Dendritic macromers for hydrogel formation: Tailored materials for ophthalmic, orthopedic, and biotech applications


  • Mark W. Grinstaff

    Corresponding author
    1. Department of Biomedical Engineering, Metcalf Center for Science and Engineering, Boston University, Boston, Massachusetts 02215
    2. Department of Chemistry, Metcalf Center for Science and Engineering, Boston University, Boston, Massachusetts 02215
    • Department of Biomedical Engineering, Metcalf Center for Science and Engineering, Boston University, Boston, Massachusetts 02215
    Search for more papers by this author


Dendritic macromolecules are well-defined highly branched macromolecules synthesized via a divergent or convergent approach. A salient feature of the macromolecules described herein, and a goal of our research effort, is to prepare dendritic macromolecules suitable for in vitro and in vivo use by focusing on biocompatible building blocks and biodegradable linkages. These dendritic macromolecules can be subsequently crosslinked to form hydrogels using a photochemical acrylate-based or a chemical ligation strategy. The properties—mechanical, swelling, degradation, and so forth—of the hydrogels can be tuned by altering the composition, crosslinking chemistry, wt %, generation number and so forth. The utility and diverse applicability is demonstrated through successful use of these hydrogels in three unique applications: hydrogel adhesives for repairing corneal wounds, hydrogel scaffolds for cartilage tissue engineering, and hydrogel reaction chambers for high throughput screening of molecular recognition events. © 2007 Wiley Periodicals, Inc. J Polym Sci Part A: Polym Chem 46: 383–400, 2008.


Dendritic macromolecules are finding ever increasing uses, and in the medical arena are being investigated as vehicles for drug delivery, contrast agents for imaging, synthetic vectors for nucleic acid transfection, sealants for tissue repair, and scaffolds for tissue regeneration among other uses.1–16 These varied uses are a consequence of the unique compositions, structures, and properties of these macromolecules. Dendrimers are highly branched macromolecules possessing three main structural components: a core, internal branching layers, and peripheral groups (Fig. 1).17–25 Unlike linear polymers where growth is accomplished by adding single monomers to the chain (1:1 growth), a dendrimer grows exponentially where each monomer is branched leading to multiple additions (1:2, 1:3, etc. growth). Each layer in a dendrimer is termed a generation (G) and thus as a dendrimer grows through the addition of new monomers, the generation number increases (G0, G1, … Gn). As the generation number of the dendrimer increases, the structure in solution adopts a globular conformation. The degree to which a dendrimer attains this globular shape is determined by the multiplicities of the core and branches, the orientation of the branching functionalities, the flexibility of the branching units, the length of the repeat unit, and the solvent environment.26–30 Examples of known peripheral groups included anionic (CO2), cationic (NR4+), neutral (NHC(O)[BOND]CH3), poly(ethylene glycol) (PEG), or alkyl chains and these functionalities play a significant role in the resultant properties. Taken all together, the chemical and structural attributes of dendrimers translate to unique chemical and physical properties (e.g., solubility, chemical reactivity, viscosity, glass transition temperature).

Figure 1.

Schematic of an idealized dendrimer.

Dendritic macromolecules are synthesized in a repetitive manner by either a divergent31–38 (from core to periphery) or convergent25, 39–45 (from periphery to core) approach. As with any synthesis requiring a series of stepwise reactions (e.g., coupling and deprotection reactions), high yields at each step are necessary to ensure preparation of ample material. Consequently, chemically well-defined, optimized, and robust reactions are used, such as amidations, esterifications, hydrogenolysis, and more recently click chemistry.46 In this highlight article, I describe (1) the synthesis of the crosslinkable dendritic macromolecules or macromers; (2) two different chemical crosslinking strategies with these macromers to prepare hydrophilic macroscopic structures (i.e., hydrogels); and (3) the successful application of these hydrogels for sealing corneal lacerations and securing corneal transplants, for repairing cartilage defects, and for creating localized hydrogel reaction chambers for high throughput screening.


We are synthesizing and evaluating degradable and biocompatible dendritic macromolecules that can be subsequently crosslinked to form hydrogels. We have reported the preparation and characterization of a variety of polyester, polyester–ether, and polyamide dendrimers and dendrons composed of biocompatible building blocks.13, 16, 47–57 We call these types of dendritic polymers “biodendrimers.” An example of a divergent synthetic approach to a generation fourth poly(glycerol-succinic acid) ([G4]-PGLSA[BOND]OH) dendrimer is shown in Scheme 1, where a glycerol-succinate monomer is added to a core using a series of stepwise esterification and hydrogenolysis reactions.48 Briefly stated, the tetra-functional G0 core, 2, was synthesized in two steps. First, the monomer, 1 (2-(cis-1,3-O-benzylidene-glycerol)succinic acid mono ester), was prepared by reacting succinic anhydride with cis-1,3-O-benzylidene-glycerol in pyridine. Next, 1 was coupled to 1.2 equivalents of cis-1,3-O-benzylideneglycerol, in the presence of two equivalents of N,N-dicyclohexylcarbodiimide (DCC) and 0.5 equivalent of 4-(dimethylamino)pyridinium 4-toluenesulfonate (DPTS) to yield the core, [G0]-PGLSA-bzld. The blzd group of the core was removed by hydrogenolysis (10% Pd/C and H2) in tetrahydrofuran to give the deprotected core, [G0]-PGLSA[BOND]OH, 2. Next, the monomer 1 was coupled to 2 in the presence of DCC and DPTS followed by hydrogenolysis to afford the G1 dendrimer. These esterification and hydrogenolysis reactions were repeated to give the higher generation dendrimers, and a G4 PGLSA dendrimer, 5, was prepared.

Scheme 1.

Divergent synthesis to a generation fourth poly(glycerol-succinic acid) ([G4]-PGLSA-OH) dendrimer.

The convergent synthesis to a generation three lysine–cysteine dendron ([G3]-(Lys)7-Cys8) is shown in Scheme 2.55 In this example, sequential amide forming condensation and deprotection reactions were used to prepare the dendron. The activated pentafluorophenol-esters of the amino acid building blocks were used, ZLys(Z)OPFP and IsoCys(Boc)OPFP as this coupling approach provided the highest yields and cleanest reactions. First, ZLys(Z)OPFP was coupled to LysOMe·2HCl in the presence of N,N-diisopropylethylamine (DIEA) and l-hydroxybenzotriazole (HOBT). The Z protecting group was removed via catalytic hydrogenolysis (10% Pd/C and H2). For the next step, BocLys(Boc) OPFP was coupled to the growing dendron. The Boc protected lysine derivative was used instead of Z protected since this gave better solubility of the larger Lys-dendron in organic solvents. Finally, the IsoCys(Boc) OPFP was added to the dendron and the Boc and Iso protecting groups of cysteine were removed using trifluoro acetic acid (TFA) and 1 N HCl in MeOH, respectively, to afford [G3]-(Lys)7-Cys8), 6. In both reaction schemes, the yields for each individual coupling and deprotection steps were 90% or better. Additional examples of recently prepared structures besides those shown in Schemes 1 and 2, include a generation four layered dendrimer composed of succinate and adapic layers (poly(glycerol-succinic-co-adapic acid) [G4,G3]-PGLAA-[G2,G1,G0]-PGLSA-OH) 7, a generation two-one PGLSA dendrimer possessing both carboxylic acid and alkyl chain peripheral groups ([G2-1]-PGLSA[BOND]CO2H[BOND]C14) 8, and a generation three hybrid dendritic-linear macromolecule ([G3]-PGLSA[BOND]OH)-PEG) 9 are shown in Figure 2. Initial cell cytotoxicity studies show that the dendritic macromolecules possess minimal toxicity and do not induce more death than what is witnessed in untreated controls.

Figure 2.

Structures of a generation four layered (poly(glycerol-succinic-co-adapic acid) [G4,G3]-PGLAA-[G2,G1,G0]-PGLSA-OH) dendrimer, a generation four poly(glycerol-succinic-acid) dendrimer possessing both carboxylic acid and alkyl chain peripheral groups ([G2-1]-PGLSA-CO2H-C14), and a generation three hybrid dendritic-linear macromolecule ([G3]-PGLSA-OH)-PEG).

Scheme 2.

Convergent synthesis to a generation three lysine-cysteine dendron ([G3]-(Lys)7-Cys8).

These synthetic routes to dendritic macromolecules, whether divergent or convergent, allow for precise compositional control within the core, internal branching layers, and peripheral groups of the macromolecule as well as the use of a wide-variety of different monomers. This level of control during synthesis enables the preparation of macromolecules possessing a unique molecular weight or very narrow molecular weight distribution, unlike most linear polymers. The narrow distribution of molecular weights allows for the correlation of a specific physical property, mechanical property, or biological response to a single specific chemical structure as opposed to a collection of different molecular weight structures, as with most linear polymers. The construction of such structure–property relationships and understandings is extremely useful in the design and evaluation of a macromolecule for an intended application.


Hydrogels are highly hydrated, crosslinked polymeric networks that are being investigated for a range of biomedical applications from drug delivery to scaffolds for tissue growth.58–62 We have reported two approaches for preparing hydrogels from these dendritic macromolecules. Importantly, both of these methods enable in situ formation of a hydrogel where an aqueous solution of the macromer (i.e., the crosslinkable derivative of the dendritic macromolecule) is delivered to an in vivo site and is subsequently crosslinked to form a three-dimensional hydrogel that conforms to the shape of the defect. In the first approach, the hydroxyl peripheral groups of the dendritic macromolecule are modified, before injection, to contain a functional group susceptible to free-radical polymerization, such as an acrylate.13 Upon free-radical polymerization, many acrylate groups on the dendritic macromolecules are crosslinked to afford a hydrogel (Fig. 3). This polymerization reaction can be initiated using a thermal- or a photo-activated catalyst. Given our interest in biomedical applications, the need to work in aqueous solutions, the requirement to work in the presence of biologics (e.g., proteins and cells), and the desire to minimize heat generation during the polymerization reaction, we have chosen to use a photochemical route. Specifically, we use a visible photoinitiating system that comprises eosin Y, 1-vinyl-pyrrilidinone, and triethanol amine. Excitation with an argon ion laser (λmax = 514 nm) of an aqueous solution containing the acrylate-modified dendritic macromolecule and a small quantity of the eosin Y photoinitiating system initiates the free radical polymerization of the methacrylate (MA) moieties on the dendritic polymer. This photoinitiating system has been used for a number of applications and is nontoxic.61, 63–65 The resulting hydrogel is hydrophilic, transparent, and, depending on the macromer and solution weight percent, can be soft and flexible or stiff.

Figure 3.

Photochemical crosslinking reaction to form the hydrogels.

In the second approach, the peripheral groups of the dendritic macromolecule are decorated with nucleophiles and subsequently reacted with another polymer containing electrophiles or vice versa. A number of nucleophile–electrophile crosslinking chemistries are available including the well known reactions of amines with N-hydroxysuccinimide or thiols with maleimide. However, we are interested in exploring crosslinking chemistry which occurs rapidly at 37 °C under neutral aqueous conditions without the generation of side-products and is amenable to preparing hydrogels with varying performance lifetimes. Moreover, the reactions must be chemoselective (i.e., only coupling between the correct partners) and possess a high tolerance to a range of other chemical functionalities (e.g., amines, thiols, carboxylic acids) that are present under physiological conditions. Consequently, we have selected reactions that belong to a family of chemical ligations, which have been applied successfully to the synthesis of a variety of proteins.66–70 We are investigating the use of thiazolidine or pseudoproline linkages, which are formed between an N-terminal cysteine and an aldehyde or an ester–aldehyde (Fig. 4).55, 56 For this approach, the dendritic polymer must contain three or more N-terminal cysteines and the PEG crosslinker must contain at least two terminal aldehyde groups, or vice versa. Specifically, we mixed aqueous solutions of a dendron containing N-terminal cysteines and a PEG-dialdehyde (PEG-DA) or PEG-diesteraldehyde (PEG-DEA) to afford a crosslinked network via formation of thiazolidine or pseudo proline linkages throughout the hydrogel, respectively.55, 56 As shown in Figure 4, the amine reacts with the aldehyde followed by thiazolidine formation, which is a reversible reaction. If an ester linkage is beta to the thiozolidine, then an O,N acyl migration occurs affording the pseudoproline—this step is irreversible. A photograph of one such hydrogel is shown in Figure 5.

Figure 4.

(Top) chemical ligation reactions yielding either a thiazolidine (I) or pseudoproline linkage (II). (Bottom) Dendritic and PEG based macromers for forming the hydrogels.

Figure 5.

Photograph of a hydrogel on top of a package of nonabsorbable suture.

Using these two crosslinking strategies, we have prepared a variety of hydrogels for characterization as well as for evaluation in specific applications. In the following sections, three successful applications of the dendritic macromolecules are described. The rationale for the selection of the specific macromer, the benefits of the chosen crosslinking approach, and the advantages of our approach over current methods are highlighted in each section. Specifically, we will describe the use of (1) peptide-based dendrons and chemical ligation crosslinking chemistry to repair corneal lacerations and to secure corneal transplants, (2) photocrosslinkable PGLSA-PEG based dendrimers for cartilage tissue engineering, and (3) Lys-PEG based dendrimers for creating localized hydrogel reaction chambers for molecular screening.


The repair of corneal wounds and the restoration of patient vision are of significant clinical importance. Corneal wounds arise from traumatic injury (e.g., perforations, lacerations), infections, and surgical procedures (e.g., transplants, incisions for cataract removal and intraocular lens implantation, laser-assisted in situ keratomileusis (LASIK)). Currently, nylon sutures are used to repair these wounds and depending on the extent of injury, multiple sutures may be required to secure the damaged tissue and restore the structural integrity of the cornea. It is estimated that globally more than 12 million procedures per year use nylon sutures to close ocular wounds. However, sutures are not ideal because the suture solely provides mechanical closure and does not actively participate in healing, in addition to the suturing procedure being inherently invasive.71–74 More specifically, sutures are suboptimal for this application because (1) the placement of the sutures inflicts additional trauma to corneal tissue, especially when multiple passes are needed; (2) sutures can act as a nidus for infection and incite corneal inflammation and vascularization increasing the incidence of corneal scarring; (3) corneal suturing often yields uneven healing, resulting in an astigmatism; (4) sutures are also prone to becoming loose and/or broken postoperatively and require additional attention for prompt removal; (5) sutures require removal by an ophthalmologist, often months after the operation creating a new opportunity for infection; and (6) suturing requires an acquired technical skill that can vary widely from surgeon to surgeon, thus influencing the overall success of the operation. Consequently, there is clinical interest in a sealant to replace or supplement sutures in the repair of corneal wounds.

There are precedents for the use of sealants and these alternative approaches have had a positive clinical impact. For example, cyanoacrylate glues were reported in the 1960s by Webster et al. for the repair of corneal perforations.75 More recently, fibrin adhesives have been explored for closing corneal wounds. However, both these glues have one or more of the following limitations including ease of application, preparation time, potential for viral transmittance, heat generation, toxic byproducts, abrasive materials, and limited effectiveness. A number of complications have been reported for cyanoacrylate glues including cataract formation, corneal infiltration, granulomatous keratitis, glaucoma, and retinal toxicity.76–84 Cyanoacrylate and fibrin glues are used “off-label” and at the discretion of the surgeon to repair the wound.

Design requirements for an idealized ocular adhesive generally fall into two main categories. The sealant must be capable of withstanding a variety of mechanical/optical constraints present in the ocular environment in addition to possessing favorable biological characteristics to prevent bacterial incursion and to promote native tissue ingrowth. The ideal material would chemically be capable of crosslinking on the moist ocular surface in a rapid and controlled manner, ideally setting in 30 s or less upon receiving the initiator signal. Additionally, a solution viscosity (<100 cP) allows for precise placement of the sealant by the technician. Upon gelation, the resultant hydrogel must provide significant closure to maintain both the structural integrity of the eye and be capable of withstanding high intraocular pressures (IOPs) (>80 mmHg). In addition, the sealant should possess elasticity greater than that of the corneal tissue to disfavor the formation of an astigmatism. The resultant hydrogel should also have a refractive index similar to that of the underlying tissue (1.42) and maintain diffusion properties to allow for gas and nutrient exchange (>2 × 10−7 cm2/s for small molecules). After successful closure for days to months depending on the extent of the wound, a characteristic that is tunable with a hydrogel, the sealant would then be either absorbed or exuded from the wound.

We have successfully used dendritic macromolecules as macromers to form hydrogel sealants via the photochemical or chemical ligation crosslinking chemistry to repair corneal lacerations and perforations,13, 16, 85, 86 seal cataract incisions,55 secure corneal transplants,56, 86 and close LASIK flaps.85, 87 To highlight the importance of crosslinking chemistry within one type of hydrogel sealant system, we will focus our discussion to full thickness corneal lacerations and corneal transplants. Corneal lacerations that are caused by trauma, infection, inflammation, or surgical procedures are an ophthalmic emergency that can lead to loss of vision. These wounds are repaired using sutures and as we have discussed earlier, suturing has significant drawbacks.

Corneal transplantation or penetrating keratoplasty (PKP) is one of the most common and successful tissue transplants.88 In a corneal transplantation, the recipient cornea undergoes a large circular full-thickness cutting to remove the damaged tissue, after which a previously cut donor corneal button is manually sutured to the recipient corneal rim. The standard of care today involves 16 running sutures to secure the new transplant tissue in place. The major disadvantages related to this procedure include delayed visual recovery, suprachoroidal hemorrhage, neovascularization, microbial keratitis, the need for postoperative suture removal (typically 9 months after transplantation), and surgically-induced astigmatism.89–92

Among the design parameters for these two indications (corneal lacerations and transplants), the lifetime of the hydrogel sealant is perhaps the one that necessitates the largest variance in overall requirements. For a corneal laceration, the sealant must remain in place for 2–4 days to allow for re-endothelization of the corneal wound site and closure of this relatively small wound (3–5 mm incision). On the other hand, a sealant for securing a full-thickness circular 8 mm corneal transplant must perform for months as the host tissue requires time to integrate with the tissue. To achieve this longevity differential, we chose to evaluate the dendron containing N-terminal cysteines ([G2]-(Lys)3-Cys4) and PEG-DA or PEG-DEA to afford a crosslinked network via formation of thiazolidine or pseudoproline linkages, respectively. Hydrogel weight loss, as a function of time at 25 °C when stored in a humidity chamber, is dramatically different for the two hydrogels. The hydrogel prepared from [G2]-(Lys)3-Cys4 and PEG-DA is intact for several days whereas the [G2]-(Lys)3-Cys4 PEG-DEA hydrogel is stable for more than 4 months.56

To determine whether a hydrogel sealant prepared from [G2]-(Lys)3-Cys4 and PEG-DA would secure a 4.1 mm full thickness corneal laceration, we performed a series of experiments. A 4.1-mm corneal laceration was made in several enucleated eyes. These wound were either left to self-seal, closed using one interrupted 10-0 nylon suture, or closed using the hydrogel sealant. For the hydrogel sealant, dendron ([G2]-(Lys)3-Cys4) and PEG-DA were mixed quickly at room temperature and then approximately 20 μL of the hydrogel sealant was applied to the wound. A hydrogel was formed upon mixing within 20–30 s as a result of the rapid formation of thiazolidine linkages. Figure 6 shows a sealed 4.1-mm corneal laceration repaired using the hydrogel sealant. Within 5 min of repairing the wound, regardless of the closure methodology utilized, saline was injected in the anterior chamber via a syringe pump until the repaired laceration leaked. In this ex vivo study, the mean leaking pressure for the hydrogel sealant, sutured, and untreated eyes (n = 3/sample) were 160, 75, and <10 mmHg, respectively. The values for the hydrogel sealant and sutures are above normal IOP of about 12 mmHg. The wound is not sealed using only the dendron or PEG-DA hydrogel precursors alone but requires the combination of both to operate effectively. Similar results were obtained in the treatment of 3-mm cataract incisions.55 The mean leaking pressure for the hydrogel sealant (n = 8) and suture (n = 2) treated eyes were 184 and 54 mmHg, respectively. Next, we evaluated if this hydrogel adhesive would prevent the influx of extraocular surface fluid into the wound.93 For these experiments, a cataract incision was made in several additional human enucleated eyes and then the wounds were either left to self-seal or treated with the sealant. India ink was applied to the ocular surface and the IOP was cyclically raised and lowered between 0 and 100 mmHg six times. Histological analysis showed that India ink entered the self-sealed wounds but not the sealant-treated corneas. During the cyclic raising and lowering of IOP, we used real-time optical coherence tomography to image the adhesive treated wound. Because of its elastic characteristics, the sealant did not dislodge but stretched to conform to the wound during the changes in IOP. No leakage was observed around the wound site. In regards to the overall efficacy, the hydrogel sealant secures the corneal wound, provides a water-tight seal, and withstands higher pressures and stresses placed on a wound than conventional suture treated wounds. The procedure with the hydrogel sealant is facile and requires less surgical time than conventional suturing, does not inflict additional tissue trauma, and does not require the use of a laser—unlike the photocrosslinkable corneal sealants—which reduces the need for additional instruments as well as eliminates the small but still present potential risk from laser eye damage.

Figure 6.

Leaking pressures for hydrogel sealant, sutured, and untreated 4.1 mm corneal lacerations. Photograph of a sealed wound (insert).

With this success, we next determined whether the hydrogel sealant prepared from dendron ([G2]-(Lys)3-Cys4) and PEG-DEA would secure the incision between the host and graft corneal tissue in a transplant. In this exvitro model, an 8-mm central corneal trephination was made in an enucleated eye and then this newly formed button was autografted back to the original eye. The host–graft tissue interface was secured using sutures, sutures combined with the hydrogel sealant, or the hydrogel sealant alone (Fig. 7). The leaking pressure for the autografted eyes was measured as we have done for the corneal laceration studies to determine the extent to which the wound was sealed.55 The leaking pressure for autografts receiving 16 interrupted 10-0 nylon sutures was 13 ± 5 mmHg (n = 4). When the hydrogel sealant was applied (33 wt %; 60 μL) to the sutured wound in addition to the 16 interrupted sutures, the leaking pressure increased to 63 ± 7 mmHg (n = 4). Increasing the macromer wt % to 50% (60 μL) with 16 interrupted sutures afforded a leaking pressure of 101 ± 5 mmHg (Fig. 7). We were unable to secure the autograft to a level above normal IOP when the hydrogel sealant was used alone indicating that this hydrogel does not possess sufficient adhesivity by itself to secure a PKP. However, an additional benefit to this hydrogel sealant, beyond closing the wound is the potential of the hydrogel barrier formed at the wound interface to prevent the flow of extraocular surface fluid and protect the wound from postoperative infections. The transport of India ink across the hydrogel can be monitored as we have done for the corneal wound study described earlier. When India ink is applied to the wound, the dye does not penetrate into the anterior chamber indicating that the wound interface is secured. The resulting crosslinked hydrogel sealants are transparent, elastic, hydrophilic, adhesive, and act as a physical protective barrier to the ocular surface.

Figure 7.

Leaking pressures for hydrogel sealant + suture, suture, and hydrogel sealant treated corneal autografts. Photograph of a sealed hydrogel sealant + 16-sutured corneal autograft (insert).


Osteoarthritis (OA) is a common form of arthritis that affects 100 million individuals in the world today. In the early stages of osteoarthritis, proteoglycans and collagenous proteins are lost from the cartilage tissue followed by the formation of small discrete lesions.94, 95 As the disease progresses, these lesions grow and eventually the subchondral bone is exposed.96–100 This degeneration of articular cartilage leads to a loss of mobility, severe and debilitating pain, and a reduction in the overall quality of life for the patients. Depending on the severity of the disease, the current clinical treatments include the chronic use of anti-inflammatory drugs, abrasion, mosaicoplasty, microfracture surgery, and chondrocyte transplantation.94, 101–106 The last resort for OA patients is total joint replacement, but this treatment is costly and traumatic. Yet, these approaches meet with varied successes due to the lack of a vascular and lymphatic system hindering the regenerative capacity of native cartilage. Cartilage tissue injuries never fully heal and only worsen with time.94, 107 Consequently, there is significant clinical interest in creating a therapy based on tissue-engineering principles to restore function to the damaged cartilage tissue site.

Typically, such strategies to repair cartilage involve a combination of a polymer-based scaffold, cells, and growth factors to create the required native cartilage.94, 108–111 The scaffold plays a key role in the repair of osteochondral defects, and it must meet a number of design criteria (1) produce a resorbable three-dimensional porous structure in vivo; (2) possess similar mechanical properties to the native tissue it is replacing; (3) support the infiltration, proliferation, and/or differentiation of the required local cell phenotype; (4) be biocompatible and nonimmunogenic in vivo; and (5) integrate with the surrounding matrix in the defect. The scaffold must ultimately guide the restoration of the tissue during healing.

Of the various scaffolds materials examined by many groups, those based on photocrosslinkable hydrogel scaffolds are showing considerable promise.59, 112–116 The in situ photocrosslinking ability of these systems is highly desirable in cartilage tissue-engineering application for a variety of reasons. First, it allows the uncrosslinked macromer solution to be mixed with cells or soluble factors, such as growth factors or cytokines, prior to defect site delivery. Second, the high water content of the scaffold allows for efficient diffusion of nutrients and oxygen into, and waste and carbon dioxide out of the hydrogel. Third, the uncrosslinked macromer solution can easily flow into irregularly shaped defects common to damaged or diseased cartilage, facilitating integration with the surrounding native tissue. Fourth, the liquid state of the macromer solution allows access to surgically inaccessible trauma sites via endoscope-assisted (micro)surgery. Lastly, these materials, once crosslinked in situ, provide immediate adhesion and mechanical integrity to the defect site at the time of implantation.

Given our interest in dendritic macromers and hydrogels, we evaluated the photocrosslinkable derivatives of the PGLSA-polyethylene glycol dendritic-linear copolymers (PGLSA[BOND]OH)2-PEG as scaffolds for cartilage tissue engineering.65 In addition to satisfying the requirements above, these dendritic macromers allow increased crosslink density of the scaffold without significantly increasing the polymer concentration when compared with linear polymer analogs. This approach leads to improved mechanical properties and minimal swelling of the hydrogel scaffold, while maintaining (bio)degradable sites such as ester linkages throughout the structure. Specifically, we modified the ([G1]-PGLSA[BOND]OH)2-PEG polymer to contain peripheral terminal MA groups ([G1]-PGLSA-MA)2-PEG (Fig. 8). Once this macromer is prepared, it can be dissolved in an aqueous solution containing the visible photoinitiating system (i.e., eosin Y, 1-vinyl-pyrrilidinone, and triethanol amine) and a hydrogel is formed upon photolysis with an argon ion laser at 514 nm. This eosin Y based photocrosslinking process is mild and has the following benefits: the vivid pink color of eosin Y in the hydrogel can be easily observed when placed in the defect site facilitating efficient filling, and the dye is bleached during the crosslinking reaction, confirming reaction completion, and uniformity of the reaction.

Figure 8.

Chemical structure of the ([G1]-PGLSA-MA)2-PEG macromer used to form the hydrogel scaffolds for chondrocyte entrapment.

Cylindrical hydrogel samples of known polymer concentration and dimensions were prepared and then used for the swelling, degradation, and mechanical testing in vitro. Hydrogels of 7.5, 10, and 15 wt % polymer showed minimal change in shape, gaining in weight only 10% over 30 days in phosphate buffered saline (PBS) at RT. This is in contrast to linear PEG dimethacrylates that can swell in excess of 100%. The equilibrium compressive modulus E was dependent on polymer wt %, as expected, with E increasing significantly from about 3 kPa at the lowest macromer concentration to 600 kPa at the highest macromer concentration (see Fig. 9). The complex shear modulus |G*| of the hydrogels showed limited concentration dependence, increasing from about 1–40 kPa, over the concentration range.

Figure 9.

Mechanical properties of the hydrogel scaffold prepared from ([G1]-PGLSA-MA)2-PEG and ([G1]-PGLBA-MA)2-PEG compared to articular cartilage.

Chondrocyte-hydrogel constructs at two different concentrations (7.5 and 15 wt %) were then prepared with freshly isolated porcine chondrocytes, placed in individual wells, and cultured in chondrocyte culture medium in a humidified atmosphere at 37 °C with 5% CO2. The chondrocyte-hydrogel constructs were harvested and processed for histology at 4 and 12 weeks (n = 3). The paraffin-embedded sections were stained with H and E, Safranin-O (marker for proteoglycans), or Masson's Trichrome (marker for collagen) for histological evaluation. Sections were also immunostained for the presence of Types I and II collagen. As shown in Figure 10, the encapsulated chondrocytes showed no signs of dedifferentiation and retained their rounded morphology. After 2 and 4 weeks of culture, Safranin-O and Masson's Trichrome staining indicated that chondrocytes encapsulated in the hydrogels at the lower macromer concentration accumulated significant amounts of extracellular matrix rich in proteoglycans and collagen, respectively, (Fig. 10). In contrast, cells encapsulated in hydrogels at the higher macromer concentration produced extracellular matrix only in the immediate area around each cell. Sections of cell-hydrogel constructs prepared from the 7.5 wt % concentration stained strongly for Type II collagen demonstrating the accumulation of extracellular matrix with molecular components present as found in native articular cartilage. No significant staining for Type I collagen was observed indicative of fibrocartilage. However, the cell-hydrogel constructs at 7.5 wt % were degrading over time and, by 4 weeks, some samples had disintegrating into several smaller fragments. This degradation behavior was not observed for the cell-hydrogel constructs formed at 15 wt %, even after extended culture time (12 weeks). Importantly, the 7.5 wt % hydrogel scaffolds were supportive of cartilaginous extracellular matrix synthesis. However, these hydrogel scaffolds possessed limited mechanical integrity.

Figure 10.

Histological sections of 7.5 and 15% macromer concentration hydrogels after 2 and 4 weeks incubation. (Left) Red indicates proteoglycans in the Safranin-O stained sections, (middle) green indicates collagen in the Masson's Trichrome stained sections and (right) red indicates Type II collagen in the immunostained sections, no significant Type I collagen was detected at either concentration. The length of the inserted bar is 100 μm.

To slow the degradation rate of the hydrogel scaffold but still retain the favorable characteristics of the 7.5 wt % hydrogel scaffold in terms of matrix accumulation, we prepared a new macromer, which contained ester as well as carbamate linkages.117 We prepared a first generation dendritic macromolecule composed of glycerol, succinic acid, β-alanine, and PEG using a divergent method as shown in Scheme 3. As before, photolysis of an aqueous solution containing the methacrylated poly(glycerol beta-alanine)-PEG macromolecule (([G1]-PGLBA-MA)2-PEG) and the eosin Y photoinitiating system afforded a crosslinked hydrogel scaffold. The hydrogel scaffolds at 5, 10, and 20% exhibited no significant swelling similar to what was seen earlier with the ([G1]-PGLSA-MA)2-PEG based hydrogel scaffolds. Next, the mechanical properties were measured over a concentration range of 5–20% w/v. The mechanical properties showed high concentration dependence with the higher polymer concentrations affording stiffer materials. Specifically, the mechanical properties of the hydrogels ranged from about 50–900 kPa for the compressive modulus, and 2–80 kPa for the complex shear modulus. With respect to native articular cartilage, as shown in Figure 9, the (([G1]-PGLBA-MA)2-PEG) based hydrogel approaches the mechanical properties of articular cartilage.

Scheme 3.

Synthesis of the ([G1]-PGLBA-MA)2-PEG macromer.

Next, we evaluated the integrity of the hydrogels in a rabbit knee confined defect model under dynamic mechanical testing. Briefly, a simulated osteochondral defect (3 mm in diameter × 10 mm in depth) was drilled in the center of excised medial femoral condyles of New Zealand white rabbits and filled with 50 μL of the macromer and then photo-crosslinked with an argon ion laser for 120 s. The cleaned femur and tibia were then mounted to the load frame of a custom designed computer controlled, servomotor-actuator system for simulating rabbit knee kinematics. The hydrogel scaffold was subjected to dynamic mechanical loading (300 cycles) with a physiologically relevant load (30 N at the end of the tibia simulating the body weight of a 3 kg rabbit). Upon completion of the loading regimen and under visual inspection, the hydrogel scaffold at 5, 10, and 20% w/v remained intact in the defect site. The integrity of the hydrogel and the hydrogel–bone interface was further assessed by magnetic resonance imaging. As shown in Figure 11, the 10 wt % hydrogel was still present and integrated with the surrounding tissue after the dynamic mechanical testing as were all the other weight percent samples.117

Figure 11.

(Left) Photograph of a biodendrimer scaffold filled osteochondral defect. (Right) Sagittal plane MR image of a biodendrimer scaffold filled osteochondral defect.

Finally, we conducted an initial in vivo experiment to evaluate the hydrogel performance in a full-thickness osteochondral defect. We selected the carbamate–ester–ether (([G1]-PGLBA-MA)2-PEG) hydrogel scaffold based on its low swelling and high E and G properties. In addition, the hydrogel scaffold prepared from this macromer will likely have a longer performance lifetime than the poly(ester)-based biodendrimers.65 The aqueous solution containing the crosslinkable biodendrimer and the photoiniating system was injected into a preformed defect located in the right knee of adult New Zealand white rabbits (n = 3). Next the solution was photocrosslinked using an argon ion laser for several minutes or until the pink color was gone confirming the crosslinking reaction was completed. A control untreated group was used in a similar defect located in the left knee of the same rabbits. At 6 months, the rabbits were sacrificed and histology was performed to determine cellularity (H and E stain), collagen (Masson trichrome), and proteoglycans (Safranin O stain) content. All three stains revealed that in the hydrogel scaffold treated defects, the hydrogel was well integrated with the surrounding tissue with strong staining for collagen and proteoglycans (Fig. 12). Importantly, the healing response in the hydrogel-filled knees exhibited morphological and biochemical characteristics consistent with normal–hyaline–tissue, whereas the unfilled controls appeared to be filled irregularly and stained less for collagen and proteoglycan content. The resulting crosslinked hydrogel scaffolds are integrated with the surrounding tissue, mechanically resilient, and promote extracellular matrix production in the wound site. These dendritic macromers possessed a number of favorable properties when used to prepare scaffolds for the repair of cartilage defects.

Figure 12.

Histological section of unfilled (top) and filled (bottom) osteochondral defects. a) Masson's trichrome and b) Safranin O staining. Scale bar = 100 μm.


Our increasing ability to access and analyze genomic and proteomic information through microarrays has afforded substantial scientific and medical advances. These advances range from greater understanding of fundamental biological processes to the evaluation of new drug targets for once untreatable diseases. The acquisition of such biological information relies heavily on high throughput and high-density screening of molecular–molecular and molecular–cellular interactions.118–127 Microarray technology is one screening method which has received considerable attention given that data can be obtained in a spatially arrayed, high-density format. However, many of the formats in use rely on covalent attachment chemistry of one component to a solid support.123, 128–148 This approach affects molecular and macromolecular properties, requires prior modification of the substrate, limits the diversity of assays, and creates unwanted molecular interactions at the surface—all of which can influence assay outcome and results. Eliminating the need for covalent attachment to the support prior to screening enables investigation of a greater number and type of molecular interactions while minimizing the biases as a result of the screening approach undertaken.

Consequently, to address these limitations and develop alternative high throughput screening approaches, we have prepared and evaluated crosslinked immobilized hydrogels as general reaction chambers for screening (bio)molecular and (bio)macromolecular interactions.149 As described earlier, dendritic macromolecules possess a number of favorable properties as macromers for hydrogel formation. For this application, we need to form small, micron-sized hydrogel reaction chambers on an aldehyde coated glass surface and thus selected (Lys)2-PEG in combination with (CHO)2-PEG as the hydrogel precursors, after a preliminary study of several potential candidates. The (Lys)2-PEG reacts with the terminal aldehydes of the (CHO)2-PEG as well as the surface immobilized aldehydes to afford Schiff-base linkages and a highly crosslinked hydrogel network adhered to the glass surface. Using an OmniGrid Accent™ microarraying robot equipped with a Stealth Printhead containing Stealth Micro Spotting Pins, we dispensed 1 nL volumes of the hydrogel precursors in a solution containing the (bio)molecule/(bio)macromolecule of interest on the aldehyde modified glass slides (Fig. 13). After printing, the slides were washed with 1% w/w bovine serum albumin (BSA) in PBS (pH = 7.4) to block the remaining surface aldehydes in the nonspecific intermediate regions. Importantly, this can all be done under mild conditions (aqueous solution; pH = 7.4; RT) without the need of prior derivatization of the printed (bio)molecule/(bio)macromolecule (Fig. 13). This screening technique is amenable to high throughput analyses as ≈50,000 torroid-like hydrogel chambers can be printed on a single 18 × 72 mm2 glass slide. A photograph of a reaction chamber is shown in Figure 14.

Figure 13.

Construction of IgG immobilized hydrogels on aldehyde coated slides using (CHO)2-PEG and (Lys-NH2)2-PEG.

Figure 14.

(Left-Top) optical and fluorescent (left-bottom) image of two hydrogel reaction chambers containing aRNA (250–5000 nt) (top row) and controls (without aRNA) (bottom row) after probing with Cy5-aRNA (60–200 nt). (Right) Example of a screening experiment performed with an array of reaction chambers.

To evaluate the capability of these hydrogel reaction chambers for screening small molecule–protein, protein–protein, and nucleic acid–nucleic acid molecular recognition, we performed a series of well-known model reactions. For small molecule–protein interactions, we prepared hydrogel reaction chambers containing biotin and then probed the chambers with Cy5 labeled streptavidin. The Cy5-streptavidin was successfully incorporated into the hydrogel chambers and bound the biotin. A twofold increase in red fluorescence relative to control chamber lacking biotin was observed. Next, we investigated a protein–protein interaction by printing goat IgG in the hydrogel chambers and then probing with Cy5 labeled protein G. A greater than a two-fold increase in red fluorescence was observed for the hydrogel chambers containing IgG relative to control hydrogel chambers without IgG indicating the formation of the IgG-protein G complex. A control experiment with BSA loaded chambers showed no red fluorescence when probed with the Cy5-Protein G, and, likewise, a reaction chamber loaded with IgG when probed with Cy5-streptavidin showed no red fluorescence. All together, these data confirm that specific protein–protein recognition within the scaffold is occurring and that the fluorescence signals are not merely a result of nonspecific physical entrapment of the protein in the hydrogel chamber during the assay.

The hydrogel reaction chambers are not limited to studying protein–protein interactions, as we were able to obtain similar results with nucleic acid–nucleic acid recognition. As shown in Figure 14, hydrogel chambers containing fragmented antisense RNA (aRNA) when probed with complementary Cy5 labeled RNA showed fluorescence intensities approximately eight-fold greater than controls (unloaded chambers). Probing the RNA loaded chambers with noncomplementary Cy5 labeled aRNA confirmed that nucleic acid complementarity was required and that noncomplementary Cy5 labeled aRNA was not trapped within the hydrogel chambers. We extended this work to assessing small DNA strands—a common screening platform. A 20-mer DNA (5′-TGAGTCTTCTAAGCTCTCCG-3′) was printed in the hydrogel and probed with its Cy5 labeled compliment (5′-Cy5-CGGAGAGCTTAGAAGACTCA-3′). After hybridization, a five-fold increase in red fluorescence was observed for hydrogel chambers containing the duplex DNA. Probing the 20-mer with noncomplementary Cy5-DNA afforded no increase in fluorescence indicating that hybridization had not occurred.

This facile and robotic screening platform using hydrogel reaction chambers comprised of dendritic macromers offers several advantages over conventional screening methods and formats. These benefits include (1) each hydrogel chamber acts as a site-isolated chamber for a specific reaction; (2) the printing and formation of the chamber occurs simultaneously with the loading of the (bio)molecule or (bio)macromolecule of interest; (3) a single platform for all molecular recognition processes from small molecules to proteins and nucleic acids; (4) the monitoring of the molecular recognition events can be achieved in an unbiased facile manner without modification or chemical attachment of the entities prior to use; and (5) the hydrogel chambers are amenable to the preparation of large arrays for high throughput screening.


In summary, dendritic macromers are versatile macromolecules for the preparation of hydrogels which are of interest and utility for a variety of applications. Herein we described three successful applications using these macromers and, importantly, each requiring a different set of design requirements—be it a corneal sealant, a scaffold for cartilage tissue engineering, or a reaction chamber for screening molecular recognition events. An underlying theme to this research is the synthesis and use of dendritic macromolecules that are biodegradable and biocompatible. For our interests, the syntheses, whether divergent or convergent, require selection of a monomer that is known to be biocompatible or degradable in vivo to natural metabolites and high reaction yields to attain material for subsequent evaluation. Dendritic macromolecules are favored in our laboratory over linear polymers because of the high level of molecular control that can be achieved during synthesis affording unique, well-defined macromolecules. This result has two significant consequences. First it enables a specific physical property, mechanical property, or biological response to be correlated to a well-defined macromolecular composition. Second, it facilitates the designing and prototyping of a macromolecule for a specific application.

With regards to the resulting crosslinked hydrogels formed from these macromers, there are a number of important points to learn. First, we can use two different types of crosslinking chemistries to prepare the hydrogels. The photochemical route allows facile “on demand” crosslinking by application of light and is adaptable to endoscope-assisted microsurgery. Prior to crosslinking, the aqueous solution of the macromer can be applied to the tissue site including those sites that are difficult to reach or are of irregular size and shape—like many trauma sites. The electrophile–nucleophile based crosslinking strategies, which begin to crosslink upon mixing and then set upon placement on the tissue, do not require additional instrumentation (such as a laser) for use but do require careful timing and placement on the surface. This is true for both the Schiff-base and chemical ligation chemistry. The chemical ligation strategy is beneficial since the reaction is performed at neutral pH, occurs quickly, produces no by-products, and is chemoselective. We have investigated both the formation of thiozolidine and pseudoproline linkages for creating these hydrogels. These two reactions demonstrate the concept of using chemical reversibility as a means to control hydrogel performance lifetimes. The formation of the thiozolidine is a reversible reaction and thus the hydrogels prepared using this linkage have limited stability in water of about a week whereas those hydrogels prepared with pseudoproline linkages, which involves an irreversible reaction, remain stable for months. This difference can be tailored for a specific application as we have shown for sealing corneal lacerations or securing corneal transplants.

Using either hydrogel formation strategy, we can vary the physical and mechanical properties of the resulting hydrogels. For example, by varying the weight percent or the dendritic structure, the mechanical properties of the hydrogels can be tuned. We have prepared hydrogels with a compressive modulus ranging from approximately 10–900 kPa. Likewise, degradation can be modulated by selecting different linkage chemistries within the dendrimer structure (e.g., ester vs. amide vs. thiozolidine vs. carbamate). The highly branched structure of the dendritic macromolecule, which possesses a multitude of crosslinkable groups, allows for efficient crosslinking and formation of hydrogels with low swelling characteristics. This is advantageous as excessive swelling can lead to dislodgement of the hydrogel from the site and/or negate the tissue sealing effect by increasing the distance between adjacent structures. The hydrogels can be formed on tissue surfaces and synthetic surfaces such as glass. In fact, many individual hydrogels can be prepared on a surface to create arrays for high throughput screening. Small molecules, proteins, nucleic acids, and cells can be entrapped within the hydrogels and once entrapped do not lose their function. Moreover, we have prepared hydrogels using a wide range of polymer wt % such that we can form hydrogels that possess from 40 to 93% water by weight. In general, we find hydrogels possessing such high water weight percents to be biocompatible and more suitable for working with biologics (e.g., tissues, cells, proteins).

These well-defined dendritic macromolecules offer a wealth of opportunities to control structure and tune properties. Our studies have enabled a basic understanding of the relationships between composition, structure, and properties as well as what design requirements are required for a specific application. It is a chemist's toolbox. We can alter composition, crosslinking chemistry, internal bonds, wt %, adhesivity, generation number and all of these effects afford diverse macroscopic results. Continued investigation and development of these dendritic macromolecules as well as other biocompatible compositions and unique architectures will increase our basic understandings and provide new solutions to chemical, biological, and medical challenges in the coming decades.


This work was supported in part by the NIH, PEW Foundation, and BU. The author thanks his collaborators Terry Kim (Duke Eye Center), Brian Snyder (Children's Hospital/Harvard Medical School), Lori Setton (Duke University), Scott Schaus (Boston University) and their fellows and students who worked on these projects. The author also thanks his graduate students and postdoctoral fellows for their hard work and dedication to these projects: Prashant N. Bansal, Jason Berlin, Michael A. Carnahan, Lovorka Degoricija, Neel Joshi, Nathanael R. Luman, Steven R, Meyers, Merredith Morgan, Abigal Oelker, Kimberly A. Smeds, Serge H. M. Söntjens, and Michel Wathier.

Biographical Information

original image

Mark W. Grinstaff

Mark W. Grinstaff is an Associate Professor of Biomedical Engineering and Chemistry at Boston University. Mark received his PhD from the University of Illinois under the mentorship of Professor Kenneth S. Suslick and was an NIH postdoctoral fellow at the California Institute of Technology with Professor Harry B. Gray. Mark's awards include the ACS Nobel Laureate Signature Award, NSF Career Award, Pew Scholar in the Biomedical Sciences, Camille Dreyfus Teacher-Scholar, and an Alfred P. Sloan Research Fellowship. He has published more than 90 peer-reviewed manuscripts and given more than 170 oral presentations. He is a cofounder of two companies that are commercializing his ideas. His current research activities involve the synthesis of new macromolecules and amphiphiles, self-assembly chemistry, tissue engineering, drug delivery, and nanotechnology.