Phage display technology relies on a peptide library to acquire functional motifs. An exogenous gene coding a short peptide was inserted into side chain gene of shell protein of a filamentous phage III (p III) or IV. A fusion protein attached to N-terminal of (p III) or IV that expressed on the shell surface of phages. Thus, specific lengths of phage aggregation of different sequences of exogenous peptides are presented, which constitute a good coverage peptide library (Smith and Scott, 1993). Phages were used to bind targets, and short peptides with highly affinity with targets were obtained after multiple screening. Finally, motifs were obtained by chemical synthesis.

In 1998, Cwirla et al. screened the phage peptide library through the human TPO receptor and obtained a 14-peptide sequence, which had similar activity with TPO in vitro (Dower et al., 1998). In the same year, Lowman et al. used insulin-like growth factors-binding protein (IGFBP) to obtain an insulin-like growth factors (IGF) related peptide motif. The acquired motif can bind to IGFBP and function in cartilage repair (Lowman et al., 1998). Subsequently, Ballinger et al. screened polypeptide C-19, which can mimic binding of basic fibroblast growth factor (bFGF) to the receptor and showed the same activity as FGF (Ballinger et al., 1999). Koishi et al. discovered a peptide sequence, HSNGLPL, with binding affinity to TGF-β1, indicating important influence on the formation of connective tissue (McLennan and Koishi, 2004). Gelain et al. screened PFSSTKT and SKPPGTSS, functional short peptides derived from bone marrow homing peptides (BMHP), to recruit stem cell for knee cartilage repair (Gelain et al., 2006). Akkiraju et al. screened and synthesized bone morphogenetic protein (BMP) functional motifs CK2.1 (Syed), CK2.2 (SLYD), and CK2.3 (SLKD). CK2.1 was the most promising peptide that induces chondrogenesis rather than osteogenesis (Akkiraju et al., 2017a). The limits of phage display technology were high cost of time and labor in phage preparation for screening motif peptides.

Highly repetitive sequences might exist in some proteins. Pierschbacher et al. (1985) found RGD sequence in fibronectin, collagen, and thrombin, which can bind to the fibronectin receptor on the cell surface and promote cell adhesion. In addition, Gelain et al. used amino acids K, p, F, S, and T to synthesize a series of short peptides with strong affinity to stem cells, including PFSSTKT and SKPPGTSS (Gelain et al., 2006). Williams et al. (2000) found neural cadherin (N-cadherin) has an evolutionarily conserved sequence, His-Ala-Val (HAV), which provides a homophile cell adhesion domain to mediate cell-cell adhesion (Blaschuk et al., 1990). Williams et al. performed a series of amino acid modifications on the HAV sequence, and found that acquired peptides showed similar binding ability to N-cadherin (Williams et al., 2001). Bian applied HAV in cartilage tissue engineering and thereby promoted the synthesis of cartilage matrix in rats model (Bian et al., 2013). However, some problems remain alongside with this method. Firstly, not all peptides possess highly repetitive conserved sequences that could be recognized. Even if a series of motifs are chosen, the sequence summarization of motifs depends on the experience of researchers, and the bioactivity of motifs needs to be verified.

For some proteins without highly repetitive sequences, microarray chip technology can be used to systematically screen specific sequences (Foong et al., 2012). In microarray chip assay, thousands of short molecules—such as DNA, peptides, small chemical molecules, and cells—are arrayed on a chip as receptor (Sun et al., 2006; Uttamchandani et al., 2006). Then, the entire sequence of the target protein is sequentially cut to equal lengths, and acquired peptides were administrated in the slide above. The peptides that show higher affinity to the receptor are chosen, and finally sequences are analyzed for further application. Owing to the outstanding advantages of miniaturization and parallelization, microarray chip technology has been widely used in peptide screening (Deng et al., 2020), antibody detection (Al‐Majdoub et al., 2013), and vaccine preparation (Gaseitsiwe et al., 2008). It should be noted that systematically screened sequences were usually short, mainly 8–15 amino acids, with partial simulation on domain structures of original protein. Improved mimic efficiency of biological function with moderate sequence is required in further motif studies before clinical application.

Microstructure simulation based on analyzing the microstructure of the functional domain, and adjusting the atomic arrangement and bonding angle to simulating the spatial structure. For example, Bhatnagar directly analyzed the structure of TGF-β, and proposed that the β-turn structure is of vital significance for its bioactivity. Then, a series of short peptides containing six or seven amino acids was developed, termed as cytomodulin (CM) family (Bhatnagar et al., 2003). Zhang et al. combined CM10 (LIANAK) with functional nanofibrous hollow microspheres (FNF-HMS), and implated subcutaneously in the backs of mice. Finally, ectopic cartilages were realized (Zhang et al., 2015a). However, the structure of functional domain is not fixed on different conditions, so as to realize the expose and block of bioactivity site. Simple structural simulation may not simultaneously satisfy the microscopic requirement in activation and inactivation of motifs.

TGF-β is a family of proteins that regulates key cellular processes involved in early embryonic development (Peng et al., 2022), cell growth (Frangogiannis, 2022), differentiation (Moreau et al., 2022), motility, and apoptosis (Weiss and Attisano, 2013). As known to all, TGF-β is vital in maintaining articular cartilage normality and joint integration. Bhatnagar selected a series of short peptides containing six or seven amino acids from the structural characteristics of TGF-β, named as the CM family (Bhatnagar et al., 2003), and main members include CM1 (ANVAENA), CM2 (LIAEAK), and CM10 (LIANAK). El-Sakka (El-Sakka et al., 1997) and Basu (Basu et al., 2009)applied CM1 and CM10 locally in mouse skin wound model, and found improvement in collagen I expression and wound strength (Figure 2A, B). Renner et al. cultured human mesenchymal stem cells (hMSCs) with free CM1 and CM2, but no significant difference was found in glycosaminoglycan (GAG) production, compared with negative controls (Figure 2C) (Renner and Liu, 2013). In contrast, Park et al. added CM2 into medium to culture human periodontal ligament stem cells (hPLSCs) and found increased expression of SOX9, ACAN, and COL2A1. Moreover, compared with the non-covalent binding mode, the covalently combined CM2 with Cx-HA exerted longer influence on GAG deposition (Figure 2D) (Park et al., 2019). Similarly, Zhang et al. covalently grafted CM10 onto FNF-HMS and found that the functionalized scaffold strengthened chondrogenic differentiation in vitro. Then the scaffold were injected subcutaneously into mice, which showed ectopic cartilage formation (Figure 2E) (Zhang et al., 2015a).

Hyaline cartilage locates at the end of the long bone, lacks blood supply and reserve cells. Its lubrication and feeding relies on articular fluid at the joint capsule. Similarly, the poor self-repair ability of articular cartilage defects is poor for that the chemokines are usually too insufficient in the defect area to guide the accumulation of stem cells (Chen et al., 2021). For example, BMHP is able to enter bone marrow and bind to BMSCs in vivo, which does favor to induce MSCs to migrate, proliferate, differentiate and synthesize matrix to repair cartilage defect (Liu et al., 2013).

Nowakowski et al. (2004) screened a series of peptides with strong affinity for stem cells, and found that the amino acids K, p, F, S, and T are crucial for biological functions Subsequently, Gelain et al. verified the bioactivities of BMHP1 (PFSSTKT) and BMHP2 (SKPPGTSS), as mimetic peptides of BMHP that rich in amino acids K, p, F, S, and T. Stem cell recruitment was realized with the two peptides as well as an improved trend of cell differentiation (Figure 3A) (Gelain et al., 2006). Lu et al. combined the PFSSTKT short peptide with RAD to produce a functionalized SAP hydrogel, which stimulated MSC proliferation, attachment, and chondrogenic differentiation in rabbit model. Acellularized cartilage matrix (ACM) scaffold was combined with SAP hydrogel to form ACM + RAD/PFS and implanted into full-thickness articular cartilage defect area, and it was found that the cartilage defect was completely covered by chondroid tissue (Figure 3B) (Cao et al., 2015; Lu et al., 2018). Sun et al. introduced the SKPPGTSS short peptide and RAD into a three-dimensional (3D) porous decellularized porcine articular cartilage matrix (DCM) scaffold to form a bone marrow-specific homing scaffold system, DCM-RAD/SKP. Experiments in vitro showed that rabbit bone marrow stem cells migrated to DCM-RAD/SKP scaffold, and the cell/scaffold system were implanted into rabbit knee cartilage defect model. Computed Tomography (CT) results showed that the cartilage defect area was filled with uniform regenerated tissue, highly similar with surrounding normal cartilage, and successfully reconstructed the subchondral bone (Figure 3C) (Sun et al., 2018). According to studies above, the peptides PFSSTKT and SKPPFTSS successfully recruited stem to specific location, in spite of original tissue. However, PFSSTKT and SKPPFTSS exerted limited effect on differentiation, indicating that differentiation depends on other conditions, such as environment.

BMPs are members of TGF-β family, and play an important role in regulating cell behavior and tissue regeneration (Zhou et al., 2022). BMPs participate in the differentiation of MSCs into bone, cartilage, ligaments, tendons, and nerves (An et al., 2010). Saito et al. Found that KIPKASSVPTELSAISTLYL, a 20-amino acid sequence from residues 73–92 of BMP2, may be one of the receptor binding sites to induce calcification of fibroblasts (Saito et al., 2003). Renner et al. used KIPKASSVPTELSAISTLYL to culture hMSCs, and found the GAG production was higher after 2 weeks (Figure 4A) (Renner and Liu, 2013). More importantly, Kim et al. found that compared with BMP2 group, cells treated with KIPKASSVPTELSAISTLYL exhibited almost no increase in hypertrophy markers but significantly increased secretion of cartilage matrix (Figure 4B) (Renner et al., 2012). In addition, Akkiraju et al. designed a novel mimic peptide CK2.1(Syed) from BMP receptor Ia (BMPRIa) and injected it into the tail veins of mice. CK2.1 promoted chondrogenesis without inducing chondrocyte hypertrophy, and the effect was better than that of BMP (Figure 4C) (Akkiraju et al., 2017a). Subsequently, Akkiraju injected CK2.1-loaded hydrogel particles (HGP) into articular cartilage defects of mice, and found that defects were filled with typical neocartilage of less hypertrophy (Figure 4D) (Akkiraju et al., 2017b).

N-cadherin is a calcium ion-dependent adhesion glycoprotein and function in maintaining cell structure and motility (Tepass et al., 2000). In chondrogenesis, N-cadherin mediates the aggregation and condensation of mesenchymal cells, like chondrogenic progenitor cells (Tavella et al., 1994; Richardson et al., 2007). Gao suggested that N-cadherin-mediated cell-cell interactions were of great significance in mesenchymal cell densification and chondrogenesis (Gao et al., 2010).

Williams et al. found that N-cadherin has an evolutionarily conserved sequence HAV (Williams et al., 2000), which provides a homophile cell adhesion recognition site and mediates cell-cell adhesion (Blaschuk et al., 1990). Relative studies clarified that inhibition of the HAV peptide weakened cell-cell adhesion (Figure 5A) (Williams et al., 2000). Williams et al. performed a series of amino acid modifications on the basic HAV sequence and synthesized peptides showed similar affinity with N-cadherin (Williams et al., 2001). Bian incorporated HAV into a MeHA hydrogel to promote the early expression of chondrogenic genes and latter cartilage-specific matrix production (Figure 5B) (Bian et al., 2013). Cagla et al. combined HAV and the amphiphilic peptide nanofiber system E-PA, and realized promoted mesenchymal cell cohesion and cartilage formation (Figure 5C) (Eren Cimenci et al., 2019). Li chemically grafted HAVDI with a KLD12 self-assembled peptide hydrogel and found that the nuclear translocation of β-catenin and the synthesis of type X collagen at the early stage of chondrogenesis. By the downregulation of the WNT/β-catenin pathway, the expression of chondrogenic genes was improved and cartilage matrix was preserved (Figure 5D) (Li et al., 2017a). Kwon et al. suggested that the HAV peptide enhanced the early expression of chondrogenic markers and promoted the long-term deposition of cartilage matrix in a strongly dose-dependent manner (Figure 5E) (Kwon et al., 2018).

Integrin is an adhesion protein on the membrane that can transmit signals by interacting with extracellular matrix (ECM), so as to regulate key cellular processes, such as cell differentiation, proliferation and migration (LaFlamme et al., 2018). Interestingly, integrin can improve the adhesion between cells and ECM, and allow cells to adapt to the surrounding environment better (Place et al., 2009). RGD sequence has been found in multiple ECM proteins that promote cell adhesion, such as fibronectin (Underwood et al., 1995), laminin (Wang et al., 2021), tenascin (Comisar et al., 2007), and thrombospondin (Benoit and Anseth, 2005). RGD binds integrins to anchor cells on the surface of matrix and enhance cell migration, adhesion, extension, and proliferation. Patterson et al. grafted RGD to polyethylene glycol (PEG) hydrogels to culture human periosteum-derived cells (hPDCs). The cells maintained high viability and promoted the expression of chondrogenic genes and synthesis of GAGs (Figure 6A) (Kudva et al., 2018). Li et al. conjugated RGD to gold nanoparticles (Au-NPs) and improved differentiation of hMSCs into chondrocytes (Figure 6B) (Li et al., 2017b). Alsberg et al. injected mixture of calcium alginate hydrogel-GRGDY-chondrocytes subcutaneously at the backs of rats, and found typical ectopic chondrogenesis (Figure 6C) (Alsberg et al., 2002). However, Zhang et al. suggested that RGD peptides promote chondrocyte proliferation and differentiation, but simultaneously lead to chondrocyte hypertrophy and differentiation (Figure 6D) (Zhang et al., 2015b). Salinas et al. believed that the influence of RGD on cells based on time. When the cleavable CPENFFGGRGDSG system was added to the PEG hydrogel loaded with hMSCs, early chondrogenesis could be induced. The removal of RGD reduced the risk of inhibited chondrogenesis (Figure 6E) (Salinas and Anseth, 2008). Li et al. cultured chondrocytes in PEG hydrogel loaded with RGD. The chondrocyte dedifferentiation was inhibited when the microscopic distance was over 70 nm. Then, 70 nm was consider the most beneficial distance of RGD stimulation to maintain the phenotype of chondrocytes (Figure 6F) (Li et al., 2015). Mechanical stimulation was also thought of as one factor that affects RGD function, according to the role of integrin in the mechanical signal transduction pathway of chondrocytes. Studies by Villanueva showed that RGD has a negative effect on the phenotype of chondrocytes without a dynamic load. However, the phenotype of chondrocytes could be maintained after dynamic compression (Figure 6G) (Villanueva et al., 2009).

Both insulin and insulin-like growth factor 1 (IGF1) play key roles in chondrogenesis. Insulin, an important component in almost all chondrogenic supplements (Puetzer et al., 2010), was reported to be significant in chondrocyte redifferentiation and could independently induce cartilage matrix synthesis of chondrogenic cell line ATDC5 (Shukunami et al., 1996). IGF1 is an important growth factor in chondrogenesis and the regulator of cartilage homeostasis. IGF1 promotes expression of chondrogenic gene, synthesis of collagen type II and proteoglycans, cell proliferation, and inhibits matrix decomposition mediated by osteoclasts (Trippel, 1995). IGF1 was found to interact with insulin receptors. Therefore, insulin-related peptides may also play a role similar to that of insulin and IGF1 for cross-reactivity (Phornphutkul et al., 2006). Renner et al. designed an insulin-mimetic peptide, GRVDWLQRNANFYDWFVAELG (GRV), which exhibited a high affinity for the insulin receptor. Then, the peptide GRV at different dosage of 0.01 µM, 0.1 µM, and 1 µM were used to culture hMSC. The results showed that differentiated chondrocytes, originating from hMSCs containing peptide GRV and TGF-β3, secreted more GAG than the control group of TGF-β3. It is believed that, with the presence of insulin and TGF-β3, the insulin functional motif GRV promoted chondrogenic differentiation and cartilage matrix deposition (Figure 6H) (Renner and Liu, 2013).

PTH is a single-chain polypeptide hormone synthesized and secreted by parathyroid cells (Suva and Friedman, 2022). PTH can regulate calcium and phosphorus metabolism at bone, kidneys and small intestine (Lombardi et al., 2020). The differentiation of MSCs into chondrocytes is usually accompanied with terminal hypertrophic differentiation (Fischer et al., 2010). PTH guides the differentiation of MSCs into chondrocytes (Chang et al., 2009) but counteracts hypertrophic differentiation (Jiang et al., 2008), so as to maintain the pehnotype of chondrocytes (Kim et al., 2008; Weiss et al., 2010). Suva et al. (Moseley et al., 1987) found that parathyroid hormone-associated protein (PTHrP) and PTH share similar sequence at residues 1–34, and exert similar effects by the same receptor, PTH1R (Weisser et al., 2002). Kafienah et al. (2007) . demonstrated that PTHrP can inhibit the hypertrophic differentiation of BMSC (Figure 7A) Fischer et al. cultured BMSCs with PTHrP in conditioned medium from human articular chondrocytes, and successfully reduced the expression of hypertrophy markers such as COL10A1 and alkaline phosphatase (Figure 7B) (Fischer et al., 2010). Meanwhile, Fischer et al. cultured MSCs in an environment of continuous or intermittent PtHrP(1–34) stimulation. The results showed that continuous stimulation inhibited chondrogenic differentiation of MSCs, whereas intermittent stimulation increased the deposition of cartilage matrix but inhibited hypertrophic differentiation (Figure 7C) (Fischer et al., 2014). In addition, the point of stimulation time on chondrogenesis was also studued. Rajagopal et al. supplemented PtHrP (1–34) into chondrogenic medium from day 4, and found that chondrogenic markers were significantly increased. At the same time, hypertrophy markers were significantly reduced compared to those treated with PtHrP from day 14 (Figure 7D) (Rajagopal et al., 2021). Kim et al. (2008) supplemented PTHrP (1–34) into chondrogenic medium for BMSCs and adipose tissue mesenchymal stem cells (ADSCs) from 14th day. The staining results indicated promoted chondrogenesis and inhibited hypertrophy in vitro (Figure 7E) (Kim et al., 2008). Zhang et al. (2013) injected PTHrP(1–40) in the articular cavity of rabbits at three time points after osteochondral injury. The results showed that, at 4–6 weeks group, articular cartilage exhibited better morphology and lower expression of hypertrophy markers than others (Figure 7F) (Zhang et al., 2013). Baron et al. cocultured human nasal septal chondrocytes with MSCs (NC/MSC) in a 1:3 ratio and added PTHrP. The results indicated that PTHrP inhibited the expression of hypertrophic markers in a dose-dependent manner and promoted secretion of cartilage matrix. PTHrP was found to eliminate cartilage calcification in vivo after the implantation of PTHrP-containing pellets into immunodeficient nude mice. In contrast, the control group exhibited obvious cartilage calcification (Figure 7G) (Anderson-Baron et al., 2021).

Natural materials, such as liposomes, collagen, and polysaccharide compounds, are popular for wide sources and low cost. In addition, they show high stability, safety, hydrophilicity, and other advantages. However, disadvantages also exist, such as low drug loading and poor adhesion to cells. Therefore, natural materials need modification or combination with other materials.

Synthetic material refers to artificially produced micromolecular monomers. Through chemical crosslinking, casting mold or 3D printing technology, the monomers assembled into macromolecular polymer materials. With excellent biocompatibility, highly standardized properties, mass production capacity, modification potential, and low immunogenicity, synthetic materials exhibit great potential as carriers in tissue engineering and regenerative medicine.

Non-covalent bonding mainly involves van der Waals forces, hydrophobic bonds, hydrogen bonds, and charge distribution. Non-covalent binding immobilizes active peptides on the surface of the material, mainly through physical adsorption force, so as to promote cell adhesion, proliferation, and differentiation. Simple dispersal is commonly used for physical adsorption. Kantlehner et al. physically immobilized RGD on titanium surface and achieved improved adhesion of osteoblasts (Kantlehner et al., 2000; Mas-Moruno et al., 2014). However, owing to the weak binding force, low adsorption rate, poor stability and repeatability, the application of non-covalent bonds is still limited.

Chemical coupling is a typical method of covalent bonding. After introducing active groups (such as -NH2, -OH, -COOH, and active hydrogen) on the surfaces of the carrier, the peptides reacts with carrier by crosslinking agents (CDI, APTES, PPY, and SMP) (Pallarola et al., 2014), leading to improved physical and chemical properties of peptides, such as stability and controlled release behavior (Zreiqat et al., 2003).

In cartilage tissue engineering, complicated techniques, such as layer-by-layer self-assembling technique, are often used to modify composite scaffold. The principle is that compounds are deposited alternately layer by layer, by interaction between monomers including the strong binding force of chemical bonds and the weak binding force of non-covalent bonds. In this way, the monomers of different layers can spontaneously form a film with stable properties and specific functions. Chua et al. used layer-by-layer self-assembly technology to prepare a HA/chitosan/PEM/titanium substrate, and chemically crosslink it with RGD, so as to improve the adhesion and proliferation of osteoblasts (Chua et al., 2008). Moreover, Yang et al. developed a polydopamine (PDA) coating layer under weak alkaline condition, which increased the adhesion of nerve growth factor (NGF) increased, and promoted the differentiation and proliferation of human neural stem cells (NSC) (Yang et al., 2012). This is based on the super-adhesive effect of the PDA coating, and the improved hydrophilicity and cell adhesion ensured the immobilization and bioactivity of peptides.

Synthetic peptides are superior to natural proteins in some ways. Firstly, functional motifs have good physical and chemical properties. 1) Functional motifs are peptide segments composed of amino acids, with simple structure and easy to adjust. 2) Functional motif can be loaded into different carriers by various methods, while maintained biological activity and stability. Zhang et al. added an extra cysteine residue to the C-terminus of CM10, which promoted the coupling of CM10 to the carrier, thereby improving the stability and function time of CM10 (Zhang et al., 2015a). Secondly, the functional motif has excellent biological properties. 1) The production of motifs by solid-phase synthesis does not require animal vectors. The low immunogenicity and no ethical issues make it possible in clinical application. 2) Excellent biocompatibility and biodegradability in natural physiological environments (Tiwari et al., 2012). 3) Potential targeting performance. Liposomes that bound with RGD peptide had been reported to exhibit targeting behaviors (Dharap et al., 2003). In a word, functional motifs can be directly synthesized in large quantities for their simple structures and low production costs. Therefore, functional motifs are increasingly popular in tissue engineering and regenerative medicine (Sekiya et al., 2002).

However, functional motifs also face some shortcomings in terms of efficacy and physical properties. As a polymer of amino acids, peptides may be inactivated for chemical degradations and physical changes under complicated environments. Moreover, some functional motifs show limited efficiency when compared with cytokines. For instance, Renner et al. supplemented TGF-β1- related motifs, CM1 and CM2, to culture hMSCs, and the results showed that cell pellets only produced significantly lower GAG compared with TGF-β1 positive controls (12%–13% for CM1 and 7% for CM2) (Renner and Liu, 2013).

Compared to macromolecular proteins, short peptide sequences are shorter and more easily to degrade by various proteases in organisms. In the future, short peptide molecules should be designed and synthesized, and the sequences should be modified to obtain more stable. For example, the stability of functional motif can be improved by ending blocking of acetylation and amidation, as well as blocking of ubiquitin modification sites.

Specific peptides of short sequence may only mimic partial structures of the functional domains. Thus, the administration dosages of short peptides are usually higher than that of whole proteins. Therefore, further researches are required to improve the simulation efficiency of biological activity with shorter sequences.

Peptides may form granular, tubular, radial, fibrous mesh, and other specific configurations by self-assembling, which can improve the adhesion and integration with defect areas. However, the microstructure also affects the bioactivity. How to obtain a balance between structure and function is a challenge in future researchers.

Short peptides and their carrier scaffolds remain the hotspots in tissue engineering researches and clinical practice. Application researches should be conducted based on the following aspects.

Firstly, the interface between cartilage and subchondral bone is calcification layer, which bears the mechanical stress of defect area. The interface layer is convex to the cartilage, leading to stress concentration and mechanical load. In future studies, we should focus on the mechanism and growth factors involved in formation of calcification interface.

In addition, endochondral ossification shares similar biological processes with cartilage degeneration and osteoarthritis progression, including chondrocyte hypertrophy, apoptosis, and degradation of the cartilage matrix. It is vital to explore that whether peptide motifs participated in the processes, and how to regulate the cartilage matrix absorption and chondrogenic regeneration.

Finally, previous studies based on the biological function of scaffold on cells or organs, but how does motifs and factors exert influence on scaffold materials need to be studied. The morphology and content of regenerative biomaterials influences the repair effect. For example, physiological regulation of pH can be used for scaffold shaping, such as self-assembly peptides, HA, and thermosensitive hydrogels, to adapt to the morphology of defect area. In addition, appropriate degradation of active groups in composite scaffolds make it possible to exert specific bioactivity at different time points.