PEGylation of NIR Cd0.3Pb0.7S aqueous quantum dots for stabilization and reduction of nonspecific binding to cells

Cd0.3Pb0.7S (CdPbS) aqueous quantum dots (AQDs) made with 3-mercaptoproprionic acid (MPA) as a ligand have the advantages of emitting near-infrared light, well above 800 nm, that completely circumvents interference from tissue autofluorescence and have significant amounts of ligands for bioconjugation. However, retaining the right amount of MPA became a challenge when using CdPbS AQDs for bioimaging because retaining too much MPA could lead to significant nonspecific staining in cell imaging while insufficient MPA could cause AQDs instability in biological systems. Here we examined PEGylation (i.e. chemically linking amine-functionalized polyethylene glycol (PEG)) to modify MPA on the AQDs surface to improve AQDs stability and reduce nonspecific staining. In addition, for conjugation with antibodies, a bifunctional PEG with a carboxyl functionality was used to permit chemical linkage of a PEG to an antibody on the other end. It was found that performing PEGylation at the thiol concentration where the zeta potential becomes saturated stabilized the CdPbS AQDs suspension and reduced nonspecific binding to cells. Furthermore, with the bifunctional PEG, the CdPbS AQDs were conjugated with antibodies and the AQD-Ab conjugates were shown to stain cancer cells specifically against normal cells with a signal-to-noise ratio of 8.


Introduction
Bioimaging traditionally uses conventional fluorescent dyes such as fluorescein isothiocyanate (FITC) and the Alexa Fluor family that have low photostability [1,2].FITC had a reduction in fluorescence over 20% in just 80 seconds of exposure under a fluorescent microscope, while Alexa Fluor 568 fluorescence decreased about 15% under the same conditions [3].Quantum dots (QDs), fluorescent inorganic semiconductor nanocrystals, are far superior to traditional dyes in photostability and brightness.QDs can also be excited by any wavelength of light less than their band-gap wavelength; a simpler requirement compared to dyes, which have a narrow excitation spectrum [4,5].There are increasing applications of QDs for bioimaging recently [6][7][8][9][10].
Previously, we have developed an aqueous synthesis method where cadmium lead sulfide (Cd 0.3 Pb 0.7 S (CdPbS thereafter)) aqueous QDs (AQDs) could be synthesized using 3-mercaptoproprionic acid (MPA) as a ligand in ambient conditions in water [11] without the need of ligand and solvent exchange as required by the traditional organic solvent-based QD synthesis methods.Once made, the surface of CdPbS AQDs were already functionalized with the ligand that could potentially be used for functionalization in an aqueous environment.More importantly, the CdPbS AQDs emit near-infrared (NIR) fluorescence well above 800 nm with a peak at 820 nm.This makes them ideal for bioimaging because its NIR emission is completely separated from any potential interference from tissue autofluorescence in the 280-700 nm range [12,13].As such, an 800 nm long-pass filter can be used to view or image the CdPbS AQD staining even under room light because only AQD fluorescence will be captured through the filter for analysis.It should be noted that even though heavy metal ions Cd 2+ and Pb 2+ are toxic to biological system, the CdPbS AQD system is intended for ex vivo imaging on excised specimens.
However, when using the CdPbS AQDs for bioimaging it was found that there was significant nonspecific binding to the cell surface including MDA-MB-231, and HT29 cell lines.In addition, the AQDs are less stable at neutral pH with a reduction in photoluminescence (PL) as pH reduced from 12 to 7.This was likely due to the short, carboxyl ligand, MPA.Because MPA did not bind strongly to the AQDs, it was necessary to include excess MPA in the suspension to maintain the AQDs stability [14][15][16].However, MPA on the AQDs could lead to nonspecific binding as the negative charge of MPA could bond to a positive charge of surface proteins.Also, the free MPA in the suspension during conjugation could compete with the MPA on the AQDs thereby reducing the chance of AQDs to link to Ab and increasing the nonspecific binding.
Previously Bentzen et al [17] showed that by coating amphiphilic poly(acrylic acid) ((AMP)-capped) CdSe/ZnS with polyethylene glycol (PEG) (MW 2000) the nonspecific binding to cells was reduced.In the literature, there were several other studies using PEG to functionalize QDs [18][19][20][21].Here we investigate the use of PEG to enhance aqueous CdPbS QD suspension stability and reduce nonspecific binding to cells.PEG is a hydrophilic and flexible ligand [22], and could lead to better suspension stability due to the hydrophilicity of the PEG and the larger separation distance between particles due to the added length of PEG.Nonspecific binding of AQDs to cell surfaces is primarily due to electrostatic and hydrophobic interactions between the AQD and the cell surfaces and AQD self-aggregation [23,24].PEG is hydrophilic and is expected to make the AQD surface less negatively charged by reacting with the carboxyl groups of the MPA.The monofunctional PEG used in the present study (MW 3000) had an amine group to react with the carboxyl functionality of surface capping molecules through carbodiimide chemistry [25].Because AQDs were made with excess MPA, the PEGylation of AQDs is different from PEGylation of QDs made by the organic route, which does not have excess free capping molecules in the solution.As a result, it was necessary to work out the condition of PEGylation that is optimized at the MPA concentration that maximizes the surface zeta potential of the AQDs.The equilibration of surface MPA and free MPA in the solution can be described by Langmuir isotherm [26][27][28] that will be helpful in guiding our experimental approach.Previously, Zhang et al [29] used homo-bifunctional PEG linkers for PEGylation of CdSe/ZnSe/ZnS QDs (emission ∼600 nm) synthesized in organic solvent.In contrast, here we used hetero-bifunctional PEGylation of NIR AQDs.We also study the specific staining of cells by conjugation with antibodies, in addition to reduction of nonspecific staining of cells.
In this study, which is based on chapter 2 of the author's thesis [30], it was shown that PEGylation led to reduced nonspecific binding in cell staining.Furthermore, using bifunctional PEG (MW 700) with a carboxyl group on the other end to conjugate with antibodies created AQD-Ab conjugates.These AQD-Ab conjugates were shown to specifically stain cancer cells against normal cells with a signal-to-noise ratio (SNR) of 8.

CdPbS synthesis
CdPbS with a nominal molar ratio 8:5:1 of [MPA]:[Cd&Pb]:[S] with [S] being 0.6 mM was prepared.First, 21 µl of MPA (Sigma-Aldrich, St. Louis, MO) was added to 47 ml of deionized (DI) water at room temperature and stirred for 5 min.Then, 300 µl of 0.08 M Cd precursor solution (1.23 g of cadmium nitrate tetrahydrate (Sigma-Aldrich) dissolved in 50 ml DI water) was added and stirred for 10 min before adding 700 µl of 0.08 M Pb precursor solution (1.32 g of lead (II) nitrate (Sigma-Aldrich) dissolved in 50 ml DI water) and stirring for 10 more minutes.The pH was adjusted to 11 by adding 510 µl of tetramethylammonium hydroxide (TMAH) (Sigma-Aldrich).This solution was stirred for 10 min before the AQDs were formed upon the addition of 375 µl of 0.08 M S precursor solution (0.96 g of sodium sulfide nonahydrate (Sigma-Aldrich) in 50 ml DI water).This suspension was stirred for 10 more minutes.The molar ratio of [MPA]: [Cd&Pb]: [S] was 8:2.6:1 at this stage.Additional 875 µl of Cd precursor was then added to fulfill the 8:5:1 [MPA]: [Cd&Pb]: [S] ratio, followed by adjusting the pH to 12 with an additional 325 µl of TMAH.The final suspension of CdPbS AQDs, at 0.6 µM concentration by particle, was stirred for 10 more minutes before storing the AQDs at 4 • C in the dark.They were used for PEGylation within 2 d of synthesis.

Excess MPA removal by centrifugation with 10k filter
To optimally react amine-PEG to the MPA on the AQD surface, the excess MPA in the suspension were removed by centrifugation at 6k relative centrifugal force for 12 min with a 10 kDa filter (MilliporeSigma, Burlington, MA) followed by refilling the retentate with DI water to restore the volume.The number of rounds of filtration determined the thiol concentration.

MPA concentration quantification
The MPA concentration was quantified by Ellman's reagent thiol assay.In the following, one may view the thiol concentration as a proxy for MPA concentration as each MPA possesses one thiol.A 20× Ellman's stock solution was first prepared in 5 ml DI water with 4.0 mg Ellman's reagent (Thermo Fisher Scientific, Waltham, MA) and 20.5 mg sodium acetate (Thermo Fisher Scientific) and a 1× Ellman working solution was prepared by diluting the 20× stock solution to 1× with 0.1 M Tris HCl (Thermo Fisher Scientific) [31].
The thiol concentration of the AQDs suspension was then measured by diluting the AQDs 100× in 1× Ellman's reagent working solution and incubating for 5 min.The absorbance of the solution was then measured at 412 nm using a UV-VIS spectrometer (Ocean Optics, Orlando, FL).The thiol concentration of the AQD suspension in question was then determined according to a standard curve created using the thiol of cysteamine (Thermo Fisher Scientific) as a model.Zeta-potential and size measured by ZetaSizer ZS90 (Malvern, Surrey, UK).PL was measured using the Photon Technology International Spectrometer (Birmingham, NJ) with excitation 468 nm.

PEGylation
PEGylation is the process of reacting the MPA on the AQDs with the PEG.Because our AQDs are capped with MPA (carboxyl), we used an aminefunctionalized PEG (MW 3000).Using carbodiimide chemistry [7,25], in which the amine will bind to the carboxyl, allowing the hydroxyl end of the PEG to interact with the outside environment.A summary of the PEGylation process using amine-PEG is shown in figure S1.The concentration of 1-(3-Dimethylaminopropyl)-3ethylcarbodiimide (EDC) (Thermo Fisher Scientific), sulfo-N-hydroxysuccinimide (sNHS) (Thermo Fisher Scientific), and PEG are important factors in the binding of amine-PEG (Sigma-Aldrich) to the AQD surface.
EDC and sNHS stock solutions were prepared.The 20 mM EDC or sNHS stock solutions were prepared in 1× phosphate buffered saline (PBS) (Thermo Fisher Scientific) immediately before use.The 2× concentrated AQDs, concentrated by 10 kDa filtering, were added to a centrifuge tube so their final concentration will be 0.6 µM by AQD.The 4 mM final concentration of EDC and sNHS were added and incubated for 15 min.This EDC/sNHS concentration was chosen because it had a minimal effect on the PL of the AQDs and did not cause aggregation in the AQDs (figure S2).The pH was raised to 6.8-7 with 10-20 µl, dependent on thiol concentration, of 20× borate buffer (Thermo Fisher Scientific) for a 2 ml total sample volume.Next, amine-PEG was added from a 30 mM stock solution in 1× PBS.The remaining volume was filled with 1× PBS.This mixture was incubated in ambient conditions.The PL stability was tracked over 6 d.After determining the optimal thiol concentration using 2 mM amine-PEG, corresponding to 3300 nominal PEG per AQD (PEG/AQD), the optimal amine-PEG concentration was studied by adding PEG ranging from 1 to 3 mM, corresponding to 1700, 3300, and 5000 nominal PEG/AQD.Samples were filtered to remove excess PEG and EDC/sNHS reagents using a 10 kDa filter.The process with bifunctional PEG (NH 2 -PEG12-proprionic acid) (Sigma Aldrich) was the same as amine-PEG but required optimization of PEG/AQD.This PEGylation process is described in figure S3.The bifunctional PEG had one end of carboxyl and the other end of amine with a MW 700.

Fluorescamine (FA) assay
One tool to determine the binding of the amine-PEG to carboxyl (of MPA) is a FA assay.FA is a molecule that, upon binding to amine groups, becomes fluorescent with excitation at 390 nm and emission at 470 nm.Increased fluorescence is correlated to an increased amine concentration.A standard curve was created using glycine (Thermo Fisher Scientific) from 0 to 5 mM.The FA (Thermo Fisher Scientific) stock was prepared as 0.3 mg ml −1 of acetone.The samples to be measured were diluted 10× with PBS and 25% of the final volume was the FA stock (7.5 µl sample + 67.5 µl PBS + 25 µl FA stock) [5].Samples were read in a 96-well plate using a BioTek Synergy plate reader (Winooski, VA) with excitation 390 nm and emission 470 nm.

Gel electrophoresis
Agarose gel was used to compare sizes of the various PEGylated AQDs.A 1.25% agarose gel was prepared by heating the 1.25% low melting point agarose (Thermo Fisher Scientific) in 1× Tris-borate-EDTA buffer (TBE) (Thermo Fisher Scientific) to 60 • C until agarose was fully dissolved and poured into a horizontal Mini-gel system (Fisher Scientific, Pittsburgh, PA).The gel was left to set for 1 h.Samples were added to wells and covered with agarose to prevent samples from leaking out of the wells.The gel was run in 1× TBE buffer for 15 min at 120 V.The UVP ChemStudio gel imaging system (Analytik Jena GmbH, Jena, Germany) was used for imaging of the AQDs fluorescence using blue light excitation and 707-752 nm emission filter.

PEGylated AQD conjugation with antibody
A schematic of the conjugation with antibody is shown in figure S4.The 1700 bifunctional PEG/AQD PEGylated AQDs were first purified by filtering 3× at 7.5k rpm for 4 min each (Sorvall Biofuge Primo, Marshall Scientific, Hampton, NH) in a 10 kDa filter, replacing with DI water after each filtering.The AQDs were concentrated to 2× (1.2 µM) during this process.A final concentration of 0.3 µM EDC and sNHS were added to a 0.4 µM final concentration AQDs at pH 6.8 and incubated for 30 min.Then, 80 nM anti-Tn antigen antibody (GeneTex, Irvine, Ca) was added and the rest of the volume added as 1× PBS and incubated for 2 h at room temperature.The conjugate was filtered 3× at 5k rpm in a 300 kDa filter (Sartorius, Göttingen, Germany) and volume lost was replaced with PBS.Final conjugate was syringe filtered in 0.22 µM filter (Sartorius).
Cells were cultured on an 8-chamber slide (MilliporeSigma) overnight.They were washed twice with PBS for 2 min each, fixed in 4% paraformaldehyde in PBS (Thermo Fisher Scientific) for 15 min at room temperature, and washed twice in PBS for 2 min each.The cells were blocked with 1% bovine serum albumin (Fisher Scientific) in PBS for 1 h at room temperature and washed 3× in PBST (PBS with 0.1% Tween-20 (Fisher Scientific, Waltham, MA)) for 5 min each before staining with AQD samples or Ab-AQD conjugates for 1 h.Finally, the cells were washed 3× with PBST for 5 min each and counterstained with 4 ′ , 6-diamidino-2phenylindole (DAPI) (Thermo Fisher Scientific) and cover-slipped for examination under a fluorescent microscope (Olympus BX51, Tokyo, Japan).Fluorescent images were analyzed using a MATLAB program, which was written to outline individual cells or groups of cells and return the intensity.The cells included in the analysis are outlined in figure S5.

Optimal MPA concentration for PEGylation
The purpose of PEGylation was for the PEG to react with MPA on the AQD surface, but not the free MPA in the suspension.The amount of MPA adsorbed on the AQD surface is equilibrated with the amount of MPA in the suspension.Qualitatively, we can describe the amount of MPA on the AQD surface as a function of MPA in the suspension by the Langmuir isotherm [26][27][28].As the concentration of MPA (represented by the thiol concentration) in suspension increases, more MPA will be adsorbed on the AQD surface.There will be an onset MPA concentration above which MPA coverage on the surface is saturated.We postulated that the onset MPA on AQDs is favorable for optimal PEGylation because below the onset concentration, there was not enough MPA on the AQD surface to stabilize the AQDs whereas above the onset, the PEG reaction with the free MPA in the suspension may be favored over the surface MPA, which could lead to instability.To identify the onset MPA concentration, we measured the zeta potential of the CdPbS AQDs at various MPA concentrations as determined by the thiol assay as each MPA contains thiol.In figure 1(a), the zeta potential of AQDs in PBS versus the MPA (thiol) concentration is shown as full squares.MPA has a carboxyl terminus, so we expect more MPA on the surface would give a more negative zeta potential [10].As can be seen, the zeta potential indeed increased in magnitude with an increase in MPA (thiol) concentration and saturated at around −30 mV at an MPA (thiol) concentration around 2.3 mM.The behavior is consistent with a Langmuir isotherm, if one considers the zeta potential as a proxy for MPA coverage.For charged nanoparticles like these CdPbS AQDs, the magnitude of the zeta potential was important for suspension stability.The knowledge that the zeta potential of the AQDs started to saturate at around 2.3 mM MPA was important to chart out the conditions for PEGylation.
To determine the optimal MPA concentration for PEGylation, 3330 amine-PEG per AQD (amine-PEG/AQD) were added to the AQDs at 1.8, 2.2, and 2.5 mM MPA (thiol concentration), around the onset thiol concentration of 2.3 mM for zeta potential saturation shown in figure 1(a) and in the presence of 4 mM EDC and sNHS.The PL, or light emission, of the PEGylated AQDs were plotted versus time in figure 1(b).While all PL decreased with time, overall, the PL of AQDs with PEGylation (dashed lines and full symbols) were higher and showed less decrease with time than those without PEGylation (solid lines and open symbols).Notably, among the PEGylated AQDs, the PL was most retained at 2.2 mM MPA at all time points, indicating that 2.2 mM MPA, which is around the onset of zeta potential saturation, was the optimal MPA concentration for PEGylation.Below this concentration, there was not enough MPA on the surface to stabilize the AQDs while around this concentration, the additional free MPA in the solution could react with amine-PEG and reduced the number of bound PEGs on the surface.In figure 1(a), we also plotted the zeta potential of the PEGylated AQDs versus the MPA (thiol) concentration when PEGylation was performed.As can be seen, the zeta potential of the PEGylated AQDs became significantly less negative as compared to AQDs without PEGylation.This is expected because the -OH end of the amine-PEG carried no charge while the negatively charged carboxyls of the MPA on the AQD surface were consumed to form a peptide bond with the amine end of the amine-PEG.Also shown were the zeta potential of AQDs with 4 mM of EDC and sNHS, but without PEG, at 1.7 mM and 2 mM thiol concentration.Clearly, with only EDC and sNHS, the zeta potential of the AQDs were not much different from without EDC and sNHS.Only in the presence of PEG with EDC and sNHS did the zeta potential became significantly less negative, supporting that PEG was chemically linked to the MPA on the AQD surface.
It is of interest to note that while the zeta potential of PEGylated AQDs became less negative than without PEGylation as shown in figure 1(a) (particularly at MPA = 2.2 mM and 2.5 mM), the PL of the PEGylated AQDs were brighter and decreased more slowly than without PEGylation.This indicates that the stability of the PEGylated AQDs was less due to electrostatic repulsion but more likely due to the extension of the capping molecules on the AQD surface, which prevents the aggregation of AQDs at neutral pH.

Optimal PEG/AQD ratio
Once we obtained the optimal thiol concentration of 2.2 mM, we varied the amount of amine-PEG to find the optimal ratio of PEG per AQD (PEG/AQD) for PEGylation where the molar concentration of AQDs is defined as that of the AQD particles (0.6 µM).In figure 2(a), we show the PL versus time of PEGylated AQDs with PEG/AQD = 0, 1700, 3300, or 5000 at a nominal thiol (MPA) concentration of 2.2 mM.While all PEGylated AQDs had a higher PL than without PEGylation (PEG/AQD = 0), the AQDs with PEG/AQD = 3300 appeared slightly more stable over 6 d as compared to those with PEG/AQD = 1700 and 5000.In figure 2 we show both the PL (2b) in black and the amount of bound amine-PEG (2c) in red versus the nominal PEG/AQD ratio used for PEGylation.The actual number of bound PEG (2c), as measured by FA, increased as the nominal PEG/AQD ratio increased.This indicates that the reaction between the amine-PEG and MPA occurred.The concentration of bound amine-PEG was calculated by subtracting the remaining amine concentration after PEGylation from the starting amine concentration.There was no significant increase in amine-PEG bound to AQD when increasing PEG/AQD ratio from 3300 to 5000.In fact, PEG/AQD = 3300 retained the PL the most over time (2b).This indicates that around 3300 PEG/AQD ratio was the optimal for the reaction in the system with 4 mM EDC and sNHS used.The FA measurement indicated that about 1.5 ± 0.3 mM out of the total MPA (which was around 2.2 ± 0.2 mM) was bound to PEG after the PEGylation process.While the FA measurement did not indicate whether the PEG is bound to the MPA on the AQD surface or to the free MPA, the assumption was that at least some portions of the bound PEG were on the surface.Using a diameter of 5 nm and assuming a lattice constant 0.59 nm and 1 MPA per cation, it was estimated that the AQDs could have up to about 0.6 mM MPA on their surface.As a result, we expect the zeta potential of the AQDs to change after PEGylation. Figure 2(d) shows the zeta potential of AQDs after PEGylation and filtering out the excess.All PEGylated AQDs became less negatively charged.The fact that the binding behavior of PEGylation (figure 2(c)) is similar to that of PL (figure 2(b)) and zeta potential measurements (figure 2(d)), supports the argument that PEGylation occurred on AQD surface rather than with free MPA.Furthermore, the fact that the 3300 and 5000 have similar degree of PEGylation shows that PEGylation saturated in the system at ∼3300 PEG/AQD and the PEGylation was with the particle surface and no further PEGylation is possible due to steric hindrance on the AQD or hydrolysis of EDC/sNHS and inability for more PEG to bind.The zeta potential of AQDs went from about −25 mV to about −12 to −10 mV after PEGylation.The carboxyl of MPA on the surface of non-PEGylated AQDs was more negatively charged at neutral pH, whereas, if PEGylated, the hydroxyl capping was mostly neutral at neutral pH [32].The inserts of figure 2(d) describe the PEGylation of the AQDs.The insert I corresponds to an MPA-capped AQD and was highly negatively charged due to carboxyl surface coverage.However, with the addition of amine-PEG, some of the MPA charge was covered by the neutral charge of the hydroxyl of amine-PEG (inserts II, III, and IV) resulting in less negative zeta potentials.

Nonspecific staining of cells with amine-PEG AQDs
Here we examine how nonspecific binding to the cell surface can be reduced by PEGylation at various PEG/AQD ratios.As mentioned earlier, nonspecific binding of AQDs to cell surfaces is primarily due to electrostatic and hydrophobic interactions between the AQD and cell surfaces [23].The CdPbS AQDs are hydrophilic when capped with MPA, but PEG is a very hydrophilic molecule.It has been demonstrated that the surface charge is less negative after PEGylation.So, we expect a reduction in both electrostatic and hydrophobic interactions with the cell surface after the AQD was successfully PEGylated.
Figure 3 shows the nonspecific staining results of PEGylated AQDs at various PEG/AQD including PEG/AQD = 0 (see figure 3(a)) on HT29 cells for comparison.Visibly, there was a reduction of staining fluorescent intensity in samples stained by PEGylated AQDs with any PEG/AQD ratio (see figures 3(b)-(d)) compared to staining with PEG/AQD = 0 (figure 3(a)).This indicates that PEG was indeed bound on the AQDs and reduced their interactions with the cell surface.The nonspecific staining intensity distributions by the AQDs with various PEG/AQD ratios are summarized in figure 3(e).Clearly, all PEGylated samples (dashed, dashed-dotted, and dotted lines) had a much lower average and standard deviation of the staining intensity.The 5000 amine-PEG/AQD sample (dotted line) showed the lowest average staining intensity, reducing the nonspecific binding by about 55% (after background subtraction) as compared to without PEG (solid line).The 1700 (dashed line) and 3300 (dashed-dotted line) ratios only reduced the nonspecific staining by about 40%.Clearly, for amine-PEG, a greater nominal PEG/AQD yielded better results for reducing nonspecific binding.In addition to offering the most reduction of nonspecific binding, the 5000 PEG/AQD ratio (dotted line) also had the tightest staining intensity distribution with the lowest standard deviation, which could be advantageous if the AQDs are eventually used for targeted imaging.

Nonspecific staining of cells with bifunctional PEGylation
While the amine-PEG showed the proof-of-concept of using PEG to stabilize AQDs and reduce their nonspecific binding, it does not have the functionality to conjugate AQDs to a biomolecule such as an antibody for specific fluorescent staining of cells.Therefore, bifunctional PEG was used for the PEGylation.The bifunctional PEG has a slightly different chemistry compared to the amine-PEG as there is an amine end and a carboxyl end.This also means that the bifunctional PEG may bind to itself.Therefore, the optimal concentration of the bifunctional PEG may not be the same as that of the amine-PEG previously utilized.
Like amine-PEG, all the bifunctional PEG concentrations showed an increase in AQD PL stability (figure 4(a)).The 1700 bifunctional PEG/AQD showed the highest PL retention at ∼450 000 overall, about 140% higher than unmodified AQDs over 6 d (at ∼190 000), with 3300 (at ∼410 000) and 5000 PEG (at ∼390 000) being 120% and 105% higher in PL, respectively.The size measurements in figure 4(b) show an increase in size with an increase in PEG/AQD ratio.Without PEG the AQD diameter was 5 ± 3 nm, but the diameter increased to 22 ± 7 nm, 54 ± 9 = 10 nm, and 73 ± 11 nm for PEG/AQD = 1700, 3300, and 5000, respectively.The PEG alone is about 5 nm.Therefore, a single layer of PEG coating on the AQD could increase the diameter by about 10 nm.The 1700 PEG amount is within the standard deviation of such an increase, supporting that there was a monolayer of PEG on the AQD surface when PEGylated with PEG/AQD = 1700.This is illustrated with insert (II) in figure 4(b) where bifunctional PEG has bound to the MPA, elongating it.As more PEG is added, the size also increases, indicating additional layers of bifunctional PEG binding to itself and creating 2 or 3 layers as shown in inserts (III) and (IV).Figure 4(e) shows size as measured by gel electrophoresis, which is consistent with the ZetaSizer measurements of increasing size with increasing PEG/AQD ratio.Figure 4(c) plots the PL after 6 d as a function of PEG/AQD ratio with the ratio of 1700 having the highest PL over this time, indicating this PEG/AQD ratio created the most stable AQD. Figure 4(d) shows the zeta potential of bifunctional PEG-functionalized AQDs versus PEG/AQD ratio.Again, the unmodified or PEG/AQD = 0 had a zeta potential of −25 ± 5 mV.All PEGylated AQDs had a slightly less negative zeta potential of about −18 ± 3 mV.This indicates bifunctional PEG binding to the surface of the AQD shielded some of the negative charge from the MPA carboxyl and made the zeta potential slightly less negative.The zeta potential results can also be explained by the inserts in figure 4(b).Inserts (II), (III), and (IV) illustrate that not all MPA may be covered with PEG, but instead, the PEG that does bind, whether one layer or several, extends beyond MPA.The PEG is what is measured for zeta potential and because there is less of it compared to unmodified AQDs (PEG/AQD = 0), the zeta potential is slightly less negative.The fact that the 1700 PEG/AQD ratio retained the most PL and had the smallest size indicates that it was the optimal PEG/AQD ratio to cover the AQD surface with a single layer of PEG.The smaller size would benefit conjugation strategies.On the other hand, further increasing the PEG/AQD ratio created a larger AQD, likely with several layers of PEG.The FA assay was not used for bifunctional PEG binding as the PEG can self-react, i.e. carboxyl of one PEG reacts to the amine of another PEG.Results of such measurements would be difficult to quantify binding of PEG to the AQD surface.
We performed TEM characterization of MPAcapped and bifunctional-PEGylated AQD with PEG/AQD = 1700 to learn more about the microscopic feature of the PEGylation.The results are shown in figure 5.It can be seen in figure 5 that PEGylation causes particles to be separated thus improving the stability of AQD.Meanwhile, the MPA-capped AQDs were aggregated.Similarly, the PL spectra of the MPA-capped AQD and that of the bifunctional PEGylated AQDs in PBS over 1 month are shown in figure S6 in the supplemental information.It can be seen that the bifunctional PEGylated AQDs are more stable thus resulting in less nonspecific staining due to aggregation.
While the PL and size measurements suggest that the 1700 PEG/AQD ratio could be optimal, the nonspecific binding results are imperative to the use of PEGylated AQDs in future bioimaging.Figure 6 shows a decrease in nonspecific staining intensity for all PEG/AQD ratios compared to unmodified PEG (figure 6(a)), like the staining results with amine-PEG.This agreed with the hypothesis that PEGylating the AQDs will reduce nonspecific binding.The 1700 PEG/AQD ratio (figure 6(b)) had the lowest nonspecific staining (at 64 ± 12), reducing it by about 40% compared to 0 PEG/AQD (at 110 ± 17), with the 3300 PEG/AQD (at 67 ± 14) having very similar staining to the 1700 PEG/AQD.The 5000 PEG/AQD ratio reduced the nonspecific binding by about 30% (at 80 ± 17).While the standard deviation of the 3300 PEG/AQD ratio was slightly less than that of the 1700 PEG/AQD ratio, the smaller size and higher brightness of AQDs with a 1700 PEG/AQD ratio may make these AQDs more suited for bioimaging applications.

Specific staining of cancer cells with bifunctional PEGylated AQD-Ab Conjugate
With the success of PEGylation in reducing the nonspecific staining of cells, we used the PEGylated AQDs for conjugation with antibody to examine whether specific staining of cancer cells could be achieved.The results of using the PEGylated AQDs to create Tnantigen specific conjugates to distinguish Tn-antigen [33][34][35][36] presenting MDA-MB-231 cancer cells, from MCF12A normal cells are shown in figure 7. The  The small size of 1700 PEG/AQD allowed multiple AQDs to be conjugated to an antibody without steric hindrance.With a nominal 1:5 Ab:AQD ratio, the staining was specific with a SNR of 8. SNR can be defined in many ways [37].Here we follow the method described by Gharia et al [38].Specifically, the signal is defined as the specific cancer cell staining on MDA-MB-231 cells and noise is considered using the nonspecific staining on MCF12A cells.The SNR is calculated by dividing the difference in staining intensity between the two cells lines divided by the standard deviation of the noise (nonspecific MCF12A staining).The SNR is acceptable though not as high as we would like to have likely because the standard deviation of the nonspecific staining was quite high.Further filtering of the conjugates may reduce some of the nonspecific staining on the MCF12A cells, thus reducing the standard deviation.This could reduce the overlap of the staining intensity distributions shown in figure 7(c), increasing the specificity of the conjugate.The optimization of the conjugation of the AQDs to antibodies to improve specific staining and minimize nonspecific staining and apply it to tissue slide staining will be examined in a future study.

Conclusions
The results of this study indicated that PEGylation was effective in stabilizing and reducing nonspecific binding of CdPbS AQDs to cells.All PEGylated AQDs showed an increase in PL stability over time compared to unmodified AQDs.Using amine-PEG created a hydroxyl-capped AQD after PEGylation.In this case, the higher PEG/AQD ratio the lower the nonspecific binding.Among the ratios used, the 5000 amine-PEG/AQD led to the least nonspecific binding with cells.On the other hand, the bifunctional PEG, which retained the carboxyl functionality on the AQD surface after PEGylation, exhibited a minimal nonspecific staining at a nominal 1700 PEG/AQD ratio.Comparing the amine-PEG and the bifunctional PEG, the bifunctional PEG is more efficient in reducing nonspecific binding to cells with a small particle size using less amount of PEG, allowing multiple (nominal 5) AQDs to conjugate to one antibody.The minimal nonspecific binding at the 1700 bifunctional PEG/AQD ratio was related to a single PEG layer on the AQD surface.For targeted staining, our study showed the potential for specific, direct staining of cancerous cells using a PEGstabilized, NIR-emitting AQD and cancer-specific antibody.We believe by further optimizing the antibody-AQDs conjugation protocol, it is possible for further improvement on this metric, which will be examined in a future study.This evidence of conjugation also paves way for future applications in bioimaging, targeted drug delivery, disease detection and diagnoses.

Figure 1 .
Figure 1.Effect of thiol (MPA) concentration and PEGylation on AQD stability by zeta potential and photoluminescent (PL) measurements: (a) Zeta potential of modified (EDC/sNHS and PEGylated) or unmodified (MPA-capped) CdPbS AQDs versus thiol concentrations in neutral pH, and (b) Photoluminescence (PL) versus time for PEGylated (full symbols and dashed lines) or unmodified (open symbols and solid lines) AQDs at various thiol concentrations.

Figure 2 .
Figure 2. Effect of amine-PEG per AQD on stability of AQD in PBS in terms of zeta potential and its binding efficiency to carboxyl groups of MPA: (a) photoluminescence (PL) stability versus time for various nominal amine-PEG/AQD ratios added to the AQDs with 2.2 mM thiol concentration, (b) PL of samples 6 d post-synthesis with varied PEG/AQD (c) the actual amount of bound amine-PEG/AQD measured by fluorescamine versus the nominal PEG/AQD added (d) zeta potential of AQD suspensions with increasing PEG/AQD.The inserts (I), (II), (III), and (IV) in (d) illustrate the elongation of PEG as consistent with the AQD size with an increasing PEG/AQD ratio.

Figure 4 .
Figure 4. Bifunctional PEG/AQD effect on (a) PL versus time over 6 d, (b) size by ZetaSizer, (c) PL 6 d post-synthesis, (d) zeta potential, and (e) size by agarose gel electrophoresis.The inserts (I), (II), (III), and (IV) in (b) illustrate the elongation of PEG as consistent with the AQD size with an increasing PEG/AQD ratio.

Figure 7 .
Figure 7. Specific staining of cancer cells using bifunctional PEGylated CdPbS AQD-Ab conjugates on (a) MDA-MB-231 breast cancer cells and (b) MCF12A normal breast cells with (c) distribution of staining intensities for each cell line with background subtracted and signal-to-noise ratio (SNR) is 8. DAPI staining for nuclei in blue, AQD in red.