WT cells to 0.16 μM in Psip1 KO cells. This finding suggested that LEDGF/p75 can compete with the ALLINI for binding to IN during the acute phase of HIV‐1 infection. However, the gain in potency observed upon the removal of LEDGF/p75 importantly raised questions regarding the relevance of the IN–LEDGF/p75 interaction in determining compound potency, indicating that this interaction is not a critical drug target for the compound [109].

Additional key insights emerged by comparing the overall antiviral activity of ALLINIs 10 and 17 with their effects on the early and late phases of the HIV‐1 life cycle [110]. Strikingly, despite its efficacy in a spreading HIV‐1 replication assay (EC50 of 5.8 μM), compound 10 did not show an antiviral effect on the early phase of infection (EC50 > 50 μM), while demonstrating potency in virus producer cells (EC50 of 1.9 μM). Similarly, compound 17 exhibited significantly higher potency in virus producer cells than in target cells, displaying an EC50 value that closely matched the value determined in the spreading replication assay (EC50 in spreading replication assay = 90 nM; EC50 in producer cells = 89 nM), indicating that compound 17 potency is principally accounted for during the late phase of HIV‐1 replication. In addition, analysis of virion morphology by thin‐section electron microscopy revealed that ALLINI treatment significantly enhanced the formation of eccentric HIV‐1 cores, indicating an underlying effect on viral core maturation.

Furthermore, to address if LEDGF/p75 affects 17 potency, the antiviral efficacy of the compound was evaluated in LEDGF/p75‐KO versus control HEK293T target cells. Consistent with previous findings [109], LEDGF/p75‐KO in target cells yielded a significant increase in compound potency. In contrast, LEDGF/p75 depletion did not affect the antiviral potency of the compound in virus producer cells [110].

These findings fundamentally provided key insights into the underlying mechanism of ALLINI action, highlighting that their potency is mainly due to their effects on the late stage of HIV‐1 replication, rather than solely on the inhibition of the IN–LEDGF/p75 interaction during the early stage, culminating in virion morphological defects.

Unraveling the Antiviral Effects of NCINIs and INLAIs

Further insights into the underlying mechanism of action of ALLINIs emerged from investigations on tBPQAs 12 and 13 (Figure 6). Owing to their dual mode of action, comprising the disruption of IN–LEDGF/p75 interaction and the IN dimer promotion, these compounds have been categorized as NCINIs [111]. A two‐part assay system was used to measure the antiviral potency at the late vs. early phase of the virus life cycle by altering the presence of compounds in cultures of virus producer and target cells. Consistent with previous findings [110], a striking functional divergence from RAL emerged during virological profiling: both compounds showed lower potency during the target cell infection phase compared to full‐cycle assay (12: EC50 vs. full cycle = 75.7 nM, EC50 vs. target cells = 6093 nM; 13: EC50 vs. full cycle = 26.4 nM, EC50 vs. target cells = 743.5 nM). Conversely, when exposure was restricted to the virus‐production stage, the antiviral activity closely mirrored the full‐cycle results (12: EC50 vs. producer cells = 84.1 nM; 13: EC50 vs. producer cells = 39.4 nM). This pronounced sensitivity during the late stages of the HIV‐1 replication cycle, coupled with the observation that the IN dimer‐interface mutation T174I confers resistance to these compounds but not to RAL, corroborated that ALLINIs and INSTIs operate through fundamentally distinct inhibitory mechanisms.

Crucially, experimental evidence demonstrated that these compounds act primarily during virus production and/or maturation, while mature virions remained refractory to these inhibitors. Furthermore, consistent with previous findings [109, 110], the variation of LEDGF/p75 expression levels in the virus producer cells had no discernible effect on either viral infectivity or compound potency. This finding indicated not only that LEDGF in producer cells is not essential for HIV‐1 infectivity, but also that the late‐stage inhibitory effect of these compounds is independent of LEDGF. Further investigations clarified that these derivatives do not interfere with Gag/Gag‐Pol processing, RT activity, or viral entry. Instead, WT virions produced in the presence of 13 displayed defective morphological defects in core and reduced infectivity, while the T174I mutation was found to revert such effects. Additionally, 12 was found to affect the oligomeric state of IN, favoring a higher proportion of IN dimers in newly produced viruses. Conversely, no such shifts were observed upon incubating mature, cell‐free virus with the compound, further corroborating the major effect of the compound on viral morphogenesis [111].

In‐depth analyses were performed to dissect particle maturation as the main step underlying the potency of quinoline‐based ALLINIs 11 and 17 [112]. Consistent with previous findings [110, 111], virus produced in the presence of the compounds displayed marked impairment of capsid morphogenesis with a high proportion of virions exhibiting “eccentric” condensates (i.e., electron‐dense aggregates located outside the capsid core). Moreover, the effect of drug treatment on reverse transcription activity closely paralleled the inhibition of mature core formation, and in turn the overall antiviral activity, indicating that their antiviral effect is due to inhibition of particle maturation. Cryoelectron tomography was used to characterize ALLINI‐treated particles, revealing important morphological differences compared to WT. Indeed, “eccentric” condensate‐containing virions from ALLINI treatment displayed higher incidence of nonconical than conical cores, indicating impairment of capsid morphogenesis. Moreover, most of the conical cores appeared relatively empty, revealing deficient incorporation of the viral ribonucleoprotein complex (RNP, assumed to be composed primarily of viral RNA, vRNA, and nucleocapsid protein NC) into the mature core, leaving it to remain outside the capsid shell. Additionally, “tomo‐bubblegram” imaging indicated that “eccentric” condensates have a high NC content, confirming that they represent nonpackaged RNPs [112].

Furthermore, to assess the role of IN in HIV‐1 maturation and the impact of ALLINIs, IN‐deficient particles were transcomplemented with IN fused to the viral accessory protein Vpr. This approach partially rescued the fraction of virions harboring WT cores with internal electron density, while this rescue effect was abolished by ALLINI treatment. Moreover, the genomic vRNA was found to be necessary for ALLINI‐induced eccentric condensate formation, as the fraction of virions exhibiting “eccentric” aggregates did not change upon ALLINI treatment when transfected with RNA‐packaging‐deficient constructs. Overall, these findings demonstrated that IN plays a direct role in initiating core morphogenesis and incorporating the RNP into the mature core during HIV‐1 maturation, in addition to critically suggesting the relevance of the IN–vRNA interaction for core morphogenesis [112].

Additional important findings emerged from investigations on a distinct series of aryl and heteroaryl 2‐(tert‐butoxy)acetic acid derivatives referred to as INLAIs [105]. The compounds were found to disrupt the interaction between the IN CCD and the LEDGF IBD as well as the interaction of full‐length LEDGF with IN. Consistently, the derivatives demonstrated significant antiviral activity, which closely correlated with their ability to disrupt IN–LEDGF interactions at submicromolar concentrations (Table 1). Among the series, Mut101 [105] (18, Figure 6) emerged as the most effective derivative, acting as the strongest inhibitor of IN–LEDGF binding and displaying the highest antiviral potency. Compound 18 was also able to interfere with LEDGF‐independent IN function, through inhibition of the ST reaction at concentrations comparable to those disrupting the IN–LEDGF interaction, despite resulting less potent than RAL. Co‐crystal structure of compound 18 bound to the IN CCD dimer (PDB: 4LH5) showed that it engages the LEDGF‐binding pocket, establishing interactions consistent with those reported for previous inhibitors.

However, unlike previously reported ALLINIs, compound 18 and the related quinoline derivative 17 (Figure 6) retained full antiretroviral activity against the HIV‐1 mutant bearing the IN A128T mutation, and the compounds proved to stabilize both the WT and mutant enzyme in a tetrameric form. In addition, compound 18 showed efficacy against a broad panel of drug‐resistant viral strains, including INSTI‐resistant variants. Interestingly, while TOA experiments indicated interference with integration, the compound demonstrated a more potent antiviral effect during the late, postintegration phase of the replication cycle. Differently, the HIV‐1 T174I mutant (i.e., mutated in the LEDGF‐binding pocket) exhibited reduced sensitivity to compound 18, resulting in the production of infectious virions [105].

Although a postintegration block in virus‐producer cells was found to be the primary mechanism of its antiretroviral activity, the compound's dual effect on both integration and postintegration was attributed to its binding at the LEDGF‐binding pocket of IN. This discrepancy in activity was explained based on the cellular localization of the cofactor: in the nucleus of target cells, LEDGF outcompetes the compound for the IN binding site [109], reducing its efficacy during integration. Conversely, in the cytoplasm (where postintegration assembly of infectious particles occurs) LEDGF is absent, allowing the compound to exert its full effect [105].

Toward the Search of Effective Quinoline‐Based ALLINIs: The Discovery of BI 224436

In the search for effective ALLINIs, extensive hit‐to‐lead optimization efforts were conducted by Boehringer Ingelheim to refine the structural determinants governing activity in quinoline‐based inhibitors. These studies established meaningful SARs within this class of compounds, culminating in the discovery of BI 224436 [96] (19, Figure 6), the first ALLINI to advance into Phase IA clinical trials [106, 113].

Starting from compound 9 (Figure 6), which was identified as 3′‐P inhibitor through high‐throughput screening, SAR studies were carried out by introducing modifications at three main regions of the molecule: the α‐position of the C3 acetic acid moiety, the arene at C4, and the B‐ring (i.e., the phenyl ring fused to the pyridine moiety of the quinoline scaffold) (Figures 8 and 9).

FIGURE 8: Structural optimization to yield compound 19.

FIGURE 9: Structural features of ALLINI binding. View of the ALLINI binding pocket and crystallographic organization of IN subunits.

In addition to exploring 3′‐P inhibition, structural insights from X‐ray crystallography studies were used to optimize interaction networks of the quinoline‐based inhibitor and the IN CCD dimer [79, 91, 99] (Figure 9).

In this framework, early optimization efforts focused on modifications at the 6‐position. While replacing the 6‐chlorine of 9 with a bromine atom improved the potency against 3′‐P, the deschloro analog exhibited lower activity. Additionally, the antiviral efficacy of 6‐bromine‐substituted derivatives was further enhanced by introducing a methyl or a chlorine group at para‐position of the 4‐phenyl group. By keeping untouched the 6‐Br and 4‐(p‐chlorophenyl) substituents, functionalization at the α‐position of the C3 acetic acid moiety was found to be generally beneficial for potency. While the methyl‐substituted derivative retained comparable activity to the acetic analog, further enlargement of the alkyl substituent led to marked improvements in antiviral activity. In particular, the insertion of a n‐propyl group yielded a derivative (i.e., the α‐propyl analog of 10) displaying potency against 3′‐P (IC50 = 0.077 µM in the LTR‐cleavage assay) and antiviral efficacy (EC50 = 0.92 µM) at submicromolar concentrations, together with negligible cytotoxicity. Based on these encouraging results, a broader range of α‐substituents was examined, revealing that alkoxy groups could confer additional gains in potency (Figure 8). Given the robust activity highlighted by the methoxy derivative (10), the racemic mixture was subsequently resolved into its individual enantiomers, with the S‐enantiomer possessing superior intrinsic and antiviral activity than the racemate.

Progressive enlargement of the alkoxy group at the C3 position to establish additional interactions within the target pocket led to the identification of the tert‐butoxy group as the optimal substituent. In particular, the S‐enantiomer of 11 proved antiviral efficacy in the nanomolar range (EC50 = 78 nM) and was used as a starting point for further optimizations [106].

To enhance the antiviral properties and explore additional binding pocket interactions, further derivatives were obtained by preserving the α‐ tert‐butoxy and the 4‐(p‐chlorophenyl) groups while replacing the 6‐Br with alternative substituents. SAR analysis revealed that the introduction of bulky aryl substituents at this position negatively affected the scaffold's multimerization‐inducing properties, leading to a progressive loss of activity. Similar detrimental effects were observed by inserting substituted phenyl rings and other bulky groups, including pyridinyl, six‐membered nonaromatic heterocycles, and five‐membered aromatic heterocycles [114]. At the same time, X‐ray analysis revealed that removing the 6‐bromine substituent caused a shift in the quinoline scaffold. This repositioned the C5 atom closer to A128 and drove the C4 substituent deeper into the highly conserved region of the binding pocket. Furthermore, derivatives lacking the 6‐bromine atom and carrying a 7‐methyl group showed antiviral efficacy against strains with 124 and 125 polymorphisms, leading to the adoption of these modifications for lead optimization [106].

Modifications to the 4‐position of the quinoline scaffold were also explored to enhance interactions with the hydrophobic pocket formed by residues W132 and L102 of subunit 2 of IN, with the aim of improving the compounds’ multimerization efficacy (Figure 9) [97]. Aromatic substituents at this position were generally favored, with substituted phenyl groups emerging as optimal. A broad exploration of para‐substituted phenyl derivatives revealed that incorporation of fluorine or fluorinated groups, such as trifluoromethyl and trifluoromethoxy, led to modest improvements in compound multimerization efficacy, whereas para‐methyl or para‐methoxy substitutions produced approximately sixfold gains in multimerization activity. The para‐chloro analog achieved the most potent inhibition within this series (EC50 ~ 100 nM in the multimerization assay), attributed to a stabilizing chlorine–π interaction with W132 (Figure 9). In contrast, alternative para‐substituents such as cyano, acetyl, acetamido, or phenyl groups reduced potency, highlighting the steric constraints of the hydrophobic pocket. Meta‐ or ortho‐substituted phenyl derivatives were less effective than their para counterparts because these positions orient the substituents away from the W132/L102 pair, attenuating hydrophobic interactions. For most substituents, the observed potency trend was _para _> _ortho _> meta. Attempts to combine para and meta substituents into disubstituted phenyl groups did not significantly improve compound multimerization efficacy, with activity either similar to that of the para isomer or even worse [97].

A more in‐depth analysis of the C4 substituent proved to be the most productive strategy for enhancing antiviral potency, as larger ring systems were generally more effective at occupying the binding pocket than smaller pentadiene‐like heterocycles such as furan or thiophene, thereby leading to increased multimerization potential. In particular, the introduction of a chromane moiety at C4 markedly improved both in vitro IN multimerization and antiviral efficacy in the nanomolar range [97, 106]. Further analogs incorporating related bulky ring systems, such as benzodioxane and substituted chromane motifs, confirmed that larger C4 arenes were particularly effective at filling the conserved binding pocket. Progressive occupation of this region also reduced the fold shift between EC50 values measured across IN variants carrying different aa124/aa125 residues, indicating a more robust interaction profile. Mechanistic investigations of the chromane‐substituted derivative confirmed its categorization as NCINI, in line with its ability to interfere with the IN–LEDGF interaction and modulate IN multimerization.

Additionally, introducing substituents that restricted rotation around the C4—C(arene) bond was found to further improve the antiviral potency, as highlighted by stable atropisomers. A key advancement was achieved through the hybridization of chromane and quinoline features, leading to tricyclic C4 arenes that possessed potent antiviral efficacy across IN variants while maintaining favorable metabolic stability (Figure 8). Finally, the removal of the 7‐methyl group of the quinoline scaffold resulted in improved serum shift, excellent metabolic stability, and favorable in vitro ADME profile, giving rise to compound 19 [106] (Figure 8). Interestingly, the derivative exhibited nanomolar antiviral activity against a broad range of mutant viruses, including variants carrying substitutions at the highly polymorphic IN residues 124 and 125 (11 nM < EC50 < 27 nM). Notably, given its excellent PK profile in animal models, such as high oral bioavailability and low clearance, compound 19 advanced into Phase I clinical trials, marking a major milestone in the field as the first ALLINI to undergo clinical evaluation [115]. Although its clinical development was later halted for undisclosed reasons, compound 19 has remained an important research tool, contributing significantly to the understanding of the mechanistic basis of ALLINI‐mediated inhibition [116, 117].

Overall, modifications of the quinoline scaffold, summarized in Figure 10, yielded important insights into the therapeutic relevance of ALLINIs, in addition to serving as key investigational tools to elucidate key mechanistic evidence, as discussed in the following paragraphs.

FIGURE 10: Overview of the main structural modifications introduced on the quinoline scaffold to investigate SARs. Substituents at different positions of the core were systematically varied to assess their impact on IN function, LEDGF–IN interaction, and antiviral activity.

Nevertheless, SAR studies demonstrate how specific structural features and the fine‐tuning of key substituents have enabled effective interference with the IN–LEDGF interaction, modulation of IN multimerization, and potent antiviral efficacy. The balance between these effects appears to be governed by the nature and steric properties of the substituents, which dictate pocket occupancy and the stabilization of specific IN oligomeric states.

Optimizing ALLINIs for Targeting IN Multimerization

The studies described in the previous sections provided important insights into the mechanism of action of ALLINIs and simultaneously delineated relevant SARs for quinoline‐based inhibitors. Strikingly, although most of these compounds were originally designed to inhibit the LEDGF/p75–IN interaction, the observation that they can induce aberrant IN multimerization, likely the key determinant of their antiviral activity, prompted a more focused investigation of this alternative mechanism in the development of new antiretroviral agents. In this context, derivatives originally designed as putative disruptors of the LEDGF/p75–IN interaction were more extensively evaluated for their potential to induce multimerization of IN. In addition, efforts were devoted to elucidating SARs that relate structural determinants with their inhibitory potency toward LEDGF/p75–IN binding and their efficacy in promoting IN multimerization, while prominent research focused on optimizing IN multimerization as a key determinant of ALLINI antiviral efficacy. Remarkably, these findings were greatly enhanced by structural insights, significantly supporting the understanding of the molecular basis for ALLINI efficacy, as discussed in the following paragraphs.

Pyridine‐Based Inhibitors

To specifically examine the role of HIV‐IN multimerization independently of the IN–LEDGF/p75 interaction, a series of selective multimerization inhibitors, referred to as MINIs, were designed and synthesized (Figure 11) [101]. Structural comparison of the crystal structures of compound 10 (Figure 6) (PDB: 4DMN) and the LEDGF/p75 IBD (PDB: 2B4J) bound to HIV‐1 IN CCD dimers revealed that both ligands occupy the LEDGF/p75 binding pocket and act as molecular bridges between the two IN subunits (Figures 7 and 12A). Specifically, each molecule establishes hydrogen bond interactions with one subunit (subunit 2) through contacts with residues E170 and H171, while the methoxy group of compound 10 further interacts with residue T174. In contrast, interactions with the other subunit (subunit 1) are mainly hydrophobic and appear constrained by the rigid, planar nature of the quinoline core (Figures 7 and 12A). To strengthen contacts with this subunit, the quinoline scaffold was subsequently modified into a pyridine‐based core, leading to the development of derivatives KF115 [101] (20, Figure 11) and KF116 [101] (21, Figure 11). X‐ray crystallographic analyses of compounds 20 and 21 in complex with the HIV‐1 IN CCD dimer (PDBs: 4O0J, 4O55) showed that, although interactions with subunit 2 are comparable to those observed for compound 10, hydrophobic contacts with subunit 1 are strengthened, largely due to the increased molecular flexibility of these derivatives. Moreover, the benzimidazole moiety of compound 21 establishes an additional interaction with subunit 1 through a hydrogen bond with residue T125 (Figure 12).

FIGURE 11: Chemical structure of ALLINIs 20–24.

FIGURE 12: Crystal structure of compounds 10, 20, and 21 bound to HIV‐1 IN CCD dimer interface (PDBs: 4DMN, 4O0J, 4O55). Cartoon representation of 10 (A), 20 (B), and 21 (C) bound to the IN CCD dimer. The IN subunit 1 and 2 are colored in light cyan and light green, respectively, compounds are represented as white sticks, with nitrogen, oxygen, bromine, and chlorine atoms highlighted in blue, red, brown, and green, respectively.

Notably, structural modifications of quinoline 10 significantly altered the biological profiles of the pyridine‐based compounds. In particular, the new derivatives exhibited higher potency and selectivity (up to 60‐fold) toward inducing IN multimerization over disrupting the LEDGF/p75–IN interaction (Table 2), in contrast with earlier ALLINIs like 10, which displayed comparable activity toward both processes (10: EC50 for aberrant IN multimerization = 4.9 µM; IC50 for IN–LEDGF/p75 binding = 1.0 µM) [101]. Interestingly, the compounds showed antiviral effects at concentrations similar to those inducing IN multimerization, suggesting that such event, rather than the inhibition of the LEDGF/p75–IN interaction, drives their efficacy (Table 2).

Compounds | IC 50 IN–LEDGF binding, μM a | EC 50 for aberrant IN multimerization, μM b | MT‐4 | References
Antiviral efficacy (EC 50) μM c | CC 50 , μM d
20 | 13.0 ± 1.5 | 0.274 ± 0.025 | 0.121 ± 0.004 | >100 | [101]
21 | 5.03 ± 0.36 | 0.086 ± 0.006 | 0.024 ± 0.003 | >100 | [101]
22 | 0.095 ± 0.005 | 0.052 ± 0.003e | 0.031 ± 0.009 | 42 ± 9 | [104]
23 | 0.19 | 0.37e | 0.45 | ND | [104]
24 | 0.0082 ± 0.00014f | ND | 0.038 | ND | [118]

While compound 21 proved more potent than 20 (Table 2), its antiviral efficacy was attenuated against the HIV‐1 IN T125A mutant, validating the functional relevance of the hydrogen bond established with the benzimidazole moiety (Figure 12). In contrast, the compound proved to maintain efficacy against the challenging HIV‐1 IN A128T mutation, effectively promoting IN multimerization and impairing HIV‐1 replication despite this substitution. While such mutation typically introduces steric and electronic constraints that hinder the binding of quinoline‐based ALLINIs (e.g., derivative 10) [91], the alternative orientation of the benzimidazole moiety of compound 21 allows it to successfully circumvent these steric effects. Consistent with previously reported ALLINIs [105, 110, 111, 112], compound 21 was more effective in producer than target cells (EC50 vs. target cells = 50.9 µM; EC50 vs. producer cells = 30 nM) at concentrations similar to those seen in the full replication cycle (EC50 = 24 nM). In addition, morphological analysis of producer cells treated with 21 revealed an impaired formation of electron‐dense cores, resulting in the production of virions characterized by “eccentric” cores lacking RNPs [101, 112]. Further experiments indicated that the binding of 21 to the IN CCD dimer interface stabilizes the IN subunits, effectively shifting the equilibrium toward aberrant, higher‐order oligomerization.

Interestingly, comparing the antiviral profiles of 21 and the quinoline‐based ALLINI 13 [102] revealed that while both compounds primarily target the late stage of viral replication, 21 is a more potent postintegration inhibitor. In contrast, 13 displayed a more pronounced effect on target cells versus producer cells compared to 21 (13: IC50 vs. target cells ~0.7 µM; 21: IC50 vs. target cells > 50 µM), which notably aligns with its higher potency in disrupting the IN–LEDGF/p75 binding relative to the pyridine‐based inhibitor (Tables 1 and 2).

Overall, these findings highlighted important divergent functional features for the two classes of compounds, indicating that the superior antiviral efficacy of the pyridine derivatives stems from their capacity to induce IN multimerization rather than interfering with LEDGF/p75 binding, consistent with their higher potency in producer cells.

In addition to compound 21, other pyridine‐based analogs have been reported as potential ALLINIs, showing improved drug‐like characteristics compared with quinoline derivatives in ADME and in vivo PK studies [119, 120]. Nevertheless, despite these encouraging properties, none of these compounds, including derivative 21, has advanced to clinical evaluation.

Thiophene‐ and Isoquinolone‐Based Inhibitors

In the pursuit of diversifying the chemical space of ALLINIs, a novel set of thiophene derivatives was identified as structural alternative to established quinoline classes [121]. Optimization of this series, driven by the necessity to maintain the pharmacophoric tert‐butoxy acetic acid moiety common to potent ALLINIs, led to the discovery of the lead compound MUT‐A [121] (22, Figure 11). This inhibitor, characterized by a five‐membered thiophene ring substituted with methyl, a gem‐dimethylcyclohexenyl, and 4‐pyridinyl moieties at positions 2, 4, and 5, respectively, of the central core, demonstrated a potent dual mechanism of action. Biochemically, nanomolar concentrations of 22 effectively disrupted the interaction between IN and LEDGF/p75, while concomitantly inducing the aberrant multimerization of the viral protein. These biochemical properties translated into robust antiretroviral activity, with the compound exhibiting antiviral efficacy within the nanomolar range with negligible cytotoxicity (Table 2). Consistent with previous findings, virions produced in the presence of 22 displayed severe infectivity defects and were noninfectious [112, 121].

However, the therapeutic potential of 22 was challenged by its sensitivity to natural polymorphisms within the IN CCD, especially at the highly variable residues 124 and 125 [104]. Interestingly, detailed virological profiling revealed that while 22 retained potency against variants bearing T125, its antiviral efficacy was drastically reduced, by approximately 50‐fold, against viruses harboring an alanine at this position. Crucially, the latter represents a highly prevalent polymorphic form, particularly within specific viral clades, which may inherently compromise the clinical utility of inhibitors sensitive to this substitution. Indeed, sequence analysis indicates that these A125‐harboring strains are dominant [122, 123]. Mechanistic decoupling studies using the A125 variants revealed a critical insight into ALLINI pharmacology: while 22 retained its ability to bind the allosteric pocket and inhibit the IN–LEDGF/p75 interaction, it completely failed to induce the requisite aberrant multimerization of the A125 IN variant. This observation underscored that affinity for the LEDGF‐binding pocket is necessary but insufficient for the late‐stage antiviral effect, which strictly depends on the induction of specific conformational changes leading to oligomerization. Crystallographic studies of the IN CCD complexed with 22 elucidated the structural basis of these findings. The X‐ray structures (PDBs: 5OI2 and 5OI8) indicated that in the context of the A125 variant, 22 binding stabilized the CCD dimer interface, preventing the formation of the aberrant higher‐order oligomers required for the inhibition of viral maturation. Conversely, in the WT T125, ligand binding destabilized the dimer, facilitating multimerization (Figure 13). Guided by these structural insights, a SAR campaign focused on the substituent at position 5 of the thiophene core was undertaken to restore activity against the A125 variant [104]. This effort led to the identification of MUT‐A03 [104] (23 Figure 11), an analog where the 4‐pyridyl moiety of 22 was replaced by a 2‐pyridyl group. Structural comparison (PDBs: 5OI2 and 5OI5) showed that unlike 22, where the pyridine nitrogen is solvent‐exposed, the nitrogen in 23 is buried within the binding site (Figure 13B). Crucially, protein interfaces, surfaces, and assemblies analysis predicted that 23 maintains the ability to destabilize the CCD dimer even in the A125 variant, correlating with its recovered ability to promote IN multimerization and its improved antiviral efficacy against this polymorphic variant compared to the parent compound. Although 23 successfully overcame the specific resistance conferred by the A125 polymorphism, it exhibited a trade‐off in potency against the WT T125 virus, highlighting the delicate structural balance required to engage the allosteric pocket effectively across diverse viral genetic backgrounds. Consequently, although these compounds provided a fundamental proof‐of‐concept (POC) for dissociating the biochemical activities of ALLINIs, their development was ultimately halted due to this inconsistent potency profile across different viral genotypes.

FIGURE 13: Crystal structures of compounds 22 (A) and 23 (B), bound to HIV‐1 IN CCD dimer interface (PDBs: 5OI2 and 5OI8). CCD dimer chains are represented with cartoons and colored in light cyan and light green, respectively. The catalytic triad residues are shown as yellow spheres. Compounds 22 and 23 are represented as white sticks, with nitrogen, oxygen, and fluorine atoms highlighted in red, blue, and pale cyan, respectively.

While the thiophene‐based series explored the impact of core diversification on polymorphic variants, parallel drug discovery efforts continued to refine the more traditional bicyclic architectures. Indeed, significant advancements were also achieved by modifying the quinoline core into more complex fused systems. A prominent example is the isoquinolinone derivative GSK1264 [118, 124] (24, Figure 11). The compound potently disrupted the LEDGF/p75 IBD–INF185K CCD interaction (IC50 ≈ 6.3 nM) and inhibited HIV‐1 replication in MT‐4 cells with an EC50 of ~38 nM in a multicycle assay. X‐ray crystallography of 24 bound to the IN CCD dimer interface (PDB: 4OJR) showed occupancy of the LEDGF pocket and engagement of both IN monomers through hydrophobic and polar interactions consistent with other ALLINIs (Figure 14). In these structural studies, the INF185K variant was employed to improve protein solubility and reduce nonspecific aggregation while preserving functional and binding properties, thereby enabling crystallographic characterization of the inhibitor‐bound complex [125, 126, 127].

FIGURE 14: Crystal structure of compound 24 bound to HIV‐1 IN CCD dimer interface (PDB: 4OJR). CCD dimer chains are represented with cartoons and colored in light cyan and light green, respectively. The catalytic triad residues are shown as yellow spheres. Compound 24 is represented as white sticks, with nitrogen, oxygen, and fluorine atoms highlighted in red, blue, and pale cyan, respectively.

As observed for related derivatives, 24 displayed lower antiviral activity during early infection compared with its potency in multicycle and late‐stage assays. Evaluation of LEDGF dependence revealed only a modest increase in activity in LEDGF KO versus WT cells (≈5‐fold), smaller than that reported for quinoline‐based ALLINIs such as 17 (≈30‐fold). Overall, these results suggest that compound 24 is less effective at displacing LEDGF from IN and primarily exerts its antiviral activity by promoting aberrant IN multimerization. In line with this mechanism, the compound induced concentration‐ and time‐dependent IN polymerization, leading to the formation of insoluble aggregates and ultimately impairing the production of mature virions [125].

Key Insights Into the Structural Basis for ALLINI Function

Discovering that not only the CCD, but also the CTD plays a pivotal role in driving IN hyper‐multimerization represented a key step forward in elucidating the underlying mode of action of ALLINIs [128, 129]. Initial molecular modeling studies suggested that these inhibitors directly bridge the interface between a CCD dimer and the CTD of an adjacent dimer. In the absence of an inhibitor, this interface is typically occupied by water molecules, which minimize direct CCD–CTD interactions. However, the binding of an ALLINI effectively fills the V‐shaped pocket, displacing water molecules and enhancing the binding interface for the incoming CTD [130].

This model was structurally validated using a full‐length HIV‐1 INY15A/F185H mutant, specifically engineered to improve solubility while retaining enough drug‐induced aggregation to allow for crystallization [131, 132] and bound to compound 24. The resulting structure (PDB: 5HOT, recently superseded by PDB: 8V9C [133]) provided a clear visualization of the open polymer architecture, where the inhibitor mediates head (CCD–CCD dimer) to tail (CTD of another dimer) interactions (Figure 15) [125]. This structure indicated that the bound compound 24 enhances and strengthens CCD–CTD interactions, promoting the formation of extended IN polymers. This event leads to aberrant multimerization, resulting in IN aggregation and subsequent enzyme inactivation, thus providing a rational explanation for the underlying mode of action of the compound [13].

FIGURE 15: Crystal structure of HIV‐1 INY15A/F185H bound to compound 24 (PDB: 5HOT). Cartoon representation of the open polymer formed upon binding of 24. Each monomer is presented in different colors and compound 24 located at CTD–CCD interfaces is shown as red spheres.

Expanding on this concept, further studies [134] demonstrated that these aggregates are not merely linear chains but complex branched polymers with fractal‐like properties. These insights revealed that while the inhibitors stabilize the CCD–CTD junction, the 3D network is further driven by homomeric CTD–CTD interactions, which act as branching points that transform the polymer into an inactive, gel‐like state.

These insightful findings paved the way for further structural investigations, many of which have focused on the quinoline‐based derivative 19 (Figure 6). Indeed, recent high‐resolution crystallographic data (PDB: 8CTA) [116] have refined the understanding of this compound's binding mode, revealing a surprising asymmetry in the ternary complex (Figure 16). The study identified two nonidentical binding sites: a well‐ordered Site 1 (volume ~781 Å3) and a more flexible, less compact Site 2 (volume ~971 Å3), which exhibits weaker electron density for key residues. Their structure highlighted a dense network of interactions. Specifically, W235 of the CTD provides a favorable π–π interaction with the aromatic pharmacophore of 19, while being itself stabilized by a cation–π interaction with R228. Furthermore, the molecule's carboxylate group is anchored through a series of hydrogen bonds involving the backbone amides of E170 and H171 in the CCD, as well as an extended rotamer of K266 in the CTD. A particularly insightful discovery was the identification of a novel binding pocket at the CCD–CTD interface, created by a significant conformational shift of Trp131, which rotates approximately 120° upon CTD binding (Figure 16B). In the crystal structure, this pocket is occupied by a molecule of ethylene glycol that lies in direct van der Waals contact with the inhibitor. This finding has a high strategic value, as it suggests that future drug optimization could involve the elaboration of chemical moieties designed to exploit this specific void, potentially increasing both affinity and the genetic barrier to resistance. Parallel to these findings, further studies [133] have shed light on how resistance emerges; mutations like W131C and N222K act as “second‐shell” perturbations that alter the quaternary orientation of the CTD. This geometric shift destabilizes the drug‐induced polymer, allowing the virus to maintain functionality even in the presence of the inhibitor [133].

FIGURE 16: Crystal structure of HIV‐1 IN bound to compound 19 (PDB: 8CTA). (A) Cartoon representation of two molecules of 19 located at CTD–CCD interfaces. (B) Focus on the novel binding pocket identified at the CCD–CTD interface. CCD dimer chains are colored in light cyan and light green, respectively, while CTD is represented in gold. The catalytic triad residues are shown as yellow spheres. Compound 19 is represented as white sticks. The two asymmetric sites are shown in red.

The specificity of these interactions was further explored by comparing the pyridine‐based ALLINI 21 and quinoline 19 [117]. While both derivatives induced higher‐order multimerization of full‐length WT IN tetramers, compound 19 showed a broader specificity, interacting with both dimers and tetramers in 2:2 ratio, whereas derivative 21 showed higher selectivity for tetramers [125]. Superimposition of the co‐crystal structures of compound 21 or 19 in complex with IN CCD aided to rationalize these observations. Similarly to 24 (Figure 11), compound 19 (PDB: 6NUJ) was found to localize deeper inside the V‐shaped pocket, allowing symmetrical CCD–19–CTD interactions between both IN dimers and tetramers. Differently, analysis of the X‐ray structure of derivative 21 (PDB: 4O55) revealed that the bulky benzimidazole group projects outside this pocket, preventing dimer–dimer symmetry while optimally positioning two IN tetramers for symmetrical CCD–inhibitor–CTD interactions.

Interestingly, while compounds proved to induce higher‐order multimerization of only full‐length IN, truncated protein constructs displayed substantial resistance to aggregation. This finding indicated that also the NTD of IN plays a role in compound‐induced higher‐order oligomerization. Indeed, although the domain does not bind the inhibitor directly, site‐directed mutagenesis of the NTD and the CCD–CTD linker compromised tetramer stability and conferred resistance to compound 21‐induced oligomerization. This suggests that the NTD contributes to stabilizing the functional tetramer by interacting with the α‐helical linker connecting the CCD and CTD.

Finally, to explore a possible eutomer effect, single enantiomers of 21 were investigated. The enantiomer (−)−21 displayed higher antiviral efficacy than (+)−21 in MT‐4 cells ((−)−21: EC50 =6.28 nM); ((+)−21: EC50 = 203.9 nM) and demonstrated improved metabolic stability in rat and human microsomes compared with both racemic mixture and compound 24. Notably, the eutomer (−)−21 showed prominent antiviral efficacy against the DTG‐resistant mutant HIV‐1NL4−3 containing IN N155H/K156N/K211R/E212T substitutions (IC50 = 0.70 nM), highlighting the potential for ALLINIs to complement INSTI‐based HIV‐1 therapies [117].

Taken together, the growing understanding of ALLINI SARs and mechanism of action, greatly supported by structural insights, has unveiled their underlying basis of function, resulting in the design of increasingly potent and pharmacologically optimized derivatives. This progress has facilitated the transition from early lead compounds to optimized preclinical candidates, some of which have undergone clinical evaluation.

Emerging ALLINIs in Clinical Development

Over time, studies have clarified the structural and molecular basis of ALLINI function, identifying inhibitor‐induced IN multimerization as the key mechanism of action. However, how this process affected viral maturation and infectivity remained unclear until the discovery of a previously unrecognized, noncatalytic role of IN in HIV‐1 biology. Specifically, IN was shown to directly bind the viral RNA genome within virions, highlighting a fundamental role in viral morphogenesis [13, 135]. These findings provided unprecedented insights into the mode of action of ALLINIs, indicating that such inhibitors primarily act by inducing aberrant IN multimerization and disrupting IN–viral RNA interactions. This leads to the mislocalization of viral RNPs outside the capsid, occupying an eccentric position between the core and the viral membrane, in contrast to mature infectious virions where RNPs are enclosed within the capsid core [112]. Importantly, representative ALLINIs, such as 11 and 17 (Figure 6), were found to impair the binding of WT IN to viral RNA in virions [135]. Consequently, ALLINI‐treated virions display mislocalized RNP complexes and are noninfectious. Notably, nucleocapsid–RNA interactions remain unaffected, indicating that disruption of IN–RNA binding is a specific and critical determinant of the observed phenotype. Mechanistically, ALLINI‐induced multimerization likely sequesters key RNA‐binding residues within the IN CTD, thereby preventing productive interaction with the viral genome. These findings clarify the molecular basis of ALLINI activity and establish that their primary antiviral effect at the late stage of replication arises from the alteration of functional IN multimerization and the consequent disruption of IN–RNA interactions, thus resulting in defects in virion maturation [13, 112, 135, 136].

However, despite significant advances in unveiling the structural and functional basis of ALLINI efficacy, the search for effective in vivo antiretroviral agents has progressed more slowly. Indeed, important limitations in ALLINI efficacy have emerged, and the encouraging in vitro potency of these compounds has not always been translated into consistent in vivo outcomes. As a result, only a small subset of ALLINIs has progressed through preclinical development to reach the stage of clinical candidates. This discrepancy has been largely attributed to safety issues associated with experimental ALLINIs, whose development was halted due to toxicity observed in preclinical animal studies [137, 138, 139]. A representative example is the highly potent in vitro pyridine‐based ALLINI 21 (Figure 11) and its derivatives, for which no further clinical development has been reported, highlighting the challenges associated with translating potent ALLINI scaffolds into viable therapeutic candidates. Similar limitations have been observed across this class, where issues related to PK, toxicity, and overall in vivo performance have hindered clinical progression [117, 137, 138, 139].

Toward the Discovery of In Vivo Effective ALLINIs

Next‐generation ALLINIs such as GS‐9695 and GS‐9822 [138] (25 and 26, Figure 17) were developed to significantly improve antiviral potency and the genetic barrier to resistance compared with first‐generation candidates like 19 (Figure 6) [138]. The compounds share a common 2‐substituted benzothiazole core incorporating the α‐(tert‐butoxy)acetic acid and p‐chlorophenyl moieties typical of most ALLINIs, while featuring alternative substituents in position 2 of the central core. Optimization of the initial lead 25 through introduction of a substituted indazolyl group at the 2‐position resulted in compound 26, which exhibited significantly improved metabolic stability and an increased barrier to resistance. The compound proved to inhibit the IN–LEDGF/p75 protein–protein interaction, demonstrating a 13‐fold increase in potency over the thienopyridine 16 [137] which notably translated into exceptional antiviral efficacy in infected cells (Table 3). Beyond acute inhibition, 26 was found to induce a “block‐and‐lock” phenotype by retargeting residual integration away from methylation marks usually highly favored by HIV toward repressive heterochromatic regions, establishing a reservoir that is highly refractory to reactivation. However, despite its prominent profile, the clinical development of these rigid scaffolds was halted due to species‐specific urothelial toxicity observed in cynomolgus monkeys, characterized by transitional cell vacuolation and bullous formation [142]. This toxicity is rooted in a unique structure‐toxicity relationship governed by the ionization state of the molecules. Indeed, 26, with pKa values of 4.2 (carboxylic acid) and 5.8 (piperidine/oxetane), and 25, with pKa values of 4 and 7.8, exist as rigid zwitterions within the acidic urinary pH range of primates, including humans (pH 5.5–7.4). In contrast, the alkaline urine of rats (pH 7.3–8.5) prevents this specific ionization. Surface activity measurements of 26 revealed an exotic transition from an adsorbed monolayer to a bilayer at the air/water interface specifically at pH = 5; this transition, driven by strong intermolecular associations between zwitterionic molecules, is highly disruptive to urothelial membrane integrity. Notably, this “detergent‐like” behavior is absent in in the quinoline‐based ALLINI 19, which lacks the piperidine/piperazine moieties and transitions directly from a neutral to a cationic species without entering a zwitterionic phase with decreasing pH [142].

FIGURE 17: Chemical structures of ALLINIs 25–30.

Compounds | IC 50 IN‐LEDGF binding, μM a | EC 50 for aberrant IN multimerization, μM b | Antiviral efficacy (EC 50 , μM) c | CC 50 (MT‐4, μM) d | References
25 | ND | ND | 0.0012 ± 0.0002 (MT‐4) | 7.1 ± 1.4 | [138]
26 | 0.07 ± 0.02 | ND | 0.0022 ± 0.0003 (MT‐4) | 4.7570 ± 0.5972 | [137]
27 | 0.190 ± 0.07 | 0.147 ± 0.02 | 0.00166 ± 0.00032e (MT‐4) | >10 | [139]
28 | ND | ND | 0.0045 ± 0.0002 (HEK293T) | ND | [140]
29 | 0.047 ± 0.014 | 0.020 ± 0.011f | 0.0087 ± 0.0028 (MT‐4) | 69 ± 8 | [141]
30 | 0.014 ± 0.005 | 0.031 ± 0.020f | 0.0031 ± 0.001 (MT‐4) | 46 ± 11 | [141]

The Discovery and Development of Pirmitegravir

In the search for potent ALLINIs with robust efficacy and safety profile in in vivo models, the identification of STP0404 [139] (pirmitegravir, 27, Figure 17) reported by ST Pharm (Seoul, South Korea) has marked a major milestone in the field [139].

Compound 27 retains the hallmark features of previously reported ALLINIs (i.e., a central scaffold with a 4‐substituted phenyl group along with a 2‐(tert‐butoxy)acetic acid moiety), while displaying a pyrrolopyridine core and an N‐methyl(pyrazolyl)methyl substituent as key structural points of divergence. The compound displayed excellent antiviral activity at low nanomolar concentrations with negligible cytotoxicity in human cells (Table 3). Due to its stereogenic center, the compound exists as two enantiomers; the S‐enantiomer displayed potency comparable to the racemate, while both outperformed the R‐enantiomer. Serial passage in the presence of the compound promoted the development of Y99H and A128T mutant strains, which exhibited cross‐resistance to the quinoline‐based ALLINI 19. By contrast, compound 27 remained effective at nanomolar concentrations against strains with variations at the polymorphic positions 124 and 125.

The X‐ray structure of HIV‐1 IN CCDF185H in complex with compound 27 (PDB: 7KE0) revealed that the inhibitor occupies the V‐shaped pocket formed by the two CCD subunits, establishing interactions comparable to those reported for other ALLINIs (Figure 18), while the pyrazole group projects outward from the core scaffold and does not directly contact the CCD dimer. Structural inspection further helped to explain the reduced activity of compound 27 against HIV‐1 variants carrying A128T and Y99H substitutions. The close proximity of residue A128 to the pyrrolopyridine moiety suggested that replacement with the bulkier, polar threonine may introduce steric constraints that hinder inhibitor binding. Likewise, Y99 lies deep within the V‐shaped pocket and plays a role in CCD–CCD contacts; mutations at this position are therefore likely to alter the pocket architecture and negatively affect ligand accommodation.

FIGURE 18: Crystal structure of compound 27 bound to HIV‐1 IN CCD dimer interface (PDB: 7KE0). CCD dimer chains are represented with cartoons and colored in light cyan and light green, respectively. Compound 27 is represented as white sticks, with nitrogen, oxygen, and chlorine atoms highlighted in blue, red, and green, respectively.

In addition, the compound proved to inhibit IN–RNA binding (IC50 = 0.020 µM) and viruses produced from cells treated with the inhibitor showed their viral RNA genomes outside of the capsid. Consistently with other ALLINIs, derivative 27 was found to induce higher‐order IN multimerization, disrupt IN–LEDGF binding, and primarily inhibit the late stages of viral replication. Importantly, the compound demonstrated favorable PK and safety profiles in preclinical models (rats and beagle dogs), including minimal interaction with drug‐metabolizing enzymes and good oral bioavailability, supporting its potential for oral once‐daily administration. The compound has demonstrated favorable safety and PK profiles in Phase 1 and nonclinical studies and is the first HIV‐1 ALLINI to enter a POC clinical trial. The interim analysis results of the first two completed cohorts (200 and 400 mg) highlighted promising results, while results of Cohort 3 (600 mg) are expected within 2026 [139, 143, 144].

To gain detailed structural insights into their mode of action, high‐resolution cocrystal structures of a CTD–CCD fusion construct with 27 and representative quinoline‐based ALLINI 17 were solved (PDBs: 8A1Q and 8A1P) (Figure 19) [145]. Both compounds localized at the CCD dimer interface, where their carboxylate groups formed bidentate hydrogen bonds with residues E170 and H171; in addition, the tert‐butoxy and bulky aromatic moieties were found to establish hydrophobic contacts with side chains from both CCD subunits (Q95, Y99, L102, T125, W132, T174, and M178). The main scaffolds of the compounds were found to protrude from the CCD dimer and engage in π–π stacking interactions with CTD residues Y226 and W235, promoting ionic interactions between K266, E170, and the inhibitor carboxylates. Additional interactions include a parallel π–π stacking between the methylpyrazole group of 27 with W235 and the hydrophobic contacts formed by the chromanyl group of 17 and the chlorophenyl substituent of 27 with the CTD residue I268 (Figure 19). By strengthening CTD–CCD contacts through the engagement of key conserved residues of HIV‐1 IN (E170, Y226, K266, W235, and I268), these compounds behave as authentic molecular glues [145].

FIGURE 19: Crystal structure of compounds 17 and 27 bound to HIV‐1 IN CTD–CCD interface (PDBs: 8A1Q and 8A1P). (A) Cartoon and superimposed representation of compounds 17 and 27. (B) Zoomed‐in, alternative perspective focusing exclusively on compound 17 to clearly illustrate its binding interactions. CCD dimer chains are colored in light cyan and light green, while the CTD is shown in gold. Compound 27 is depicted in red and transparency, while 17 is depicted as white sticks, with nitrogen, oxygen, and chlorine atoms highlighted in blue, red, and green, respectively.

Unlike classical molecular degraders, which stabilize interactions between E3 ubiquitin ligases and their target proteins to exploit the ubiquitin–proteasome system for target elimination, ALLINIs behave as nondegrading molecular glues [146]. As such, these inhibitors enhance the glue‐like stabilization of an otherwise transient or unfavorable interaction interface, thereby inducing the rapid, higher‐order multimerization and hyper‐aggregation of IN, which ultimately blocks IN functionality and affects viral maturation.

The growing interest of drug discovery campaigns in the search for nondegrading molecular glues, combined with the key structural insights recently described regarding the ALLINI‐induced glue‐like stabilization, will serve as key drivers for the rational design of next‐generation inhibitors.

Further analyses of resistance‐associated mutations selected under compound pressure, particularly Y99H and A128T, revealed reduced viral infectivity and a marked decrease in susceptibility, especially in double mutants [140]. Although these substitutions had limited impact on direct inhibitor binding to the CCD, they significantly disrupted CTD recruitment to the CCD–inhibitor complex. Structural comparisons indicated that A128T introduces steric clashes with CTD residues (notably Y226 and I268) and alters local positioning through additional interactions, ultimately impairing CCD–CTD assembly. Moreover, the orientation of the 3‐methyl group toward residues A128/T128 may further restrict access of the inhibitor to the V‐shaped pocket, contributing to reduced efficacy. To enhance deeper positioning within the binding pocket, a desmethyl analog of compound 27, referred to as EKC110 [140] (28, Figure 17), was synthesized. This derivative showed approximately 14‐fold higher potency against the HIV‐1(Y99H/A128T IN) mutant and effectively induced aberrant IN multimerization in both WT and mutant enzymes. Structural analyses of 28 in complex with WT and mutant CCD (PDB: 8S9Q), as well as CTD–CCD constructs (PDB: 8T5B), revealed that the absence of the 3‐methyl group allows 28 to shift its chlorobenzene moiety deeper into the CCD–CCD cavity toward L102 (Figure 20). This repositioning facilitates a unique anchoring of the CTD, where the inhibitor's carboxylate establishes a direct ionic interaction with K266, effectively bypassing the unfavorable T128‐T124 hydrogen bond network induced by the A128T mutation. Compared with 27, compound 28 promotes a distinct and more stable CTD orientation that is resilient to resistance‐associated structural shifts. Overall, these findings clarify the structural basis for the improved inhibitory potential of 28, which could serve as a robust alternative to 27 due to its superior genetic barrier to resistance. Furthermore, these insights provide deeper guidance for the rational design of new, nondegrading molecular glues endowed with enhanced efficacy and resistance barriers.

FIGURE 20: Crystal structure of compounds 27 and 28 bound to HIV‐1 IN CTD–CCD interface (PDBs: 8A1Q and 8T5B). (A) Cartoon and superimposed representation of compounds 27 and 28. (B) Zoomed‐in, alternative perspective focusing exclusively on compound 28 to clearly illustrate its binding interactions. CCD dimer chains are colored in light cyan and light green, while the CTD is shown in gold. Compound 27 is depicted in red and with transparency, while 28 is depicted as white sticks, with nitrogen, oxygen, and chlorine atoms highlighted in blue, red, and green, respectively.

The Discovery and Development of BDM‐2

To further expand the chemical space for this class of inhibitors, a new series of ALLINIs with highly promising properties has been recently reported by Biodim (Romainville, France). While sharing some similarities with previously described inhibitors, these derivatives differ in that they lack the typical central heterocyclic core. Instead, they feature a phenylacetic scaffold bearing either a 2‐_tert‐_butoxy (e.g., BDM‐2, 29 [141], and its methyl analog MUT871, 30 [141], Figure 17) or a 2‐cyclopropyloxy group [141]. Additionally, they incorporate a cyclopropyl group on the central phenyl core while retaining the 4‐chromanyl moiety (or related substituents) found in previously described ALLINIs, such as compound 18. Comparative studies including previously reported ALLINIs and such phenylacetic derivatives showed distinct activity profiles. While compound 30 was the most effective at disrupting the IN–LEDGF/p75 interaction, derivative 29 demonstrated the highest potency in promoting IN aggregation (Table 3). Overall, the compounds were more potent in inducing IN multimerization than inhibiting IN–LEDGF binding, which is consistent with the concept that ALLINIs primarily act as IN–IN molecular glues rather than simple disruptors of the IN–LEDGF interaction. Both derivatives 29 and 30 also demonstrated strong antiviral efficacy against clinical isolates, outperforming RAL in this panel. Consistent with previous findings, the inhibitors were more active in multiple‐round assays, with compound 30 being the most active (EC50 = 3.1 nM). While the reference ALLINI 19 exhibited a 27‐fold difference in EC50 values between single‐ and multiple‐round assays, the new derivatives showed significantly higher ratios, ranging from 100 to 200. Interestingly, a direct correlation between activity in single‐round infection assays and the ability to disrupt IN–LEDGF binding was observed. Accordingly, the most potent inhibitor of IN–LEDGF/p75 interaction (30) exhibited the strongest antiviral activity in the single‐round infection assay (EC50 = 0.63 µM); conversely, the weakest inhibitor of the series demonstrated the lowest antiviral effect in the single‐round infection assay (EC50 = 9.2 µM), resulting in a remarkable EC50 ratio of 767 between the two assay types.

Notably, derivatives 29 and 30 retained their efficacy against primary isolates carrying polymorphisms at IN positions 124 and 125. Resistance mutations selected by compound 29 included T174I, Y99H, A128T, H171Q, and N222K, with T174I exerting the strongest negative impact on antiviral activity and viral fitness. Conversely, analogs of derivative 29 endowed with a α‐cyclopropyloxy group on the acetic moiety exhibited a lower EC50 fold change relative to WT HIV‐1, indicating that they possess a higher genetic barrier toward such mutation compared to 29. Structural analysis of compound 29 in complex with previously described CTD–CCD construct (PDB: 8CBR) further clarifies these observations. The inhibitor is anchored by hydrogen bonds with the main‐chain amides of E170 and H171, while the T174 side chain establishes both hydrogen‐bonding and van der Waals interactions with the tert‐butoxy moiety of the inhibitor, explaining the loss of activity upon T174I substitution due to significant steric hindrance. Furthermore, this structure reveals how 29 acts as an efficient molecular glue by recruiting the CTD through a salt bridge between the inhibitor's carboxylate and K266. The compact benzene scaffold also allows W235 to approach the CCD more closely than in other ALLINI complexes, resulting in a distinct pivoting of the entire CTD domain that likely contributes to the unique multimerization profile of this series. In contrast, 2‐cyclopropyloxy analogs were less affected by the T174I mutation, likely owing to their more compact structure (Figure 21).

FIGURE 21: Crystal structure of compound 29 bound to HIV‐1 IN CTD–CCD interface (PDB: 8CBR). Chains are represented with cartoons. CCD dimers are colored in light cyan and light green, respectively, while the CTD is shown in gold. Compound 29 is depicted as white sticks, with oxygen atoms highlighted in red.

Importantly, the INSTI EVG retained full activity against ALLINI‐resistant mutants, and no antagonism was observed between compound 29 and approved antiretrovirals, while moderate to strong synergy was detected. Owing to its favorable profile, compound 29 advanced to clinical development: phase 1 studies in healthy volunteers reported good tolerability and PK properties, with no serious adverse events. Thus, together with derivative 27, compound 29 currently represents one of the most advanced HIV‐1 ALLINIs under clinical investigation [141, 147].