Synthesis, Internalization and Visualization of N-(4-Carbomethoxy) Pyrrolidone Terminated PAMAM [G5:G3-TREN] Tecto(dendrimers) in Mammalian Cells †

Tecto(dendrimers) are well-defined, dendrimer cluster type covalent structures. In this article, we present the synthesis of such a PAMAM [G5:G3-(TREN)]-N-(4-carbomethoxy) pyrrolidone terminated tecto(dendrimer). This tecto(dendrimer) exhibits nontraditional intrinsic luminescence (NTIL; excitation 376 nm; emission 455 nm) that has been attributed to three fluorescent components characterized by different fluorescence lifetimes. Furthermore, it has been shown that this PAMAM [G5:G3-(TREN)]-N-(4-carbomethoxy) pyrrolidone terminated tecto(dendrimer) is able to form a polyplex with double stranded DNA, and is nontoxic for HeLa and HMEC-1 cells up to a concentration of 10 mg/mL, even though it accumulates in endosomal compartments as demonstrated by its unique NTIL emission properties. Many of the above features would portend the proposed use of this tecto(dendrimer) as an efficient transfection agent. Quite surprisingly, transfection activity could not be demonstrated in HeLa cells, and the possible reasons are discussed in the article.


Historical Overview
Over the past four decades, dendrimers and dendritic polymers have created international interest among a diverse range of disciplines and researchers. During the first decade (i.e., 1980s), after the polyamidoamine (PAMAM) dendrimer discovery [1], there arose some skepticism over theoretical claims that these structures could be synthesized as truly monodispersed products that amplified according to a mathematically predictable model at each generation. As such, there was

Dendrimer Properties of Value in Drug Delivery and Nanomedicine
One of the first synthesized and hence most studied dendrimer families is the poly(amidoamine) (PAMAM) dendrimers. It is now widely recognized that these Tomalia type PAMAM and other dendrimers possessing interiors derived from "symmetrical branch cells" manifest unique interior void spaces/cavities suitable for hosting guest molecules [9]. In contrast, Denkewalter poly(l-lysine) (PL) dendrimers, which contain "asymmetrical branch cells" do not exhibit interior void space for hosting guest molecules. As such, symmetrical branch cell dendrimers have been used extensively as "unimolecular micelles" for the encapsulation and delivery of a wide range of drugs and active pharmaceutical ingredients [10]. Furthermore, many poly(valent) dendrimer surface moieties (i.e., amino or carboxylic, etc. moieties) have been exploited to bind various molecules/drugs/particles of interest either by direct covalent conjugation or supramolecular exo-complexation. These features make PAMAM dendrimers excellent candidates as drug carriers/encapsulators, imaging agents, gene carriers [11][12][13][14] or as actual active drugs/pharmaceutical agents [15,16].
A critical requirement for using any dendritic structures in drug delivery systems is that they exhibit biocompatibility, do not adversely stimulate/suppress the immune system nor show unacceptable cytotoxic effects [17,18]. The most widely recognized source of dendrimer toxicity is generally associated with cationic surface charge and has been comprehensively demonstrated in the case of amine terminated PAMAM dendrimers. In order to enhance biocompatibility, various surface modifications have been applied [19], wherein surface glycosylation or PEGylation are most commonly used to improve these properties; however, neither of these options is perfect. For example, glycosylation may create both anti-inflammatory and proinflammatory properties [20,21]. On the other hand, PEGylation has been shown to undergo radical based oxidation to produce fragmentation products known to elicit immune responses [22]. As a consequence, there is currently an active quest for PEGylation alternatives in the drug delivery area. That withstanding, such a PEGylation alternative has recently been reported by Tomalia et al. [23]. This alternative involves facile conversion of PAMAM dendrimer surface amines to N-(4-carbomethoxy) pyrrolidone moieties which appear to manifest very promising biocompatibility features [24]. It was shown that such a modification dramatically inhibits activation of proinflammatory signals in human monocytes [17] as well as substantially reducing cytotoxicity [25,26]. By analogy to the term PEGylation, these transformations have been referred to as "pyrrolidonylations" [23] and have been proposed as an alternative protocol to PEGylation for cloaking either dendrimer carriers, proteins or drugs. The objective of this cloaking process is to render drugs/drug carriers stealthy to in vivo protein opsonization, thereby reducing complement activation properties, while enhancing drug circulation times and reducing cytotoxicity. Furthermore, it should be noted that dendrimer surface "pyrrolidonylation" substantially enhances so-called "non-traditional intrinsic luminescence"(NTIL) properties [23,27,28]. This fluorescence enhancement allows direct in vivo bioimaging of the dendrimer nanoparticles without any need for conjugating traditional dyes or external staining. These studies demonstrated that pyrrolidone modified PAMAM-dendrimers manifest dramatically enhanced emission intensities (i.e., >50 fold) compared to unmodified amine terminated PAMAM dendrimers [28,29]. Recent proposed mechanisms to account for the NTIL phenomenon have concluded that these inexplicable blue emissions are largely due to the aggregation/clustering and/or physico-chemical confinement of normally non-emissive, electron-rich, hetero-atomic, functionalized moieties, (HASLs) (i.e., heteroatomic subluminophores) [30]. As such, the architectural immobilization of certain HASLs within the interior of PAMAM dendrimers led to some of the first reported examples of NTIL in the literature [31,32]. A logical extension of these NTIL mechanism principles suggests that covalent aggregates of PAMAM dendrimers (i.e., megamers or core-shell tecto(dendrimers)) might be expected to produce prototypes exhibiting substantially enhanced NTIL properties. In order to test this premise, a highly congested tecto(dendrimer) prototype was conceived and synthesized for evaluation as an in vivo NTIL imaging agent in biological cells as well as for its potential as a drug delivery vector.

Core-Shell Tecto(dendrimers); Synthesis and Property Evaluation for Life Science Applications
Within the context of architectural types, core-shell tecto(dendrimers) are well-defined dendri-clusters/ assemblies and constitute a subset of poly(dendrimers) referred to as "megamers" [33][34][35]. The first examples of core-shell tecto(dendrimers) were reported by the Tomalia group in 1994 and 2002; however, they have received relatively little attention until the present. The first synthesis of so-called "shell saturated" core-shell tecto(dendrimers) was published in 2000 [36]. In this case, a limited amount of amine terminated PAMAM dendrimer (i.e., referred to as the core reagent) was combined with a large excess of carboxylic acid terminated PAMAM dendrimer, (i.e., referred to as the shell reagent) in the presence of LiCl. This combination yielded a charge neutralized core reagent dendrimer that was completely surrounded (i.e., surface saturated) by shell reagent dendrimer to produce a highly organized, core-shell type supramolecular structure. This core-shell supramolecular assembly was then covalently fixed by adding a carbodiimide coupling reagent to produce amide linkages between the contacting core and shell components. This general protocol produced an ordered subset of megamers referred to as shell saturated, core-shell tecto(dendrimers). This tecto(dendrimer) category was coined to designate the fact that essentially all available surface area on the core dendrimer reagent was systematically and totally occupied by dendrimer shell components. This was possible due to charge neutralization interactions between core and shell components that could be annealed to provide the most efficient parking order for the shell components before covalently fixing core-shell surface positions with amide forming carbodiimide reagents.
Subsequently, a second semiordered category of megamers referred to as partial shell filled, core-shell tecto(dendrimers) was reported in 2002 [37]. This class of core-shell tecto(dendrimers) is defined by less ordered, more random core-shell attachment sites. This category is synthesized by a direct reaction of reactive core-shell components to produce covalent linkages. More specifically, this protocol does not involve the formation of a charge neutralized, core-shell supramolecular assembly that may be annealed to establish ordered and complete saturation of the core surface prior to covalent core and shell conjugation. As such, it is differentiated from the "saturated shell" type tecto(dendrimer) category by directly forming random covalent linkages between the core and shell components until all accessible reactive core surface space is filled. Due to these random, unordered covalent binding events, the core reagent surface area is less efficiently filled by the shell reagent dendrimers. In fact, only 40-66% shell saturation is observed [37] relative to total shell saturation levels that may be mathematically predicted by the Mansfield-Tomalia-Rakesh equation [38]. As a consequence, well-defined domains of either nucleophilic/electrophilic nanoclefts or nanocusps are created due to the random, unordered conjugation of shell components to the core. In this present work, nucleophilic clefts (i.e., occupied by primary-NH 2 ) are expected to result by attaching electrophilic (i.e., ester terminated) dendrimer shell components to a nucleophilic (i.e., amine terminated) dendrimer core component, as described in Figure 1. A comprehensive review of these superstructured poly(amidoamine) dendrimer based nanoconstructs and their applications in nanomedicine has been recently reported by [39]. Subsequently, a second semiordered category of megamers referred to as partial shell filled, core-shell tecto(dendrimers) was reported in 2002 [37]. This class of core-shell tecto(dendrimers) is defined by less ordered, more random core-shell attachment sites. This category is synthesized by a direct reaction of reactive core-shell components to produce covalent linkages. More specifically, this protocol does not involve the formation of a charge neutralized, core-shell supramolecular assembly that may be annealed to establish ordered and complete saturation of the core surface prior to covalent core and shell conjugation. As such, it is differentiated from the "saturated shell" type tecto(dendrimer) category by directly forming random covalent linkages between the core and shell components until all accessible reactive core surface space is filled. Due to these random, unordered covalent binding events, the core reagent surface area is less efficiently filled by the shell reagent dendrimers. In fact, only 40-66% shell saturation is observed [37] relative to total shell saturation levels that may be mathematically predicted by the Mansfield-Tomalia-Rakesh equation [38]. As a consequence, well-defined domains of either nucleophilic/electrophilic nanoclefts or nanocusps are created due to the random, unordered conjugation of shell components to the core. In this present work, nucleophilic clefts (i.e., occupied by primary-NH2) are expected to result by attaching electrophilic (i.e., ester terminated) dendrimer shell components to a nucleophilic (i.e., amine terminated) dendrimer core component, as described in Figure 1. A comprehensive review of these superstructured poly(amidoamine) dendrimer based nanoconstructs and their applications in nanomedicine has been recently reported by [39]. Dendrimers exhibit numerous properties which distinguish them as excellent candidates for drug delivery agents: it is most notable that they are well-defined nanoparticles possessing modular, modifiable structures [40]. As such, they can be fine-tuned both with regard to interactions with potential cargo (complex formation, encapsulation, conjugation) and with target cells (nontoxicity, bio-orthogonality, internalization mechanism and kinetics, intracellular fate) [41][42][43][44]. Since unmodified amine-terminated PAMAM dendrimers are largely suboptimal for many direct biological applications due to their toxicity, there is growing attention focused on the engineering of their "critical nanoscale design parameters" (CNDPs) in order to enhance certain properties [25,[45][46][47]. As part of this quest, we turned our attention to tecto(dendrimers), a ternary conjugate of one Dendrimers exhibit numerous properties which distinguish them as excellent candidates for drug delivery agents: it is most notable that they are well-defined nanoparticles possessing modular, modifiable structures [40]. As such, they can be fine-tuned both with regard to interactions with potential cargo (complex formation, encapsulation, conjugation) and with target cells (nontoxicity, bio-orthogonality, internalization mechanism and kinetics, intracellular fate) [41][42][43][44]. Since unmodified amine-terminated PAMAM dendrimers are largely suboptimal for many direct biological applications due to their toxicity, there is growing attention focused on the engineering of their "critical nanoscale design parameters" (CNDPs) in order to enhance certain properties [25,[45][46][47]. As part of this quest, we turned our attention to tecto(dendrimers), a ternary conjugate of one large (core) dendrimer with a number of smaller ones at the periphery (shell) [36,37,48]. We decided to combine this approach with surface modification by pyrrolidone moieties, since they are known to confer biotolerance and enhance nontraditional intrinsic luminescence (NTIL) with the potential for direct bioimaging [24,26,29].

Synthesis
In this present work, a core-shell tecto(dendrimer) series bearing nucleophilic nanoclefts was prepared by allowing a limited amount of nucleophilic dendrimer core reagent (i.e.,-NH 2 ) to react with an excess of electrophilic functionalized dendrimer shell reagent (i.e.,-CO 2 Me) to produce an amide linked partial shell filled, core-shell tecto(dendrimer), bearing nucleophilic clefts (i.e., primary-NH 2 ) and electrophilic cusps (i.e.,-CO 2 Me), (Figure 1). In our earlier work, we observed 53-66% shell filling leading to the synthesis of Structure (A). This represents the attachment of 8-10 shell components (i.e., PAMAM G2.5 shell attachment to a PAMAM G5 core) out of the theoretical prediction of 15 shell components according to the Mansfield-Tomalia-Rakesh equation [38]. In this present work, we observed the attachment of 10 out of 15 shell components according to electrophoretic (polyacrylamide gel electrophoresis; PAGE) analyses, thus allowing us to propose stoichiometries and calculate estimated molecular weights for Structures (A)-(C) as described in Figure 1.
The new "partial shell filled" PAMAM [G5:G3-TREN]-N-(4-carbomethoxy) pyrrolidone terminated core-shell (tecto)dendrimers were synthesized by slightly modifying our earlier published procedures [36,37]. This minor modification involved using tris-N,N ,N"[2-(aminoethyl)amine] (TREN) versus ethylene diamine to advance Structure (A), (G2.5) ester terminated core-shell tecto(dendrimer) to the next generation level (i.e., G3-TREN, amine terminated),Structure (B). Since the tris-N,N ,N"[2-(aminoethyl)amine] (TREN) reagent contains a branch juncture, this modification produces an additional terminal group amplification by doubling the primary amine groups on the surface of Structure (B). More specifically, this modification involved the following three steps: Step ( . In our previous work, we observed shell saturation levels for the core-shell Structure (A) that varied between 53 and 66% (i.e., 8-10 shell tectons) versus a theoretical prediction of 15 shell tectons for a 100% saturation level as predicted by the Mansfield-Tomalia-Rakesh equation [36,38]. In this present work, Structure (A) appeared to possess~9-10 dendrimeric shell tectons as determined by comparison to monomeric PAMAM G6 (MWt = 58.0 kDa) and PAMAM G7 (MWt = 116.5 kDa), respectively, using PAGE analyses (Supplementary Material, Figure S3). Excess dendrimer shell reagent G2.5 was removed by ultrafiltration to yield Structure (A). Note, it is very important not to isolate Structure (A) in a neat form. This product must be handled as a solution or rapid oligomerization, including cross-linking, which may occur due to amidation reactions involving nanocleft amino moieties and nanocusp ester groups.
Step (3) Allowing Structure (B) to react with a slight excess (i.e., 10%) of dimethyl itaconate converted all accessible primary amines to N-(4-carbomethoxy) pyrrolidone moieties to yield the final PAMAM [core (G5):shell (G3)-(TREN)]-N-(4-carbomethoxy) pyrrolidone terminated tecto(dendrimer), Structure (C). This final modified product was obtained as a white solid after ultrafiltration. It differs from all previously reported core-shell tecto(dendrimers), wherein the branch cells involved in the formation of the final (G3) shell layer are derived from tris-[2-(aminoethyl)amine], (TREN). This product was characterized by FTIR, 1 H/C 13 -NMR and electrophoresis (PAGE). A comparison with standard protein markers, other authentic relevant core-shell tecto((dendrimers) and appropriate N-(4-carbomethoxy) pyrrolidone terminated PAMAM monomeric dendrimers corroborates the proposed Structure (C), which appears to be present as a mixture of monomeric, dimeric and trimeric species as determined by PAGE analyses (Supplementary Material, Figure S4). A comparison with standard protein markers revealed three discrete electrophoretic bands with estimated molecular weights of~115,~225 and~325 kDa, respectively. These values correspond to molecular weights expected for monomeric, dimeric and trimeric forms, respectively, for core-shell tecto(dendrimer), Structure (C).
A survey of the literature reveals that fewer than a handful of core-shell (tecto)dendrimers have been reported to date [37,49,50]. However, core-shell (tecto)dendrimers have been examined as efficient, multivalent nanoscale platforms for the delivery of pharmaceuticals [51,52]. Due to their dendrimeric multiplicities and larger aggregate dimensions, they possess substantially more surface groups and internal cavity space. This enhanced interior void space endows them with variable sizes and well-defined chemistry, which could enable the transfer of diverse or multiple drug combinations [50]. Quite surprisingly, it has been shown that certain PAMAM core-shell tecto(dendrimers) possess intrinsic antitumor activity, wherein they exhibit selective cytotoxicity to melanoma cells but do not affect healthy cells [53].
In summary, activity in this area is in its infancy, and there remains a need for the synthesis and development of new core-shell tecto(dendrimer) structures. It is hoped these tecto(dendrimer) structures may combine desirable features such as: high biocompatibility, higher multiplicity, polyvalent features (i.e., external binding/interior encapsulation sites) as well as enhanced intrinsic fluorescence properties (i.e., NTIL emissions) suitable for in vivo bioimaging in combination with various drug delivery applications [30]. In fact, PAMAM [G5:G3-(TREN)]-N-(4-carbomethoxy) pyrrolidone terminated tecto(dendrimer) (C) was able to form a polyplex with double stranded DNA and convert its initially strongly negative zeta potentials to positive values with only two times excess ratio (weight).
Molecules 2020, 25, x FOR PEER REVIEW 11 of 19 cells were observed after incubation with the PAMAM [G5:G3-(TREN)]-N-(4-carbomethoxy) pyrrolidone terminated tecto(dendrimer) (C)-plasmid DNA polyplex even after numerous attempts at various DNA concentrations and N/P ratios. However, as shown in Figure 6B, Structure (C): plasmid DNA polyplex was effectively internalized by the cells and was visualized thanks to its characteristic fluorescence.

Discussion
We characterized the basic physicochemical properties (i.e., relevant to our biological goals) of the novel pyrrolidone-modified tecto(dendrimers) and found them to be consistent with expectations and needs for further application development. Thanks to their special core-shell aggregate architecture, they display a combination of advantageous traits observed for both nonmodified (i.e., amine terminated) and pyrrolidone modified (i.e., pyrrolidonylated) monomeric PAMAM dendrimers. The core-shell structures possess accessible nanoclefts bearing free primary amino groups, which appear to confer cationic surface charge in neutral aqueous solution. However, this cationic property is significantly weaker than in the case of basic unmodified, amine terminated PAMAM. On the other hand, pyrrolidonylation (i.e., 4-CMP modification) of the core-shell tecto(dendrimer) structure confers strongly enhanced NTIL emission features compared to the pyrrolidone modified monomeric PAMAM dendrimers. More specifically, fluorescence lifetime analysis of core-shell tecto(dendrimer) structures confirmed that they manifested relevant NTIL fluorophore features that behaved analogously to previously characterized monomeric, N-(4carbomethoxy pyrrolidone) (4-CMP) terminated PAMAM dendrimers [29]. For example, two relatively strong but comparable fluorophore centers, (i.e., corresponding to fluorescence decay curve components 2 and 3 in this study) are distinctly noted in this analysis [28]. It has previously been demonstrated that the brightest fluorophore (component 3, in our tecto(dendrimers) contributing half of total fluorescence output) is derived from interactions of tertiary amino groups and imidic acid moieties along the terminal PAMAM branches and that its brightness and quantum yield are enhanced by branch stabilization [54][55][56]. In the case of tecto(dendrimers), this stabilization is probably conferred both by the 4-CMP modification as well as by the enhanced rigidity of the core-

Discussion
We characterized the basic physicochemical properties (i.e., relevant to our biological goals) of the novel pyrrolidone-modified tecto(dendrimers) and found them to be consistent with expectations and needs for further application development. Thanks to their special core-shell aggregate architecture, they display a combination of advantageous traits observed for both nonmodified (i.e., amine terminated) and pyrrolidone modified (i.e., pyrrolidonylated) monomeric PAMAM dendrimers. The core-shell structures possess accessible nanoclefts bearing free primary amino groups, which appear to confer cationic surface charge in neutral aqueous solution. However, this cationic property is significantly weaker than in the case of basic unmodified, amine terminated PAMAM. On the other hand, pyrrolidonylation (i.e., 4-CMP modification) of the core-shell tecto(dendrimer) structure confers strongly enhanced NTIL emission features compared to the pyrrolidone modified monomeric PAMAM dendrimers. More specifically, fluorescence lifetime analysis of core-shell tecto(dendrimer) structures confirmed that they manifested relevant NTIL fluorophore features that behaved analogously to previously characterized monomeric, N-(4-carbomethoxy pyrrolidone) (4-CMP) terminated PAMAM dendrimers [29]. For example, two relatively strong but comparable fluorophore centers, (i.e., corresponding to fluorescence decay curve components 2 and 3 in this study) are distinctly noted in this analysis [28]. It has previously been demonstrated that the brightest fluorophore (component 3, in our tecto(dendrimers) contributing half of total fluorescence output) is derived from interactions of tertiary amino groups and imidic acid moieties along the terminal PAMAM branches and that its brightness and quantum yield are enhanced by branch stabilization [54][55][56]. In the case of tecto(dendrimers), this stabilization is probably conferred both by the 4-CMP modification as well as by the enhanced rigidity of the core-shell tecto(dendrimer) architecture. This last feature is consistent with a recent NTIL phenomenon mechanism proposed by us [30], which invokes immobilization due to rigidity as a significant parameter for enhancing NTIL emission properties.
Since the pyrrolidone-modified core-shell Structure (C) appeared to be fulfilling certain predicted needs as a potential nanoscale vector for drug delivery in mammalian cells, we next performed a series of critical biochemical/cellular measurements to verify these desirable properties. Firstly, Structure (C) exhibited extremely low toxicity properties when exposed to relatively delicate cells (i.e., derived from vascular endothelium) even at concentrations significantly higher than normally used for practical applications. This finding was surprising and exceeded all expectations. This was especially true in view of its PAMAM chemical structure as well as its larger nanoscale dimensions. Earlier work verified that dendrimers bearing cationic primary amino groups, especially when presented on large, higher generation PAMAM dendrimers, were very deleterious to biocompatibility and generally manifested strong toxicity properties [57]. On the other hand, this result corroborates the basic ideas behind the structural design of PAMAM [G5:G3-(TREN)]-N-(4-carbomethoxy) pyrrolidone terminated tecto(dendrimer) (C). As expected, 4-CMP modification successfully masks the external (G3-linked) primary amine groups, which are directly involved in cell membrane/surface receptors interactions that lead to toxicity. On the other hand, internal nanocleft (G5-linked) primary amine groups appear to remain available as cationic sites for charge conferral and biomolecular cargo interactions without exhibiting any toxicity features. Furthermore, these observations may also confirm a broadly held tenet that pyrrolidone-modified molecules (including biomaterials) usually have unexpectedly good biocompatibility properties compared to their basal forms due to the bio-orthogonality of the pyrrolidone moiety [58].
Since nanoparticle based drug delivery applications require a complex sequence of biochemical interactions with the cargo as well as the cell surface to fulfill their role [59], we decided to perform certain mechanistic evaluations before embarking on a phenomenological experiment. As such, we examined the capacity of PAMAM [G5:G3-(TREN)]-N-(4-carbomethoxy) pyrrolidone terminated tecto(dendrimer) (C) for binding with DNA by observing changes in surface charge distribution around the macromolecule using zeta potential measurements. This allowed us to not only to confirm the efficacy of DNA polyplex formation, driven by the nanocleft based, cationic amino group charge of the tecto(dendrimer), but also to monitor its stoichiometry with DNA. Considering that plasmid DNA is a very large molecular structure, it is significant that the DNA: core-shell tecto(dendrimer) complexation (i.e., at a 1:1 weight ratio) produced a net negative phosphate charge from the DNA. It was determined that the saturation point for neutrality resided around 1:1.5 (DNA: dendrimer weight ratio).
Subsequently, we verified the capability of cells to take up the tecto(dendrimer) nanoparticle. Here, the nontraditional intrinsic luminescence (NTIL) properties of the tecto(dendrimer) are a crucial feature since they allow us to use direct confocal microscopy to confirm the subcellular localization of cell-bound molecules. While NTIL emission intensities for unmodified, amine terminated PAMAM dendrimers are too weak to allow efficient bioimaging against the background of autofluorescent cellular components, we show that PAMAM [G5:G3-(TREN)]-N-(4-carbomethoxy) pyrrolidone terminated tecto(dendrimer) (C) produces NTIL emissions that are bright enough to be effectively imaged within cellular structures without significant interference from background fluorescence. Thus, these core-shell tecto(dendrimer) structures join the ranks of very few available nanoparticles manifesting intrinsic fluorescence emissions strong enough for label-free bioimaging. Although inorganic quantum dots represent such an intrinsically fluorescent nanoparticle category, it is also well known that they exhibit very dangerous heavy metal toxicity properties. This is in sharp contrast to these core-shell tecto(dendrimers) which exhibit virtually no cytotoxicity features at the cellular level. Thus, the usual nanoparticle imaging conundrum (i.e., nontoxic nanoparticles require conjugated labeling to be fluorescent, while fluorescent particles need to be coated for biocompatibility) may be avoided. This now appears to be possible with a single, well-defined nanoparticle type such as PAMAM core-shell tecto(dendrimers). Even more importantly, we show that these core-shell tecto(dendrimer) particles are efficiently internalized by mammalian cells and end up concentrated in an endosomal compartment, pointing to endocytosis rather than pinocytosis as the major uptake mechanism. This highly efficient internalization property is quite surprising, considering its relatively large polymeric structure, thus giving high optimism for future applications in drug delivery systems.
Our final phenomenological experiment concerning DNA delivery, as verified by gene transfection assessment, produced a rather surprising result. More specifically, although we were able to observe very efficient internalization of the DNA-tecto(dendrimer) polyplex, we were unable to detect any significant level of DNA transfection activity based on the lack of reporter protein expression from the DNA plasmid. In the case of unmodified, (G5) amine terminated PAMAM-(NH 2 ) 128 dendrimer, which is strongly toxic to cells, some of the surviving cells were indeed transfected, confirming the literature data that these structures are generally suitable for gene transfection [60]. However, it appears that tecto(dendrimer) architecture changes the DNA binding properties in such a manner as to prevent the efficient delivery of the cargo plasmid to the nucleus, where it has to be transported and located for expression. This may be due to one of several reasons: either the interaction between tecto(dendrimer) and DNA is too strong for efficient release within the cell, the dendrimer-DNA complex is trapped within the lysosomal pathway and cannot be released to cytoplasm or the DNA-tecto(dendrimer) polyplex is too large to enter the nucleus, which are all prerequisite steps in the transfection process. It may be possible that the lack of flexibility of the pyrrolidone-decorated tecto(dendrimer) structure, may restrict the endosomal escape of the DNA, which is also crucial for effective transfection activity [61]. Although this assessment does not indicate the PAMAM [G5:G3-(TREN)]-N-(4-carbomethoxy) pyrrolidone terminated tecto(dendrimer) (C) functions in its present form as an effective DNA transfection vector, it does suggest many other possibilities for drug delivery as well as possible structural modifications for enhancing effective DNA transfection, which we intend to explore in the future. MeOH in a dropwise manner over 5 min. while stirring. This reaction mixture was sealed under N 2 and heated at 40 • C for 25 days. Periodic analysis by SEC revealed the production of the higher molecular weight core-shell tecto(dendrimer) product, which was essentially equivalent in size to a PAMAM (G7) dendrimer. This was further confirmed by PAGE analysis which indicated a molecular weight of~116 kDa (Supplementary Material, Figure S3). Formation of the desired ester terminated, core-shell (tecto) dendrimer was monitored by 13 C/ 1 H-NMR, SEC, PAGE and FTIR spectroscopy and found to be essentially complete after 25 days based on no further amide formation. This ester terminated core-shell tecto(dendrimer) product could be isolated by ultrafiltration using an Amicon 30 kDa cut-off, regenerated cellulose membrane and after collecting a total of 9 L permeate. Note, this product should either be immediately pacified by treatment with tris(2-hydroxymethyl)aminomethane as described in [37] or used directly for the synthesis of core-shell tecto(dendrimer), Structure (B) as described below. The crude reaction mixture of Structure (A), including remaining G2.5 ester terminated dendrimer shell reagent, was added to a stirred solution of excess tris-(2-aminoethyl) amine (TREN); (i.e., 100 equivalent excess of TREN relative to the amount of original "ester terminated dendrimer shell reagent" used) in 50 mL methanol. The amidation of all ester functions was monitored by FTIR and found to be complete after 3 days at 4 • C and 8 h at room temperature. The excess TREN was first extracted away with several (10 mL) toluene washings. Complete removal of TREN together with any remaining amine terminated G3, dendrimer shell reagent was accomplished by exhaustive ultrafiltration through an Amicon 30 KDa cut-off, regenerated cellulose membrane and after collecting at least 8 L permeate. This methanol soluble product could be isolated as a white solid after lyophilization and gave the 13 C-NMR as shown in Supplementary Material Figure S1  10 , Structure (B) reaction product was allowed to react with dimethyl itaconate (i.e.,~10% excess/amino group) in methanol for 12 h at room temperature to give the final PAMAM, [G5:G3-(TREN)-N-(4-carbomethoxy) pyrrolidone] 10 terminated, Structure (C). This methanol soluble product was obtained as a white solid after lyophilization and confirmed by 13 C-NMR and electrophoretic (PAGE) analyses (Supplementary Material, Figures S2 and S4).

Cell Culture
HeLa and HMEC-1 cell lines were purchased from American Type Culture Collection and maintained under standard conditions. HeLa cells were cultured in DMEM medium with high glucose (4.5 g/L) and 10% fetal bovine serum (FBS). HMEC-1 cells were cultured in MCDB131 medium supplemented with epidermal growth factor (10 ng/mL), hydrocortisone (1 µg/mL), glutamine (10 mM) and 10% FBS. Cells were subcultured three times per week.

Cytotoxicity Assay
Cells were seeded into 96-well black plates at a density of 2 × 10 4 cells per well and treated with increasing concentrations of tecto(dendrimer) for 24 h in respective fully supplemented culture media. Following the incubation and single wash with medium, resazurin was added to the culture medium to a final concentration of 10 µg/mL and plates were incubated at 37 • C in darkness to allow conversion of resazurin to resorufin. Resulting fluorescence of metabolized resazurin was measured after 30, 60 and 90 min at 530 nm excitation and 590 nm emission using a microplate reader. Cell viability was derived from the slope of resorufin fluorescence intensity increase in time and was presented as a percentage of untreated control.

Conclusions
In summary, we observed interesting and somewhat unexpected physical/biochemical properties for the pyrrolidone-modified PAMAM [G5:G3-(TREN)]-N-(4-carbomethoxy) pyrrolidone terminated tecto(dendrimers), such as Structure (C). This large, somewhat complex synthetic structure exhibits many similarities to biological protein-based complexes, especially its lack of toxicity despite the presence of unique nanocleft based primary amine moiety site. It is important to remember that this is an example of an architecture type which can and will be further developed, with easily modifiable facades as well as internal cavities that can be a platform for interaction with various binding partners. We demonstrate three crucial properties warranting further studies in drug delivery, namely these tecto(dendrimers) exhibit intense nontraditional intrinsic fluorescent (NTIL) emission properties suitable for in vivo applications, they complex DNA efficiently, they enter biological cells readily and they exhibit low cytotoxicity. The complexity and rigidity of the core-shell tecto (dendrimer) structure may be crucial for each of these characteristics and deserves further examination. Although the direct use of these core-shell tecto(dendrimer) structures in their current form failed as vectors for DNA transfection applications, the unique nontoxic cationic character manifested by these structures as well as their highly enhanced NTIL emission properties allows for their use as biocompatible, highly fluorescent, NTIL emissive nanoparticles for direct biological imaging.