A bimolecular modification strategy for developing long-lasting bone anabolic aptamer

The molecular weight of nucleic acid aptamers (20 kDa) is lower than the cutoff threshold of the renal filtration (30–50 kDa), resulting in a very short half-life, which dramatically limits their druggability. To address this, we utilized 3-(2,5-dioxo-2,5-dihydro-1H-pyrrol-1-yl)-N-(4-hydroxy-2-oxo-2H-chromen-6-yl)propenamide (HC) and 12-((2,5-dioxopyrrolidin-1-yl)oxy)-12-oxododecanoic acid (DA), two newly designed coupling agents, for synergistic binding to human serum albumin (HSA). Both HC and DA are conjugated to a bone anabolic aptamer (Apc001) against sclerostin to form an Apc001OC conjugate with high binding affinity to HSA. Notably, HC and DA could synergistically facilitate prolonging the half-life of the conjugated Apc001 and promoting its bone anabolic potential. Using the designed blocking peptides, the mechanism studies indicate that the synergistic effect of HC-DA on pharmacokinetics and bone anabolic potential of the conjugated Apc001 is achieved via their synergistic binding to HSA. Moreover, biweekly Apc001OC at 50 mg/kg shows comparable bone anabolic potential to the marketed sclerostin antibody given weekly at 25 mg/kg. This proposed bimolecular modification strategy could help address the druggability challenge for aptamers with a short half-life.


INTRODUCTION
Aptamers are short, single-stranded DNA (ssDNA) or RNA (ssRNA) oligonucleotides, typically composed of 25-80 nucleotides, which can form a range of complex three-dimensional structures.These unique structures enable them to selectively bind to specific targets, including peptides, proteins, small molecules, and even living cells.2][3][4][5][6][7][8] The thrombin aptamer NU172 failed in the phase 2 clinical trial due to the short half-life in vivo (10 min). 9The development of a pharmacokinetics-related aptamer design strategy could help address their druggability challenge.
There are some post-selection modification strategies that could prolong the half-life of aptamers. 10Accordingly, introducing a bulky moiety is an efficient strategy for averting the rapid renal clearance of aptamers and extending their time in circulation.Among the various substantial functional groups utilized to increase the dimensions of aptamers are polyethylene glycol (PEG), proteins, cholesterol, liposomes, as well as organic and inorganic nanomaterials.2][13][14][15] Riccardi et al. engineered NU172-T H 9 with an extended in vivo half-life of 28.6 h through PEGylation, resulting in enhanced thrombin inhibitory activity. 9Nonetheless, the PEG moiety constitutes a substantial portion of the PEG-aptamer conjugate (over 75%).7][18][19] In light of these druggability challenges, it is desirable to seek innovative coupling agents to develop long-lasting therapeutic aptamers with increasable dosages.
Human serum albumin (HSA), the predominant plasma protein (ranging between 35 and 50 g/L in human serum), has a molecular weight of 66.5 kDa. 20It is synthesized at an approximate rate of 0.7 mg/h per gram of liver, equating to 10-15 g daily, and exhibits a mean half-life of 19 days. 21Strategically, HSA has large hydrophobic interface cages, which could bind to some low-molecular-weight coupling agents (LMWCAs) to form molecular complexes with an average mass above the TVr. 22Clinically, several strategies have been adopted to develop long-lasting therapeutics by conjugating a small molecular albumin-binding moiety to either small molecular drugs or peptide drugs. 22Mechanistically, LMWCAs can be employed to modify aptamers, producing conjugates that, upon binding to HSA, form molecular complexes with an average mass exceeding the TVr for extending the half-life. 13Advantageously, the molecular weight of LMWCAs is substantially lower than that of the aptamer, allowing the aptamer to account for a large proportion of the LMWCA-aptamer conjugate (over 95%).Two reported molecules have strong binding affinity to HSA.One is warfarin and the other is fatty acids (FAs). 23,24However, the anticoagulation effect of warfarin makes it unsuitable to be an LMWCA. 25,26The benzylacetone moiety on warfarin is responsible for this anticoagulant action.Consequently, we designed a novel coumarin derivative (3-(2,5-dioxo-2,5-dihydro-1H-pyrrol-1-yl)-N-(4-hydroxy-2-oxo-2H-chromen-6-yl)propenamide [HC]) exhibiting a potent affinity for HSA without the anticoagulant effect.8][29][30][31] However, OA could also bind to FA-binding proteins (FABPs), which are widely distributed Figure 1.A synergetic interaction between HC and DA to HSA was found by an interaction analysis (A) The predicted interactions between the coumarin moiety of HC (4-hydroxycoumarin, colored in green) and HSA, including two hydrogen bonds with Y146 (2.0A ˚) and R253 (2.1 A ˚), respectively.(B) The predicted interactions between the FA moiety of DA (dodecanedioic acid [12C], colored in light blue) and HSA, including three hydrogen bonds with K16 (2.4A ˚), K132 (2.3A ˚), and K158 (1.8A ˚), respectively.(C) The predicted interactions between the coumarin moiety of HC and HSA, with the presence of the FA moiety of DA near the binding pocket of HC to HSA, including two hydrogen bonds with K191 (3.9A ˚) and K195 (2.3A ˚), respectively; two cation-p interactions with R218 (3.2A ˚) and H238 (4.3A ˚), respectively; and two hydrogen bonds with R253 (2.1 and 2.7 A ˚) alone.The above data predicted that the conformation of the binding pocket of HC to HSA might be significantly altered upon binding of DA, implying a dramatically enhanced binding affinity of HC to HSA.Note: R253 (colored in gray) was the only unaltered predicted binding site between HC and HSA, with or without the presence of 12C.À12C, before molecular dynamic simulation; +12C, after molecular dynamic simulation.
in various tissues, resulting in the decrease of drug amounts in blood circulation. 32The length of FA chains could play an important role in the binding to FABP.Thus, we designed a novel FA derivate, 12-((2,5-dioxopyrrolidin-1-yl)oxy)-12oxododecanoic acid (DA), with low binding affinity to FABP but high binding affinity to HSA.
][35] Remarkably, the long-lasting sclerostin aptamer for osteogenesis imperfecta (OI) was granted Rare Pediatric Disease Designation (RPD-2022-667) and Orphan Drug Designation (DRU-2022-9087) by the US Food and Drug Administration (FDA) in 2022.In this study, both HC and DA are conjugated to Apc001 to form an Apc001OC conjugate with high binding affinity to HSA.Notably, we found that HC and DA could synergistically facilitate prolonging the halflife of the conjugated Apc001 and promoting its bone anabolic potential.Furthermore, our mechanism studies indicated that the synergistic effect of HC-DA on pharmacokinetics and bone anabolic potential of the conjugated Apc001 is achieved via their synergistic binding to HSA (Figure 1).The proposed innovative long-lasting modification strategy could help in addressing the druggability challenge of aptamers with short half-life.

RESULTS
Two novel compounds were designed as candidate LMWCAs with high binding affinity to HSA The crystal structure of warfarin complexed with HSA has revealed that the interaction between warfarin and HSA appears to be dominated by both hydrophobic contacts and several specific electrostatic interactions. 24However, the intrinsic bioactive properties of warfarin pose hemorrhagic concerns in susceptible individuals, rendering warfarin unsuitable for application as a long-acting coupling agent. 25The benzylacetone group on warfarin was implicated in decreasing blood clotting ability through the inhibition of vitamin K epoxide reductase, an essential enzyme in the synthesis of vitamin K-dependent coagulation factors. 26Consequently, we designed a novel coumarin derivative without anticoagulation effect.A computer-assisted high-throughput virtual screening was employed to identify 4-hydroxycoumarin (integral for HSA binding), exhibiting the highest predicted binding affinity to HSA among over 1,000 coumarin derivatives lacking the benzylacetone group (Figures S1  and S2A). 36Given that 4-hydroxycoumarin is unsuitable for direct aptamer conjugation, an aptamer-specific linker was introduced to the coumarin structure to form the candidate LMWCA, HC.
OA holds the strongest binding affinity to HSA among the published FAs. 22,27,28Besides, it has been reported that the conformation of warfarin-binding pocket in HSA is significantly altered upon FA binding, which dramatically enhances the binding affinity of warfarin to HSA. 29 However, OA could also bind to FABPs, which are widely distributed in various tissues, resulting in the decrease of drug amounts in blood circulation. 32The length of the fatty chains plays an important role in the binding of FAs to FABP. 32Therefore, we designed a novel FA derivative with low binding affinity to FABP but high binding affinity to HSA.Virtual molecular docking identified the dodecanedioic acid (the key binding structure to HSA) with the lowest predicted binding affinity to FABP but comparably high predicted binding affinity to HSA from a series of FA derivatives varying in length and side-chain modifications (Figure S2B).To transform dodecanedioic acid into a suitable coupling agent for aptamer conjugation, an aptamer-specific linker was introduced to the FA structure, generating the agent DA.

Experimental validation of HC and DA as candidate LMWCAs with high binding affinity to HSA
To validate the binding affinity of HC to HSA, surface plasmon resonance (SPR) assay was performed.Our data demonstrated that HC had high binding affinity to HSA (dissociation constant [K d ] = 6.54 nM) (Figure S2A).To validate the abolished anticoagulant effect of HC, the anticoagulant assay was performed.The anticoagulant assay showed that warfarin markedly reduced the clotting time, whereas HC had no anticoagulant effect (Figure S2A).As shown by the above data from both SPR assay and anticoagulant assay, HC could demonstrate high binding affinity to HSA but without anticoagulant effect.
To validate the binding affinity of DA to HSA and FABP, respectively, SPR assay was performed.Our data demonstrated that DA had low binding affinity to FABP (K d = 90.64nM) but still retained high binding affinity to HSA (K d = 40.52nM) (Figures S2C and S2D).Moreover, the binding affinity of DA to FABP was significantly lower than that of OA to FABP (K d = 8.62 nM).As shown by the above data from the SPR assay, DA could demonstrate low binding affinity to FABP but high binding affinity to HSA.

Synthesis of Apc001OC conjugate
1 (amino-mercapto-modified Apc001 [Apc001-SH-NH 2 ]) was conjugated with both HC and DA to form Apc001OC.The detailed synthesis route of Apc001OC is described below.Synthesis of Apc001OC (Figure 2C) commenced with 1 and HC in the mixture of doubledistilled H 2 O (ddH 2 O) and DMF (pH = 6.75) to provide 2. Without further purification, 2 was reacted with DA in the presence of NaHCO 3 , providing Apc001OC.Apc001-SH was conjugated with HC to form Apc001HC (Figure S3A).Amino-modified Apc001 (Apc001-NH 2 ) was conjugated with DA to form Apc001DA (Figure S3B).The structures of key intermediates were confirmed by high-performance liquid chromatography (HPLC) and mass spectrometry (Figures S13-S15).Apc001OC was characterized by HPLC and mass spectrometry (Figure S16).The structures of Apc001HC and Apc001DA were confirmed by HPLC and mass spectrometry (Figures S17 and S18).
Both HC and DA could synergistically facilitate the binding affinity of the conjugated Apc001 to HSA The binding affinity of Apc001OC, Apc001HC, and Apc001DA to HSA was analyzed by SPR.Our data demonstrated that the Apc001OC conjugate (K d = 0.75 nM) had dramatically stronger binding affinity to HSA than Apc001HC (K d = 7.06 nM) and Apc001DA (K d = 46.6 nM).A synergetic interaction mechanism between HC and DA was predicted using a binding-mode-based interaction analysis by molecular dynamic simulation (Figures 1 and 3A-3E).Without the presence of DA near the binding pocket of HC to HSA, there were two predicted interactions between HC and HSA, including two hydrogen bonds with Y146 (2.0 Å) and R253 (2.1 Å), respectively (Figure 1A).When DA bound HSA alone, there were three predicted interactions between DA and HSA, including three hydrogen bonds with K16 (2.4 Å), K132 (2.3 Å), and K158 (1.8 Å), respectively (Figure 1B).Notably, with the presence of DA near the binding pocket of HC to HSA, there were six predicted interactions between HC and HSA, including two hydrogen bonds with K191 (3.9 Å) and K195 (2.3 Å), respectively; two cation-p interactions with R218 (3.2 Å) and H238 (4.3 Å), respectively; and two hydrogen bonds with R253 (2.1 Å, 2.7 Å) alone (Figure 1C).To efficiently demonstrate the increased binding affinities of different Apc001 conjugates to HSA, we normalized the binding affinity of Apc001DA to HSA (K d = 46.6 nM) as 100%.As a result, the binding affinity of Apc001HC to HSA (K d = 7.06 nM) showed an increase of 660% compared to the binding affinity of Apc001DA to HSA.More importantly, the binding affinity of Apc001OC to HSA (K d = 0.75 nM) had an increase of 6,200% compared to the binding affinity of Apc001DA to HSA (Table S1).This means that the increase in the binding affinity of Apc001OC to HSA was dramatically higher than the sum of the increases in the binding affinity of Apc001HC to HSA and Apc001DA to HSA (6,200% > 760%, where 760% = 660% + 100%).Therefore, we concluded that HC and DA could synergistically facilitate the binding affinity of the conjugated Apc001 to HSA.
To elucidate the binding affinity of the conjugated Apc001 to sclerostin, SPR assay was performed.Our data demonstrated that the Apc001OC conjugate (K d = 35.6 nM) had a comparable binding affinity to sclerostin when compared to Apc001 (K d = 37.1 nM) (Figure 3F), suggesting that the introduction of both HC and DA did not perturb the inherent affinity of Apc001 to sclerostin.
Four blocking peptides were designed to block the binding of HC, DA, and HC-DA combination and HC-DA interaction to HSA, respectively The coumarin structure could contribute to the high binding affinity to HSA. 24 Based on the calculated binding sites of HSA to the coumarin moiety of HC, it was predicted that Y146 and R253 on HSA were involved.Consequently, a peptide (peptide A) including these residues was designed and its binding affinity to HC was validated via SPR (K d = 15.56 nM) (Figure S4A).The FA structure could contribute to the high binding affinity to HSA. 22,27,28 Based on the calculated binding sites of HSA to the FA moiety of DA, it was predicted that K16, K132, and K158 on HSA were involved. 37A corresponding peptide (peptide B) including the above binding sites was designed and its affinity to DA was validated by SPR (K d = 40.15nM) (Figure S4B).Based on the above prediction, a peptide sequence including the above binding sites that could bind both HC and DA was designed (peptide C).Then we verified its binding affinity to both HC and DA by SPR analysis (K d = 13.78 nM for HC, K d = 30.62nM for DA) (Figure S4C).Furthermore, based on the calculated binding sites of HSA to HC in the presence and absence of DA (Figure S5A), it was predicted that residues Y146 and R253 on HSA were involved in the absence of DA, while, in its presence, residues K191, K195, R253, R238, and H238 on HSA were involved in the presence of DA.Of note, K191, K195, R238, and H238 were the newly participated binding sites in the presence of DA (Figure S5B).
Then, a peptide sequence (peptide D) including the above newly participated binding sites was designed and its affinity to HC was validated by SPR (K d = 26.82nM) (Figure S5C).We designed the negative peptide sequences (NC1 for peptide A, NC2 for peptide B, NC3 for peptide C, NC4 for peptide D) by mutating the predicted binding amino acids sites to alanine on each blocking peptide.
The binding affinity of Apc001OC conjugate to HSA was significantly reduced or abolished when it was pre-saturated with our designed blocking peptide sequences The above data provided the following validated research tools: peptide A (designed to block the binding of HC to HSA), peptide B (designed to block the binding of DA to HSA), peptide C (designed to block the binding of HC-DA combination to HSA), and peptide D (designed to block the binding of HC-DA interaction to HSA).Mechanistically, the binding affinity of the Apc001OC conjugate to HSA was significantly reduced when it was pre-saturated with either peptide A (from K d = 0.75 nM to K d = 27.8 nM) or peptide B (from K d = 0.75 nM to K d = 23.5 nM) (Figures S6A and S6B).The binding affinity was abolished when it was pre-saturated with peptide C (Figure S6C).Together, our data indicated the important roles of HC, DA, and HC-DA combination in Apc001OC conjugate binding to HSA, respectively.Moreover, both HC and DA could synergistically facilitate the binding affinity of the conjugated Apc001 to HSA.Importantly, the synergistically enhanced binding affinity of the conjugated Apc001 to HSA by HC and DA was abolished when pre-saturated with peptide D (from K d = 0.75 nM to K d = 12.61 nM) (Figure S6D).Our findings indicated that DA could alter the conformation of the HC binding pocket in HSA and drive more amino acid residues to bind to HC, including K191, K195, R238, and H238, which could explain the contribution of the interaction between HC and DA to the synergistic effect.

HC and DA could synergistically prolong the half-life of the conjugated Apc001 toward 289.94 h in normal rats
From our pharmacokinetics studies in normal rats, a single subcutaneous administration of Apc001OC conjugate (6.25 mg/kg), Apc001HC conjugate (6.25 mg/kg), Apc001DA conjugate (6.25 mg/kg), PEG40modified Apc001 (Apc001PE) (6.25 mg/kg), and Apc001 (6.25 mg/ kg) was performed. 12,33The plasma concentrations of the aptamers at each time point after the subcutaneous administration were analyzed by molecular beacon method (Figure 4A).The Apc001OC conjugate (t 1/2 = 289.94h) had a longer half-life compared with Apc001HC conjugate (t 1/2 = 162.93h), Apc001DA conjugate (t 1/2 = 98.81 h), and Apc001PE conjugate (t 1/2 = 64.83h) (Figure 4B).The half-life of the non-conjugated Apc001 was only 1.92 h (Figure 4B).Moreover, the Apc001OC conjugate (AUC 0-t = 13,397 (mg*h)/L) had a lower clearance rate in comparison to Apc001HC conjugate (AUC 0-t = 7,804.508(mg*h)/L), Apc001DA conjugate (AUC 0-t = 6,732.577(mg*h)/L), Apc001PE conjugate (AUC 0-t = 936.090(mg*h)/L), and Apc001 conjugate (AUC 0-t = 60.084(mg*h)/L) (Figure 4C).The t 1/2 value of 289.94 h could afford an interval of 2 weeks at most for the administration of the Apc001OC conjugate, whereas the t 1/2 value of 56 h could only afford an interval of 3 days at most for romosozumab. 38,39Similarly, the t 1/2 value of 64.83 h could only afford an interval of 3 days at most for Apc001PE.It suggested that the half-life of the Apc001OC conjugate in normal rats was at least four times longer than that of both romosozumab and Apc001PE.More importantly, the half-life of the Apc001OC conjugate (t 1/2 = 289.94h) was even longer than the addition of the half-life of the Apc001HC conjugate (t 1/2 = 162.93h) and the half-life of the Apc001DA conjugate (t 1/2 = 98.81 h).Our findings indicated that both HC and DA could synergistically facilitate prolonging the circulation half-life of the conjugated Apc001 in normal rats.
The HC-DA combination effect on prolonging the half-life of the conjugated Apc001 was reduced or abolished when it was presaturated with our designed blocking peptide sequences in normal rats From the above binding validation data, peptide A, peptide B, and peptide C could block the binding of HC, DA, and HC-DA combination to HSA, respectively.Then, the normal rats were subcutaneously administered with Apc001OC, which is pre-saturated by different blocking peptides (denoted as peptide A/B/C@Apc001OC) or these blocking peptides alone, respectively, to investigate the role of HC-DA combination in prolonging the half-life.The half-life of the different blocking peptides was less than 1 h (Figure S21).Our data demonstrated that the HC-DA combination effect on prolonging the half-life was reduced when it was pre-saturated with either peptide A (from t 1/2 = 289.94h to t 1/2 = 96.3h) or peptide B (from t 1/2 = 289.94h to t 1/2 = 170.1 h) (Figure S7).Importantly, the HC-DA combination effect on prolonging the half-life was abolished when it was pre-saturated with peptide C (from t 1/2 = 289.94h to t 1/2 = 6.4 h) 12,33 (Figure S7), indicating the indispensable contributions of both HC and DA in Apc001OC conjugate to synergistically facilitate prolonging the circulation half-life of the conjugated Apc001.
The HC-DA interaction effect on prolonging the half-life of the conjugated Apc001 was abolished when it was pre-saturated with peptide D in normal rats From the above binding validation data, the peptide D could block the binding of HC-DA interaction to HSA.Then, the normal rats were subcutaneously administered with Apc001OC, which is pre-saturated by peptide D (denoted as peptide D@Apc001OC) to investigate the role of HC-DA interaction in prolonging the half-life.Our findings demonstrated that the HC-DA interaction effect on prolonging the half-life was abolished when it was pre-saturated with peptide D (from t 1/2 = 289.94h to t 1/2 = 196.8h) (Figure S7), indicating the important role of the HC-DA interaction in the Apc001OC conjugate in synergistically prolonging the circulation half-life of the conjugated Apc001.
HC and DA could synergistically facilitate the conjugated Apc001, promoting bone formation, increasing bone mass, and improving bone microarchitecture integrity in Col1a2 +/G610C mice To evaluate the effect of Apc001OC on bone mass and bone microarchitecture in OI mice, 6-week-old Col1a2 +/G610C mice (OI) were  subcutaneously administered with 25 mg/kg of Apc001OC conjugate, Apc001HC conjugate, Apc001DA conjugate, scrambledApc001OC (scApc001OC) conjugate, Apc001PE conjugate, or non-conjugated Apc001 once every 2 weeks (q2wk) for a 12-week period. 12,33Micro-computed tomography (micro-CT) was utilized for the measurement of trabecular bone (below the growth plate) at the metaphysis of the distal femur in Col1a2 +/G610C mice.Before treatment, trabecular volume per total volume (Tb.BV/TV), trabecular volumetric mineral density (Tb.vBMD), and trabecular number (Tb.N) at the above site were significantly lower in the OI-baseline group in comparison with the wild-type (WT) baseline (group (Figure 6).It indicated substantially lower bone mass and worse bone microarchitecture for the trabecular bone in distal femur of Col1a2 +/G610C mice compared to WT mice.The micro-CT analysis showed that the Apc001OC group had significantly higher Tb.vBMD (+73.8%,p < 0.01), Tb.BV/TV (+128%, p < 0.01), and Tb.N (+70.6%, p < 0.01) compared to the OI-Veh group.However, our data demonstrated no differences in the levels of Tb.vBMD, Tb.BV/TV, and Tb.N among the OI-Veh, Apc001, and scApc001OC groups (Figure 5A).The levels of Tb.vBMD, Tb.BV/TV, and Tb.N in Apc001OC group were significantly higher than those in the Apc001PE group.More importantly, the levels of the above parameters in the Apc001OC group were even higher than the addition of those in Apc001HC group and Apc001DA group.
To examine the effect of Apc001OC on bone formation in OI mice, bone histomorphometric analysis was used for measurement of trabecular bone (below the growth plate) at the distal femur.Before treatment, trabecular bone formation rate/bone surface (Tb.BFR/BS) and trabecular mineral apposition rate (Tb.MAR) at the distal femur were significantly lower in the OI-baseline group compared to WT-baseline group, indicating substantially lower bone formation for trabecular bone of Col1a2 +/G610C mice when compared to WT mice.The bone histomorphometric analysis showed that the Apc001OC group had significantly higher BFR/BS (+208.8%,p < 0.0001) and MAR (+140.5%,p < 0.0001) compared to the OI-Veh group (Figure 5B).The Apc001HC group also had higher BFR/BS (+104.4%,p < 0.0001) and MAR (+87.9%,p < 0.005) compared to the OI-Veh group (Figure 5B).However, our data demonstrated no differences in the levels of BFR/BS and MAR among the OI-Veh, Apc001, and scApc001OC groups (Figure 5B).The levels of BFR/BS and MAR in the Apc001OC group were significantly higher than those in Apc001PE group.More importantly, the levels of the above parameters in the Apc001OC group were even higher than the addition of those in Apc001HC group and Apc001DA group.
The HC-DA combination effect on facilitating the conjugated Apc001 promoting bone formation, increasing bone mass, and improving bone microarchitecture integrity was reduced or abolished when it was pre-saturated with our designed blocking peptide sequences in Col1a2 +/G610C mice From the above binding validation data, peptide A, peptide B, and peptide C could block the binding of HC, DA, and HC-DA combination to HSA, respectively.To investigate the role of HC-DA combination effect in facilitating the conjugated Apc001 promoting bone formation, increasing bone mass, and improving bone microarchitecture integrity, Col1a2 +/G610C mice were subcutaneously administered with Apc001OC, peptide A/B/C@Apc001OC, or blocking peptides alone, respectively.The trabecular bone in distal femur was examined by micro-CT and bone histomorphometric analysis, respectively.The micro-CT analysis showed that the Apc001OC group had significantly higher Tb.vBMD (+73.8%,p < 0.01), Tb.BV/TV (+128%, p < 0.01), and Tb.N (+70.6%, p < 0.01) compared to the OI-Veh group (Figure 5A).However, our data demonstrated no differences in the above parameters between the OI-Veh group and the other groups (Figures 5A and  S10A) (p > 0.05).Our above data indicated that the HC-DA combination effect on facilitating the conjugated Apc001 increasing bone mass and improving bone microarchitecture integrity was reduced when it was pre-saturated with either peptide A or peptide B. Importantly, the HC-DA combination effect on facilitating the conjugated Apc001 increasing bone mass and improving bone microarchitecture integrity was abolished when it was pre-saturated with peptide C, indicating the important role of HC-DA combination effect on facilitating the conjugated Apc001 promoting increasing bone mass and improving bone microarchitecture integrity in OI mice.The bone histomorphometric analysis showed that the Apc001OC group had significantly higher BFR/BS (+208.8%,p < 0.0001) and MAR (+140.5%,p < 0.0001) compared to the OI-Veh group (Figure 5B).The peptide B@Apc001OC group also had higher BFR/BS (+96.7%, p < 0.0001) and MAR (+81.6%, p < 0.0001) compared to the OI-Veh group (Figure 5B).However, our data demonstrated no differences in the above parameters between the OI-Veh group and the other groups (Figures 5B and  S10B) (p > 0.05).Our data demonstrated that the HC-DA combination effect on facilitating the conjugated Apc001 promoting bone formation was reduced when it was pre-saturated with either peptide A or peptide B (from +208.8% to +96.7% for BFR/BS, from +140.5% to +81.6% for MAR).Importantly, the HC-DA combination effect on facilitating the conjugated Apc001 promoting bone formation was abolished when it was pre-saturated with peptide C, indicating the important role of HC-DA combination effect bind HC; peptide B, a peptide sequence within HSA including the predicted binding sites that could bind DA; peptide C, a peptide sequence within HSA including the predicted binding sites that could bind both HC and DA; peptide D, a peptide sequence within HSA including the predicted newly participated binding sites that could bind HC with the presence of DA.Data are expressed as mean ± standard deviation followed by one-way ANOVA with Tukey's post hoc test, n = 6 per group.*p < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.0001.on facilitating the conjugated Apc001 promoting bone formation in OI mice.
The HC-DA interaction effect on facilitating the conjugated Apc001 promoting bone formation, increasing bone mass, and improving bone microarchitecture integrity was abolished when it was pre-saturated with peptide D in Col1a2 +/G610C mice From the above binding validation data, the peptide D could block the binding of HC-DA interaction to HSA.To investigate the role of HC-DA interaction effect in facilitating the conjugated Apc001 promoting bone formation, increasing bone mass, and improving bone microarchitecture integrity, Col1a2 +/G610C mice were subcutaneously administered with Apc001OC, peptide D@Apc001OC, or peptide D, respectively.The trabecular bone in distal femur was examined by micro-CT and bone histomorphometric analysis, respectively.The micro-CT analysis showed that the Apc001OC group had significantly higher Tb.vBMD (+73.8%,p < 0.01), Tb.BV/TV (+128%, p < 0.01), and Tb.N (+70.6%, p < 0.01) compared to the OI-Veh group.The peptide D@ Apc001OC group also had notably higher Tb.vBMD (+54.2%, p < 0.01), Tb.BV/TV (+88.2%, p < 0.01), and Tb.N (+53%, p < 0.01) compared to the OI-Veh group (Figure 5A).Our data demonstrated that the HC-DA interaction effect on facilitating the conjugated Apc001 increasing bone mass and improving bone microarchitecture integrity was abolished when it was pre-saturated with peptide D (from +73.8% to +54.2% for Tb.vBMD, from +128% to +99.5% for Tb.BV/TV, from +70.6% to +47.3% for Tb.N), indicating the important role of HC-DA interaction effect on facilitating the conjugated Apc001 increasing bone mass and improving bone microarchitecture integrity in OI mice.The bone histomorphometric analysis showed that the Apc001OC group had significantly higher BFR/BS (+208.8%,p < 0.0001) and MAR (+140.5%,p < 0.0001) compared to the OI-Veh group.The peptide D@Apc001OC group also had notably higher BFR/BS (+124.2%,p < 0.0001) and MAR (+104%, p < 0.0001) compared to the OI-Veh group (Figure 5B).Our data demonstrated that the HC-DA interaction effect on facilitating the conjugated Apc001 promoting bone formation was abolished when it was pre-saturated with peptide D (from +208.8% to +124.2% for BFR/BS, from +140.5% to +104% for MAR), indicating the important role of HC-DA interaction effect on facilitating the conjugated Apc001 promoting bone formation in OI mice.

DISCUSSION
Design and validation of HC and DA as two candidate LMWCAs with high binding affinity to HSA The SPR result demonstrated that HC exhibited significant high binding affinity to HSA, indicating that the SPR data validated the prediction data.The coumarin structure on HC could contribute to its high binding affinity to HSA.Importantly, the anticoagulant study demonstrated that HC had no anticoagulant effects compared to warfarin.Prior studies have implicated that the benzylacetone moiety is essential for the anticoagulant effect of warfarin. 26Thus, the abolished anticoagulant effect of HC could be attributed to the absence of this benzylacetone group.Our data supported that HC could be a suitable candidate LMWCA rather than warfarin.
The SPR result demonstrated that DA possessed a low binding affinity to FABP but high binding affinity to HSA, indicating that the SPR data validated the prediction data.The length of FA chains is pivotal in the binding of FAs to FABP. 32Consistently, the above data regarding the low binding affinity of DA to FABP but high binding affinity of DA to HSA could be explained by the fatty chain of DA being shorter than OA.Consequently, our data supported that DA could be a suitable candidate LMWCA rather than OA.
A synergetic interaction was predicted between HC and DA on binding to HSA by the molecule-protein computational analysis and experimentally validated by the SPR assay According to the molecule-protein computational analysis data, there were more predicted binding sites (K191, K195, R253, R238, and H238) of HSA to HC with the presence of DA near the binding pocket of HC to HSA than those (Y146 and R253) of HSA to HC without the presence of DA.This observation is consistent with SPR findings, where the Apc001OC conjugate exhibited markedly enhanced binding affinity to HSA than either Apc001HC or Apc001DA.The synergetic interaction between HC and DA could be explained by the FA structure on DA potentially modulating the conformation of the HC binding pocket on HSA, facilitating the engagement of more amino acid residues, in which the binding affinity of HC to HSA was dramatically enhanced. 29The above data indicated that both HC and DA could synergistically facilitate the binding affinity of the conjugated Apc001 to HSA.
HC and DA could synergistically facilitate prolonging the half-life of the conjugated Apc001 and promoting its bone anabolic potential in OI mice via their synergistic binding to HSA Apc001OC demonstrated longer half-life and enhanced bone anabolic potential than either Apc001HC or Apc001DA at the same dosage and dosing interval.Mechanistically, the prolonged half-life and enhanced bone anabolic potential of Apc001OC was reduced or abolished when it was pre-saturated with our designed blocking peptides.Logically, the longer half-life of Apc001OC could be explained by HC and DA synergistically binding to HSA and consequently exhibiting dramatically stronger binding affinity to HSA than either Apc001HC or Apc001DA.Moreover, the enhanced bone anabolic potential of Apc001OC could be further explained by the half-life of Apc001OC (t 1/2 = 289.94h) being longer than either Apc001HC (t 1/2 = 162.93h) or Apc001DA (t 1/2 = 98.81 h).Together, the above data indicated that HC and DA could synergistically facilitate prolonging the half-life of the conjugated Apc001 and promoting its bone anabolic potential in OI mice via their synergistic binding to HSA.
Apc001OC with 50 mg/kg q2wk showed comparable bone anabolic potential in OI mice when compared to the marketed sclerostin antibody with 25 mg/kg opw It has been reported that the marketed sclerostin antibody (romosozumab) was proved to improve skeletal parameters in OI mice. 40In this study, Apc001OC with 50 mg/kg q2wk, showed comparable effect on promoting bone formation, increasing trabecular bone mass, and improving trabecular microarchitecture at distal femur in OI mice when compared to romosozumab with 25 mg/kg opw.Although Apc001OC was administered less frequently than romosozumab, the therapeutic effect of both drugs was comparable.This could be explained by that the half-life of Apc001OC was four times longer than romosozumab, which allows Apc001OC to achieve comparable therapeutic effect with a lower dosing frequency than romosozumab when the total dosage of both drugs is equal during the treatment.For the unmodified aptamers with short half-life, frequent administrations at a short interval could largely reduce the clinical treatment compliance of patient.Notably, this proposed bimolecular modification strategy could afford an interval of 2 weeks for the administration of the conjugated aptamer, which could potentially improve the clinical treatment compliance, especially for the treatment of chronic diseases. 19

Perspective
In general, this proposed bimolecular modification strategy could greatly extend the half-life of the conjugated sclerostin aptamer (up to 12 days) compared to that of PEGylation (only 2.7 days), which could potentially improve the clinical treatment compliance. 12,33oncurrently, the bimolecular modification strategy could remarkably increase the proportion of aptamer within the aptamer conjugate (up to 95%) compared to that of PEGylation (low to 25%), which increases the subcutaneous dosage for the aptamer moiety at a fixed subcutaneous administration volume.Accordingly, the synergistic effect between HC and DA may pave the way for development of all long-lasting therapeutic aptamers targeting circulating proteins with increasable dosages.Although existing studies corroborate the superior efficacy of multiple FAs in enhancing the binding of warfarin to HSA compared to singular FA, the optimal amount of DA required to enhance the binding of HC to HSA remains ambiguous.Thus, the tailored combination of HC with varying DA amounts needs to be further explored for pharmacokinetics-related aptamer optimization.On the other hand, it is worth screening different combinations of HC derivatives and DA derivatives to enhance their binding affinity to HSA for pharmacokinetics-related aptamer optimization.However, combinations of HC derivatives (>10 5 ) and DA derivatives (>10 5 ) could be at least 10 10 , all of which cannot be experimentally implemented due to the high cost in both labor and time.Artificial intelligence could be employed to generate both HC derivatives and DA derivatives with predicted binding affinities to HSA for further predicting the most promising HC derivative/DA derivative combination. 41Moreover, integrating computational chemistry with structural biology is imperative for a comprehensive understanding of the Apc001OC@HSA complex.

Conclusions
In summary, two innovative coupling agents are designed, HC and DA, which could synergistically bind to HSA with high affinities.The in vivo study indicated that HC and DA could synergistically facilitate prolonging the half-life of the conjugated Apc001 and promoting its pro-anabolic potential in OI mice via their synergistic binding to HSA.Importantly, Apc001OC showed dramatically longer half-life and enhanced pro-anabolic potential compared to Apc001PE.Furthermore, Apc001OC with 50 mg/kg q2wk shows comparable bone anabolic potential to the marketed sclerostin antibody with 25 mg/kg opw.The proposed innovative long-lasting modification strategy could help address the druggability challenge of aptamers with short half-life.Future studies would aim to use artificial intelligence to generate both HC derivatives and DA derivatives with predicted binding affinities to HSA for further predicting the most promising HC derivative/DA derivative combination.

Molecular-protein docking by Schrodinger
Schrodinger_Suites_2021-2 is a suite of software for simulating small molecule and macromolecule systems.It was used to perform docking simulations of the coumarin derivatives to full-length HSA (PDB: 1AO6).Some docking modes were generated.HSA were prepared using Protein Preparation Wizard with default parameters.All small molecules were prepared using LigPrep.Additionally, the ionization of LigPrep was set to generate possible states at target pH 7.0 ± 0.5 using Epik, and the checkbox to include the original state was checked.Before running Glide Grid Generation, the grid box was set to cover the whole macromolecule.Before running Glide Ligand Docking, the precision of Glide ligand docking was set to XP (extra precision) and the checkbox of Add Epik state penalties to docking score was unchecked.The output parameters of Glide ligand docking were selected as follows: the maximum number of poses per ligand written out was selected as 25 and the number of poses per ligand to include was selected as 500.All coumarin derivatives in this study were selected from the PubChem database.The output format of the coumarin derivatives was based on REST-style version of Power User Gateway (PUG) (https://pubchemdocs.ncbi.nlm.nih.gov/pugrest), and all compounds were further refined offline.Due to the size of the binding pocket of the receptor and Lipinski rule of five, molecules with molecular weight higher than 550 were removed.The final screening library contained 14,991 coumarin derivatives without a benzylacetone group.AutoDock Vina was used for the virtual screening.The top eight conformations with comparably low binding energy to HSA were chosen for subsequent extra precision docking by Schrodinger.The top eight predicted coumarin derivatives with comparably high predicted binding affinity to HSA are shown in Table S2.Among the eight coumarin analogues, 4-hydroxycoumarin possessed the highest predicted binding affinity (energy, À6.91 kcal/mol; rank, 1) to HSA.

Molecular-protein docking by AutoDock
AutoDockTools 1.5.6 is a suite of software for simulating small-molecule and macromolecule systems.It was used to perform docking simulation of the truncated OA derivatives to full-length HSA (PDB: 1AO6) and FABP (PDB: 1G5W), respectively.Some docking modes were generated.Both HSA and FABP should delete water and add hydrogens before setting as macromolecules.All small molecules should add hydrogens before setting as ligand.Additionally, the number of rotatable bonds of ligands was selected by software automatically.Before running autogrid, the grid box should be set to cover the whole macromolecule, and the ligand should be outside the grid box.Before running autodock, macromolecule should be set as a rigid receptor.Both user-specified initial position and initial relative dihedral offset (quat0) were random.Genetic algorithm parameters were selected as follows: number of GA Runs was selected 20; maximum number of evals was selected 2,500,000; maximum number of generations was selected 27,000.All the docking parameters were set to defaults, and Lamarckian Genetic Algorithm 4.2 (LGA) was selected to calculate the lowest energy of the system.The docking results are shown in Table S3.Among all the tested truncated OA derivatives, dodecanedioic acid demonstrated the lowest predicted binding affinity (energy, À0.23 kcal/mol; rank, 1) to FABP (PDB: 1G5W) but comparably high predicted binding affinity to HSA (PDB: 1AO6).

Molecular dynamics simulation
Amber20 is a suite of biomolecular simulation programs for simulating small-molecule and macromolecule systems.It was used to perform molecular dynamics simulations of the coumarin moiety of HC to full-length HSA (PDB: 1AO6) with the presence of the FA moiety of DA.All the small molecules were prepared by the antechamber module and then checked by the parmchk2 module separately.The HSA was prepared by the pdb4amber module and the reduce module.After the preparation of small molecules and HSA, all the structures were combined and prepared using the LEaP module.Additionally, the protein force field was set as ff14SB and the water model was set as TIP3P.Then the PMEMD.CUDA module was used as the molecular dynamics engine to perform molecular dynamics simulations.Cluster analysis was performed once the molecular dynamics simulation process was completed.The binding pocket of HC was located inside the HSA structure.Both HC and DA were conjugated the Apc001 to form an Apc001OC; therefore, the binding sites between DA and HSA should be near the HC binding pocket, which was within the distance of 30 Å.As a result, three binding pockets (Z1, Z2, Z3; Figure S19) within a distance of 30 Å from the HC binding pocket were selected.First, we conducted molecular dynamics simulations with a short timescale (13 ns) to predict the binding modes between HC and HSA, with or without the presence of DA in the three binding pockets.The results are shown in Figure S20.With the presence of DA in pocket Z3, HC possessed the highest binding affinity to HSA.There were four predicted binding sites between HC and HSA: Y146, K195, H238, and R253.In order to obtain more reliable results, we further conducted a molecular dynamics simulation with a longer timescale (30 ns) for precise prediction of binding affinity of HC to HSA, with the presence of DA in pocket Z3.The predicted results are summarized in Figure 1.Without the presence of DA, there were two predicted binding sites between HC and HSA: Y146 and R253.Notably, with the presence of DA in pocket Z3, there were five predicted binding sites between HC and HSA: K191, K195, R218, H238, and R253.Also, R253 was the only unaltered predicted binding site of HC to HSA, with or without the presence of DA in pocket Z3.

Preparation of DA
In a flame-dried 100-mL flask, dodecanedioic acid (2.30 g, 10.0 mmol, 1.0 equiv) was dissolved in 30 mL of anhydrous tetrahydrofuran (THF) by stirring at 0 C. DCC (2.68 g, 13.0 mmol, 1.3 equiv) was added slowly to the mixture.After 10 min, the solution of 1-hydroxypyrrolidine-2, 5-dione (1.15 g, 10.0 mmol, 1.0 equiv) in 5.0 mL of anhydrous THF was added to the mixture at the same temperature.After 20 min, 4-DMAP (20.0 mg) was added to the mixture.The resulting mixture was warmed up slowly to room temperature and stirred overnight.The solid was filtered off and washed with THF.The filtrate was concentrated in vacuo and products were purified by column chromatography (eluent: dichloromethane/MeOH = 40:1-20:1, v/v) to afford the pure product DA (2.05 g, 62.6% yield), a white solid.High resolution electrospray ionization mass spectrometry (ESI-HRMS) calculated for C 16

Preparation of the Apc001-SH-NH 2
The Apc001-SH-NH 2 (sequence: NH 2 -C4-(SH-C6C1)C2-C(OMe) G(OMe)G(OMe)G(OMe)GTGTGGGTTCGTCGTTAGCTTGATT TGGCAGCU(OMe)G(OMe)C(OMe)C(OMe)-idT) was synthesized on a 32-mmol scale on an ÄKTA oligopilot plus 100 standard DNA/RNA synthesizer using commercially available 3 0 -IDT CPG, 5 0 -O-DMT-2 0 -deoxynucleoside (A Bz , C Ac , G iBu , and T) phosphoramidite monomers, 5 0 -O-DMT-2 0 -O-methyl nucleoside (A Bz , C Ac , G iBu and U) phosphoramidite monomers, amino-modifier C7 CE phosphoramidite monomer, and thiol-modified C6 S-S phosphoramidite monomer.All oligonucleotides were synthesized in DMT-OFF mode.After completion of the synthesis, the solid support was suspended in 28% ammonium hydroxide solution and heated at 60 C for 4 h to release the product from the support and to complete the removal of all protecting groups.The solid support was filtered, and the filtrate was desalted/buffer exchanged into ddH 2 O and lyophilized to dryness.The mixture was then dissolved in 10 mL of pH = 8.4 ddH 2 O by stirring at room temperature and 100 mM dithiothreitol (DTT) solution was added to the mixture.The resulting mixture was stirred for 2 h.After the reaction was completed, the resulting mixture was lyophilized to dryness.

Preparation of Apc001-SH
The Apc001-SH (sequence: SH-C6-C(OMe)G(OMe)G(OMe)G(OMe) GTGTGGGTTCGTCGTTAGCTTGATTTGGCAGCU(OMe)G(OMe) C(OMe)C(OMe)-idT) was synthesized on a 32-mmol scale on an ÄKTA oligopilot plus 100 standard DNA/RNA synthesizer using commercially available 3 0 -IDT CPG, 5 0 -O-DMT-2 0 -deoxynucleoside (A Bz , C Ac , G iBu , and T) phosphoramidite monomers, 5 0 -O-DMT-2 0 -O-methyl nucleoside (A Bz , C Ac , G iBu and U) phosphoramidite monomers, and thiol-modified C6 S-S phosphoramidite monomer.All oligonucleotides were synthesized in DMT-OFF mode.After completion of the synthesis, the solid support was suspended in 28% ammonium hydroxide solution and heated at 60 C for 4 h to release the product from the support and to complete the removal of all protecting groups.The solid support was filtered, and the filtrate was desalted/buffer exchanged into ddH 2 O and lyophilized to dryness.The mixture was then dissolved in 10 mL of pH = 8.4 ddH 2 O by stirring at room temperature and 100 mM DTT solution was added to the mixture.The resulting mixture was stirred for 2 h.After reaction was completed, the resulting mixture was lyophilized to dryness.

Preparation of the Apc001-NH 2
The Apc001-NH 2 (sequence: NH 2 -C6-C(OMe)G(OMe)G(OMe) G(OMe)GTGTGGGTTCGTCGTTAGCTTGATTTGGCAGCU(OMe) G(OMe)C(OMe)C(OMe)-idT) was synthesized on a 32-mmol scale on an ÄKTA oligopilot plus 100 standard DNA/RNA synthesizer using commercially available 3 0 -IDT CPG, 5 0 -O-DMT-2 0 -deoxynucleoside (A Bz , C Ac , G iBu , and T) phosphoramidite monomers, 5 0 -O-DMT-2 0 -O-methyl nucleoside (A Bz , C Ac , G iBu and U) phosphoramidite monomers, and amino-modifier C6 CE phosphoramidite monomer.All oligonucleotides were synthesized in DMT-OFF mode.After completion of the synthesis, the solid support was suspended in 28% ammonium hydroxide solution and heated at 60 C for 4 h to release the product from the support and to complete the removal of all protecting groups.The solid support was filtered, and the filtrate was desalted/buffer exchanged into ddH 2 O and lyophilized to dryness.

Purification of the modified Apc001
The crude modified Apc001 was dissolved in 40 mL of ddH 2 O, and then the mixture was filtered with 0.45-mm polytetrafluoroethylene (PTFE) membrane to obtain clear filtrate.The crude product was purified by preparative liquid chromatography with the Autotide 100 Oligo Purification System to obtain the modified Apc001 (condition: column, Autotide-RP-20 C18, 250 Â 22 mm, 1 mm; 26 C; a flow rate of 6 mL min À1 ; a linear gradient of 5-35% v/v solvent B in 30 min, A was 0.05 M triethylamine acetate [TEAA] solution, B was HPLC grade acetonitrile).

Preparation of the Apc001OC conjugate
In a flame-dried 100-mL flask, HC (240 mg, 100 equiv) was dissolved in 4 mL of DMF by stirring at room temperature.Then the solution of 100 mg of Apc001-SH-NH 2 in 6 mL of 1Â PBS solution (pH = 6.71) was added to the mixture.The resulting mixture was stirred overnight.The solution of NaHCO 3 (150 mg) in 3 mL of ddH 2 O was added to the mixture.Then the solution of DA (250 mg, 100 equiv) in 2 mL of DMF was added to the mixture.The resulting mixture was stirred overnight again.After reaction was completed, the mixture was concentrated in vacuo to obtain the crude Apc001OC conjugate.

Preparation of the Apc001HC conjugate
In a flame-dried 100-mL flask, HC (240 mg, 100 equiv) was dissolved in 4 mL of DMF by stirring at room temperature.Then the solution of 100 mg of Apc001-SH-NH 2 in 6 mL of 1Â PBS solution (pH = 6.71) was added to the mixture.The resulting mixture was stirred overnight.After reaction was completed, the mixture was concentrated in vacuo to obtain the crude Apc001HC conjugate.

Preparation of the Apc001DA conjugate
In a flame-dried 100-mL flask, NaHCO 3 (150 mg) was dissolved in 4 mL of ddH 2 O by stirring at room temperature.Then the solution of DA (250 mg, 100 equiv) in 2 mL of DMF was added to the mixture, and 100 mg of Apc001-NH 2 in 6 mL of ddH 2 O was subsequently added.The resulting mixture was stirred overnight.After reaction was completed, the mixture was concentrated in vacuo to obtain the crude Apc001DA conjugate.

Purification of theApc001OC/HC/DA conjugate
The crude Apc001OC/HC/DA conjugate was dissolved in 20 mL of ddH 2 O, and then the mixture was filtered with 0.45-mm PTFE membrane to obtain clear filtrate. 42,43The crude product was purified by preparative liquid chromatography with the Autotide 100 Oligo Purification System to obtain the pure Apc001OC/HC/DA conjugate (condition: column, Autotide-RP-20 C18, 250 Â 22 mm, 1 mm; 26 C; a flow rate of 6 mL min À1 ; a linear gradient of 5-35% v/v solvent B in 40 min, A was 0.05 M TEAA solution, B was HPLC grade acetonitrile).

SPR analysis
Every protein (or peptide) was immobilized on the CM5 chip (GE) by the method of amine coupling.Conditions for immobilization of protein (or peptide) contained several parameters as follows: surface activation and ligand attachment were performed at 25 C; specified contact time was 420 s; the concentration of protein or peptide was 50-80 mg/mL (every protein or peptide was dissolved in 10 mM acetate, pH = 4.5 or 5.0, according to the results of immobilization pH scouting); the aqueous buffer of aptamer conjugate ligands was 1Â PBS; and the aqueous buffer of small-molecule ligands was 1Â PBS with 5% DMSO.Other chemicals required for immobilization included 50 mM NaOH solution, ethanolamine, EDC, and NHS.The regeneration scouting contained several parameters as follows: number of regenerations was one; the Prime before run option was selected; contact time was 60 s; stabilization period was 5 s; number of conditions was 3 s; and number of cycles for each condition was five.The regeneration solution for regenerating protein (or peptide) was 10 mM glycine-HCl solution (pH = 3.0 or 3.5).The mode of multi-cycle kinetics with a 1:1 fit was selected to perform experiments.Different concentrations of ligands were used for analyzing the binding affinities to proteins.The buffer and flow cell-interaction effects have been subtracted from the binding curves.

The design of blocking peptides
Based on the calculated binding sites of HSA to the coumarin moiety of HC, it was predicted that Y146 and R253 on HSA were involved.Then, we designed a peptide sequence (peptide A) including the above binding sites and verified its binding affinity to HC by SPR analysis (Figure S4A).Based on the calculated binding sites of HSA to the FA moiety of DA, it was predicted that K16, K132, and K158 on HSA were involved.Then, we designed a peptide sequence (peptide B) including the above binding sites and verified its binding affinity to DA by SPR analysis (Figure S4B).Based on the above prediction, a peptide sequence including the above binding sites that could bind to HC-DA combination was designed (peptide C).Then, we verified its binding affinity to both HC and DA by SPR analysis (Figure S4C).Based on the calculated binding sites of HSA to HC in the presence and absence of DA (Figure S5A), it was predicted that Y146 and R253 on HSA were involved in the absence of DA and K191, K195, R253, R238, and H238 on HSA were involved in the presence of DA.Interestingly, K191, K195, R238, and H238 were the newly participated binding sites in the presence of DA (Figure S5B).Then, we designed a peptide sequence (peptide D) including the above newly participated binding sites and verified its binding affinity to HC by SPR analysis (Figure S5C). 6,7All the peptide sequences were purchased from GL Biochem (Shanghai).

Blood sampling
After subcutaneous injection of the Apc001OC conjugate, the Apc001DA conjugate, the Apc001HC conjugate, peptide A@Apc001OC, peptide B@Apc001OC, peptide C@Apc001OC, and peptide D@Apc001OC, 500 mL of blood was taken via orbital vein from each rat at different time points and was collected into tubes containing sodium heparin as an anticoagulant (0.9-mL vacutainers, BD Biosciences) and then were put into wet ice immediately. 44

Plasma isolation
Plasma was isolated by centrifugation at 6,000 Â g for 10 min at 4 C within 1 h after collection and stored at À80 C until analysis.After allowing each sample to thaw, the plasma was mixed and an aliquot (100 mL) transferred to a polypropylene tube.Methanol (1.4 mL) was added to each sample and, after mixing, the plasma protein was removed by centrifugation at 14,000 Â g for 10 min.An aliquot of the supernatant was transferred to a 250-mL limited-volume insert vial for Aptamer conjugate quantification.Then the supernatants were dried and reconstituted for aptamer quantification.Standards will be prepared in blank rat plasma containing sodium heparin. 45

Aptamer conjugate quantification
All the Apc001OC conjugate, the Apc001DA conjugate, and the Apc001HC conjugate could be assayed by MB technology. 46The MB sequence was Cy3-CCGCGCATCAAGCTAACGACGAAGCGCGG-BHQ2.The sequence of Apc001 was CmGmGmGmGTGTGGG TTCGTCGTTAGCTTGATTTGGCAGCUmGmCmCm-idT (the underlined sequence is the complementary sequence between MB and Apc001).The 96-well plate was used for loading samples.For each well, 5-mL plasma samples, 5 mM MB sequence, and 85 mL of PBS were added.Then, the reagents were mixed by gently tapping the sides of the plate.The plate was incubated and protected from light for 1 h at 37 C.The Apc001 could be detected when hybridization occurred between the complementary sequence of MB and Apc001.This caused the separation of the stem and hence of the fluorophore and the quencher.Once the fluorophore was no longer next to the quencher, illumination of the hybrid with light resulted in the fluorescent emission.The presence of the emission showed that the event of hybridization had occurred and hence the Apc001 sequence was present in the test sample.The plate was read using a fluorescence plate reader at excitation 550 nm and emission 575 nm.The concentrations of Apc001OC conjugate, the Apc001DA conjugate, and the Apc001HC conjugate were calculated according to the standard curves.Standard curves were obtained by analyzing different concentrations of Apc001 that were dissolved in plasma and processed as well as being analyzed using the same procedure.

The pharmacokinetics analysis
The Apc001OC conjugate, the Apc001DA conjugate, and the Apc001HC concentrations versus time profile were plotted and analyzed for each rat by software DAS 2.0 (BioGuider, Shanghai, China).All reported concentrations of the Apc001OC conjugate, the Apc001DA conjugate, and the Apc001HC conjugate were based on the area under the curve (AUC).Using the data from standards, calibration curves were generated for the Apc001OC conjugate, the Apc001DA conjugate, and the Apc001HC conjugate, respectively.Half-life (t 1/2 ) of the Apc001OC conjugate, the Apc001DA conjugate, and the Apc001HC conjugate were calculated according to the time it takes for the Apc001OC conjugate, the Apc001DA conjugate, and the Apc001HC conjugate to eliminate half of the maximum plasma concentration. 47Maximum plasma concentration (C max ) and the time to maximum plasma concentration (T max ) were obtained according to pharmacokinetics curves.9][50] Subsequently, the dosing intervals in multiple doses of the Apc001OC conjugate was calculated according to the equation D = 1/1 À e ÀKe$t (Ke is elimination constant, Ke is ln2/T 1/2 t is dosing interval; D is dosing ratio = loading dose/maintenance dose). 51,52The results are shown in Figures 4 and S7.

An OI mouse model for examining bone
Six-week-old Col1a2 +/G610C mice and 6-week-old WT littermates were employed to examine the bone anabolic potential of Analysis of the trabecular bone in distal femur was performed with micro-CT (version 6.5, vivaCT40, Scanco Medical, Bassersdorf, Switzerland).Images of femur were reconstructed and calibrated at the isotropic voxel size of 12.5 and 17.5 mm (70 kVp, 114 mA, 200 ms integration time, 260 thresholds, 1,200 mg HA/cm 3 ).Every measurement used the same filtering and segmentation values.Using the Scanco evaluation software, regions of interest (ROIs) were defined for trabecular parameters.For the trabecular bone, a central region was selected equivalent to 70% of the vertebral body height and extended from proximally to the end of the distal growth plate toward the vertebral body.We drew freehand the trabeculae ROI on 100 sequential slices to ensure it was within the endosteal envelope.Trabecular bone parameters, including Tb.vBMD, Tb.BV/TV, Tb.N, trabecular thickness (Tb.Th), and trabecular spacing (Tb.Sp) were calculated. 53,54ne histomorphometric analysis After sacrificing, the left distal femoral metaphysis and the left femoral mid-shaft were fixed in 4% paraformaldehyde for 48 h; dehydrated in increased 10%, 20%, and 30% concentrations of sucrose (dilution in 1Â PBS) for 24 h in each concentration; and embedded without decalcification in an optimal cutting temperature compound (Sakura Finetek, Tokyo, Japan).After embedding, the proximal region of the samples was sectioned longitudinally, and the histomorphometric analyses of trabecular bone was performed at the above four sites.The frozen tissue specimens were obtained at a thickness of 5 mm with CryoStar NX50 (Thermo Fisher Scientific, Waltham, MA, USA).The sites were consistent with the selected sites of micro-CT.Fluorescence micrographs for the bone sections were captured by a Q500MC fluorescence microscope (Leica, Bensheim, Germany).The parameters of bone dynamic histomorphometric analysis for trabecular bone and cortical bone included BFR/BS and MAR.The analysis was performed using professional histomorphometric analysis system (BIOQUANT OSTEO, Nashville, TN, USA), and the parameters were calculated and expressed according to the American Society for Bone and Mineral Research standardized nomenclature for bone histomorphometry. 55

Statistical analysis
All variables were expressed as mean ± standard deviation.One-way ANOVA with Tukey's post hoc test was performed to determine the inter-group differences in the study variables, including for anticoagulant assay, micro-CT parameters, and bone histomorphometric parameters.All the statistical data were analyzed by GraphPad Prism (version 8; GraphPad Software, San Diego, CA, USA), and p < 0.05 was considered to be statistically significant.For the in vivo experiments, sample size was pre-determined by a power calculation according to our previously published protocol.The animals were grouped randomly and blindly to researchers.The animals in poor body condition were excluded.

Figure 3 .
Figure 3.Both HC and DA were conjugated to a bone anabolic aptamer against sclerostin (Apc001) to form an Apc001OC conjugate with a high binding affinity to HSA (A) Schematic diagram of the binding mode of Apc001OC to HSA. (B) The multi-cycle kinetics analysis of binding affinity of Apc001 to HSA. (C) The multi-cycle kinetics analysis of binding affinity of Apc001DA to HSA (K d = 46.6 nM).(D) The multi-cycle kinetics analysis of binding affinity of Apc001HC to HSA (K d = 7.06 nM).(E) The multi-cycle kinetics analysis of binding affinity of Apc001OC to HSA (K d = 0.75 nM).(F) Left: the multi-cycle kinetics analysis of binding affinity of Apc001 to sclerostin (SOST) (K d = 37.1 nM).Right: the multi-cycle kinetics analysis of binding affinity of Apc001OC to SOST (K d = 35.6 nM).

Figure 5 .
Figure 5. HC and DA could synergistically facilitate the conjugated Apc001 promoting bone formation, increasing bone mass, and improving bone microarchitecture integrity in OI mice (A) Representative images showing three-dimensional trabecular architecture by micro-CT reconstruction at the distal femur.Scale bars, 0.1 mm (the upper panel).Bar charts of the structural parameters of Tb.vBMD, Tb.BV/TV, and Tb.N from ex vivo micro-CT examination at the distal femur for the OI mice injected subcutaneously with 25 mg/kg of Apc001OC conjugate, Apc001HC conjugate, Apc001DA conjugate, scApc001OC conjugate, Apc001PE conjugate, Apc001, peptide A@Apc001OC, peptide B@Apc001OC, peptide C@Apc001OC, or peptide D@Apc001OC once every 2 weeks (q2wk) for 12 weeks, respectively (the lower panel).(B) Representative fluorescent micrographs of the trabecular bone sections showing bone formation at the distal femur visualized by double calcein labels.Scale bars, 30 mm (the upper panel).Analysis of dynamic bone histomorphometric parameters of Tb.BFR/BS and Tb.MAR at the distal femur for the OI mice (the lower panel).BS, baseline; Veh, vehicle; Peptide A/B/C/ D@Apc001OC, Apc001OC that is pre-saturated by different blocking peptides; peptide A, a peptide sequence within HSA including the predicted binding sites that could (legend continued on next page)

Figure 6 .
Figure 6.Apc001OC with 50 mg/kg q2wk showed comparable effect on promoting bone anabolic potential at distal femur in OI mice when compared to sclerostin antibody with 25 mg/kg opw (A) Representative images showing three-dimensional trabecular architecture by micro-CT reconstruction at the distal femur.Scale bars, 0.1 mm (the upper panel).Bar charts of the structural parameters of Tb.vBMD, Tb.BV/TV, and Tb.N from ex vivo micro-CT examination at the distal femur for the OI mice injected subcutaneously with 50 mg/kg of Apc001OC conjugate q2wk or 25 mg/kg of romosozumab for 12 weeks, respectively (the lower panel).(B) Representative fluorescent micrographs of the trabecular bone sections showing bone formation at the distal femur visualized by double calcein labels.Scale bars, 30 mm (the upper panel).Analysis of dynamic bone histomorphometric parameters of Tb.BFR/BS and Tb.MAR at the distal femur for the OI mice (the lower panel).BS, baseline; Veh, vehicle; OC-50q2wk, 50 mg/kg of Apc001OC conjugate q2wk; mab-25opw, 25 mg/kg romosozumab opw.Data are expressed as mean ± standard deviation followed by one-way ANOVA with Tukey's post hoc test, n = 6 per group.*p < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.0001.