SC80 Master of Applied Science MSc Biofabrication Final Thesis - … · Quentin Clément PEIFFER . P a g e 1 | 78 STATEMENT OF ORIGINAL AUTHORSHIP The work contained in this thesis

UU student #: 5582733

QUT student #: n9827102

Iris OTTO

Dr.Phong TRAN

Dr.Ir.Jos MALDA

D/Prof.Dietmar W. HUTMACHER

UMC – Utrecht University, Netherlands

IHBI - Queensland University of Technology, Australia

01/09/2015 – 31/08/2017

SC80 Master of Applied Science

MSc Biofabrication

Final Thesis

“Biofabrication: tools for new therapeutics in

regenerative medicine and drug delivery.” Submitted in fulfilment of the requirement for the degree of SC80 Master of Applied Science

Science and Engineering Faculty

Quentin Clément PEIFFER

P a g e 1 | 78

STATEMENT OF ORIGINAL AUTHORSHIP

The work contained in this thesis undertaken between QUT and Utrecht University has not

been previously submitted to meet requirements for an award at these or any other higher

education institution. To the best of my knowledge and belief, the thesis contains no material

previously published or written by another person except where due reference is made.

QUT Verified Signature

P a g e 2 | 78

FOREWORD

This Master thesis is the result of the collaboration between the Utrecht University (UU) and

the Queensland University of Technology (QUT). As such, this document is divided in two

independent section: the first part “Biofabrication of an auricular cartilage” is presenting the

work carried out at UU while the second part “ Microporous polycaprolactone scaffolds for drug

delivery” is presenting the work carried out at QUT. As such, each part features its own ab-

stract, keywords, abbreviation, acknowledgement, table of content and bibliography.

Biofabrication of an auricular cartilage………………………………p3 - 43

Microporous polycaprolactone scaffolds for drug delivery…………p44-78

P a g e 3 | 78

Biofabrication of an auricular

cartilage implant

Quentin Clément PEIFFER

UU student #: 5582733

MSc Biofabrication

Minor Research Project

RMC Utrecht

09/11/2015 – 10/06/2016

Daily supervisor: Dr.Phong Tran

Examiner: Dr.Ir.J.Malda

P a g e 4 | 78

LAYMAN’S SUMMARY

Facial malformations like ear loss due to cancer, burns, trauma or even birth defects can heav-

ily affect the relationship between an individual and their relatives or society. Current treat-

ments present severe drawbacks with highly variable aesthetic results. With the capacity to

combine different materials in a precise manner, 3D printers appeared in biomedical sciences

during the past few years as novel tools able to bring new solutions, such as scaffolds, that

can overcome all previous treatments. A scaffold in tissue engineering is a construct made of

a material compatible with the human body to repair damaged tissues. Yet, to provide these

new clinical solutions, 3D printers require considerable work for research and optimization. In

this work, the deposition of two materials is studied to combine them in an ear shape cartilage.

The first material is a thermoplastic that provides mechanical strength and consequently, the

printed scaffolds is not destroyed after grafting. The second material is a hydrogel; a hydrogel

is a gel able to absorb a high quantity of water and therefore, provides conditions for cells to

proliferate. To combine these two materials, it is first necessary to optimize the deposition of

each of them individually. The first part of this work is to study the potential of a new tool for

the deposition of the thermoplastic material. Since the printing process can kill the cells, the

second step of this work is to review the literature to predict how to deposit the hydrogel while

preserving the highest cell survival. The thermoplastic and hydrogel laden with cells were com-

bined in square constructs and analyzed in the third step of this work. The final phase of this

work focuses on the computer work related to the control of the printer, to assess which soft-

ware would be the most useful for carrying out the printing of two materials in an ear shape.

ABSTRACT

Microtia or ear loss are facial malformations for which no current treatments are perfectly

adapted. Additive manufacturing is a growing field and is expected to provide medical applica-

tions in the near future, especially by the creation of intricate scaffolds. This study explores the

co-manufacturing of hybrid PCL/gelMA scaffolds, specifically for ear cartilage engineering.

This research with a step-by-step approach aims to present the different challenges that

represent co-manufacturing and how they could be overcome. This includes a description of

Fused-deposition modeling (FDM) and Pressure-Assisted Bioprinting (PAB) with attention

given to the preservation of cell-viability. If the combination of FDM and PAB is not a technical

challenge, this work illustrates the importance of characterizing materials rheological proper-

ties to have control over the fabrication process. Therefore, after experimenting and backed

by literature, it appears that spraying cell with low inlet-pressure is the approach that preserves

the highest cell viability when dispensing a cell-laden hydrogel. At last, this work points out the

importance of considering the computer science behind additive manufacturing, and which

otherwise can rapidly become a limitation for tools capacities

P a g e 5 | 78

ABBREVIATIONS 3D: Three Dimensional

CAD: Computer-Aided Design

DMEM: Dulbecco's Modified Eagle Medium

ECM: Extracellular Matrix

EXC: Experimental Cartridge

FDA: Food and Drug Administration

GelMA: Gelatin methacryloyl

HMI: Human Machine Interface

PBS: Phosphate-buffered saline

PCL: Polycaprolactone

RGD: Arginylglycylaspartic acid

STL: Standard Tessellation Language

ACKNOWLDEGMENT Jos Malda

Pedro da Costa

Iris Otto

Kim van Dorenmalen

Riccardo Levato

Maarten Blokzjil

Sarah-Sophia Carter

Madeline Hintz

Noël Dautzenberg

KEYWORDS

PCL; Auricular cartilage, GelMA, Dual printing ; 3D printing ; Biofabrication;

P a g e 6 | 78

TABLE OF CONTENTS

ABSTRACT ............................................................................................................................................ 4

ABBREVIATIONS ................................................................................................................................. 5

ACKNOWLDEGMENT ......................................................................................................................... 5

KEYWORDS .......................................................................................................................................... 5

1. INTRODUCTION .............................................................................................................................. 7

1.1 Materials .............................................................................................................. 7

1.2 The 3D printing approach ..................................................................................... 8

2. CHAPTER 1: PCL PRINTING ...................................................................................................... 11

2.1 Introduction .........................................................................................................11

2.2 Material and methods..........................................................................................11

2.3 Results ................................................................................................................13

2.4 Discussion ..........................................................................................................17

2.5 Conclusion ..........................................................................................................18

3. Chapter 2: Spraying vs Deposition .............................................................................................. 18

3.1 Introduction .........................................................................................................18

3.2 Conclusion ..........................................................................................................21

4. Chapter 3: Cell viability and printing ............................................................................................ 21

4.1 Introduction .........................................................................................................21

4.2 Material and method ...........................................................................................22

4.4 Troubleshooting ..................................................................................................29

4.5 Discussion ..........................................................................................................30

4.6 Conclusion and Further Experiments ..................................................................33

5. Chapter 4: Auricular shape and biofabrication .......................................................................... 33

5.1 Introduction .........................................................................................................33

5.2 Material and method ...........................................................................................34

5.3 Results and troubleshooting ................................................................................35

5.4 Discussion and Conclusion .................................................................................36

6. Chapter 5: General conclusion and prospective work .............................................................. 36

REFERENCES .................................................................................................................................... 40

ANNEX 1 .............................................................................................................................................. 41

ANNEX 2 .............................................................................................................................................. 42

ANNEX 3 .............................................................................................................................................. 43

P a g e 7 | 78

1. INTRODUCTION

Facial malformations hinder the relationship between a patient and society, resulting in a social

and psychological burden, that when surgically treated can highly improve psychosocial as-

pects and consequently the quality of life of the patient. Auricular malformations as a result of

congenital anomalies (microtia), cancer, burns or even trauma, are part of these facial malfor-

mations. Current treatments options for auricular malformation include ear prostheses, syn-

thetic implants and auricular reconstruction using skin flaps or autologous rib cartilage. Be-

cause of the complex three-dimensional (3D) shape of the auricle, auricular reconstruction with

autologous costal cartilage is a challenging procedure with a highly variable aesthetic outcome;

not to mention the significant operating time and donor site morbidity. In response, efforts have

been made towards creating pre-fabricated synthetic auricular implants (Medpor®). Medpor®

appears to be a good solution, even though as a foreign body it can potentially lead to implant

exposure or infection risks. However, the great majority of plastic surgeons prefers the use of

autologous cartilage frameworks, that is the current gold-standard over synthetic implants.

The convergence of regenerative medicine and biofabrication brings new alternatives that

would overcome limitations associated with current treatments such as donor site morbidity

while improving aesthetic and functional outcomes. It allows the possibility to engineer func-

tional cartilage using patient-derived or donor cells, and to create custom-designed cell-laden

implants with intricate architectures and complex shapes.

Biofabrication is particularly interesting since it offers the opportunity to combine different cell

types and materials to produce the ideal auricular scaffold. In addition, it may someday reach

a higher level of complexity, by incorporating fatty tissue or perichondrium. To be successful

the ideal auricular engineered scaffold should:

- Be strong enough to withstand the contractive forces of the skin and durably maintain

the same shape than the contralateral auricle.

- Incorporate autologous chondrocytes or stem cells that will recreate a matrix with the

natural elastic bending properties of the auricle.

- Be slowly degradable while new cartilaginous matrix replaces it, maintaining its original

shape.

Therefore, the project was based on I. Otto’s work [1] and the knowledge from literature.

1.1 Materials

1.1.1 Soft hydrogel: GelMA

P a g e 8 | 78

Chondrocytes typically thrive in a soft hydrogel that allows unimpeded nutrient diffusion and

provides a homogenous microenvironment promoting cellular migration, proliferation, and dif-

ferentiation, and to this purpose, GelMA was used, a water-soluble protein that mimics the

Extracellular Matrix (ECM). Biodegradable, biocompatible and able to form hydrogels, many

features make GelMA a good candidate in biomedical science and these are documented in

the literature. Its functionalization with unsaturated methacryloyl, combined with a photoinitiator

and under exposition of UV-light enable the GelMA to form covalently cross-linked hydrogels,

that can be cultured with encapsulated cells.

1.1.2 Stiff polymer: PCL

To withstand the contractive forces of the skin and durably maintain shape, the scaffold will

need to have a high degree of stiffness. A soft hydrogel such as GelMA, even if cross-linked,

cannot by itself reach the degree of stiffness required. Therefore, a stiff polymer, in our case

polycaprolactone (PCL), will be deposited with GelMA to enhance the mechanical properties

of thereof. PCL is a biodegradable polymer, non-toxic, with a broad miscibility, and a mechan-

ical compatibility with many polymers. Furthermore, it provides adhesion to a broad spectrum

of substrates, can be modified to create microporous fibers, or be grafted with the cell adhesion

site (such as RGD). PCL is approved by the FDA for specific biomedical applications and

widely used in research, especially in 3D printing where its mechanical properties and low-

melting-point make it an ideal printable polymer.

These materials are very common in the regenerative medicine landscape, especially in

biofabrication where their features are highly appreciated. The next section below provides a

more technical description of the project.

1.2 The 3D printing approach

1.2.1 Biofabrication approach

A cell-laden GelMA hydrogel solution is deposited between strips of PCL (Fig. 1) with robotic

dispensing technology using layer-by-layer deposition according to a computer-aided design.

This leads to the creation of custom hybrid constructs that combines the cell compatibility of

GelMA and the stiffness of PCL. This approach is already described in the literature and has

been proved viable [2], the thermal requirement for printing PCL was shown to be compatible

P a g e 9 | 78

with cell viability in the hydrogel [3]. Furthermore, the real potential of this approach lies in its

versatility, by combining different cells, materials or architecture all kind of applications are

possible [4].

1.2.2 Printers

Several tools are available in the laboratory, depending on the system the biofabrication pro-

cedure will be different slightly different, more details can be found in Chapter 4. In the labor-

atory, two robotic dispensing system is available, the Bioscaffolder and the 3D Discovery.

First the Bioscaffolder (SYS+ENG) (Fig 2 A): The Bioscaffolder is a 3-axis dispensing system

with an automatic tool change function, controlled by a Human Machine Interface (HMI) called

PrimCAM. PrimCAM allows the user to import Standard Tessellation Language (STL) files but

provides poor drawing tools. The Bioscaffolder can possess up to 5 dispense heads allowing

Fig.1. Schematic overview of the hybrid bioprinting process. A three-dimensional design is

translated to a deposition protocol which alternates steps of printing polymer and cell-laden

hydrogels to yield hybrid constructs. Source : [2]

Fig.2 (a) Image of the Bioscaffolder printer and its two printheads (b), PCL auger screw

pump on the left and syringe piston on the right. (c) Visual representation of the missing

heating piece.

(a) (b)

à

(c)

P a g e 10 | 78

the printing of a wide range of material from medium to high viscosity biomaterials (biopoly-

mers) to low and medium viscosity biomaterials such as hydrogels and silicones. The Bioscaf-

folder has been the first choice for the realization of the project since it was the only system in

the lab, at that time, able to dispense PCL and a GelMA based hydrogel ink in the same struc-

tures. Usually, the GelMA is heated up to 37°C and dispensed by a mechanical piston which

applies pressure over time to a syringe plunger (Fig. 2 b). Unfortunately, the piece ensuring

the heating of the syringe’s nozzle is not available in the lab (Fig. 2 c), consequently, GelMA

is cooling down in the nozzle impeding a proper deposition between the PCL strands. Finally,

after several attempts to overcome this issue, it was decided to change the tool used.

The 3DDiscovery (regenHU) (Fig 3 a & b): Compared to the Bioscaffolder, the 3D discovery

is a pneumatic based system that can host 4 different printheads in addition to a UV-light tool.

All of them work in a coordinated motion since they are connected to the same robotic arm;

thanks to this feature the 3DDiscovery has a work speed significantly higher than the Bioscaf-

folder. The HMI of the 3D Discovery use its own G-code, therefore it can only import .iso files

created in software provided by regenHU, BioCAD and MMconverter (see chapter 4 for more

details). The 3D Discovery can print a wide range of materials due to its 4 types of the

printhead, nevertheless, the printer can only print thermopolymers and highly viscous media

with an HM-300H printhead (Fig. 3 c). However, our lab made the choice to install two DD-

135N and two CF-300H (Fig. 3). In absence of the HM-300H, it is technically impossible to

print highly viscous material, such as PCL, with the 3Ddiscovery, and it was the main reason

Fig. 3. Image of the 3DDiscovery printer. (a) Enclosed in a flow box to print under sterile conditions. (b) A

close-up view of the 3DDiscovery: (1) flow box; (2) 3D Discovery; (3) air pressure regulators; (4) printheads;

(5) tool charger; (6) building platform; and (7) console. (c) Range of printheads and tools of the 3D Discovery.

Source : [14]

(b) (c) (a)

P a g e 11 | 78

why the 3D Discovery was not considered as a possibility when building the project plan. Re-

cently, RegenHU developed an experimental cartridge suitable for polymers with higher ther-

mal requirements such as PCL (Fig. 3), that fits in a DD-135N printhead. Considering the

difficulties with the bioscaffolder, it was decided to determine the potential of this cartridge for

the realization of the project.

This introduction presented the background of the project, the materials, the biofabrication ap-

proach and the tools involved. Below is the correct project process step by step:

- Implement and optimize PCL printing with the new cartridge

- Optimise the dual printing of PCL + GelMA in simple a shape in order to practice assays

- Optimise the dual printing of PCL +GelMA in an auricular shape

Therefore, for more clarity, this report is divided into 5 chapters, each of them dedicated to one

aspect of the project.

2. CHAPTER 1: PCL PRINTING

2.1 Introduction

As mentioned earlier this project is based on the ability of the printer to dual print PCL and

GelMA, thus the very first step was to set up PCL printing with the new experimental cartridge

(Fig. 3) of the 3D Discovery. The performance of this new cartridge was evaluated in relation

to the Bioscaffolder.

2.2 Material and methods.

2.2.1 Inks

Two different PCL types were used, the 704105 Polycaprolactone (Sigma Aldrich) (average

Mn 45,000) (Mw 48,000-90,000) and the medical grade PURASORB® PC 12 (Corbion Purac

Biomaterials, Spain) (IV midpoint 1.2 dl/g) (Mw 120,000[5]) (Mw = 130490, Mn = 79760

[6]).Only the 704105 Polycaprolactone from Sigma Aldrich was used with the Bioscaffolder.

2.2.2 Bioscaffolder

Manufacturing was performed at room temperature; 704105 PCL was heated up to 80°C and

printed with the extrusion printhead (Auger Screw Pump) through 330 µm (inner diameter)

nozzle (Fig.2) on the stationary platform covered with 2090 blue tape scotch 3M. PCL cannot

adhere properly to a microscope slide and deforms in absence of 2090 blue tape scotch 3M or

warming plate.

P a g e 12 | 78

Only the 704105 sigma PCL was printed with the bioscaffolder

since the barrel is hard to clean and the settings were already

optimized by previous work. (Precise settings in annex 1)

2.2.3 3D Discovery

In absence of the extruding printhead (HM-300H), it was neces-

sary to characterize the ability of the new cartridge to print PCL

(Fig. 3). From now on the experimental cartridge will be called

“EX”. We determined printing settings through a trial and error ap-

proach for three different nozzles; 3221130, 3221132 and

32211303 (Disposable ThinWall Insert Cores, Dispensinglink)

(Refer to: annex 1) and with two different PCL; 704105 PCL

(Sigma Aldrich), and the medical grade PURASORB PC 12 (Cor-

bion Purac). Manufacturing was performed with the EX cartridge

placed in a DD-135N printhead at room temperature, PCL heated up to 80°C and printed with

a pressure of 4.4 bar. Printing was completed on a warming plate (Thermobase platform

heater, regenHU) heating the support up to 32°C. 704105 Sigma PCL was printed on micro-

scope slides, whilst PURASORB PC 12 PCL was printed either on 2090 blue tape scotch 3M

or in a petri dish. Printing settings are present in the results. It is important to note that for both

the Bioscaffolder and the 3D Discovery the values presented in tables for “layer thickness” or

“strand interspace” are theoretical values typed into the drawing software (PrimCAM/Bio-

CAD/MMconverter). The layer thickness is the height the printer has to move up for every

additional layer, and the “strand interspace” is the distance between two strands of the same

layer but from their center (Fig. 4).

2.2.4 Characterising resolution of PCL printing - 3D Discovery versus Bioscaffolder

To characterise the resolution of PCL printing and make an accurate comparison between the

3D Discovery and the Bioscaffolder, a 2-layer design (Fig. 5 A) with decreasing strand inter-

space (1.6/1.4/1.2/1/0.8/0.6/0.4/0.2/0.1 mm) was printed, imaged with a stereomicroscope

(Olympus SZ61), and the strand thickness was calculated as the average of 10 values meas-

ured using ImageJ. In addition, PCL scaffold was printed (Fig. 5 B, C) at least 2 mm high with

a 2.25 mm strand interspace to analyze layer stacks and the overall geometry of the scaffold.

For both the 704105 sigma PCL and PURASORB PC 12 PCL, scaffolds were printed only with

the small (3221133) and medium nozzle (3221132). Pictures of the printed scaffold to measure

strand thickness can be found in Annex 2

.

Fig.4. Theoretical representation

of a PCL scaffold cross-section,

d1, d2 and d3 will be respectively

referred in our work by « meas-

ured strand thickness », « strand

interspace » and « layer thick-

ness ». Source : [8]

P a g e 13 | 78

2.2.5 Printing process

The printing process from modeling to printing is further explained in Chapter 4.

2.3 Results

2.3.1 Bioscaffolder - 704105 sigma PCL

The Bioscaffolder is a very reliable system for PCL printing, indeed with its extrusion-based

system, we can see (Fig. 6) that it can produce 465 µm fibers at the feed rate of 2.5 mm/sec.

Even though 2.5 mm/sec seems slow, printing PCL at such a resolution and speed is adequate.

When printing a scaffold, the bioscaffolder also performs well. We can see in Fig. 6 that the

overall square shape is true to the STL file, layers are perfectly stacked, the deposition is

homogeneous, and strands are parallel. However, in the absence of a warming plate, the print-

ing must be done on blue tape, and that is an obstacle to maintain sterility.

2.3.2 3DDiscovery

a) 704105 sigma PCL

The 704105 Sigma PCL was the first attempt to print PCL with the 3D Discovery, at that time

it was known that in absence of warming plate, the blue tape was essential to prevent PCL

scaffolds to deform during the printing. However, the team was not aware yet of the significant

impact of the support's surface on the deposition of PCL, and the thermoplate of the 3D Dis-

covery being fully functional, it was not deemed necessary to print on blue tape. In addition,

unlike blue tape, to print directly on microscope slides does not hinder sterility, and therefore,

Fig. 5. (a) two-layer design used to measure strand thickness, the strand interspace decreases on the X-axis

in the first layer and on the Y-axis in the second layer. The yellow mark indicates the first strand interspace of

1.6mm. (b) Visual representation of a PCL scaffold the red circle is pointing out what we are going to call the

single layer zone, while the green circle is pointing out the double layer zone (c) Theoretical representation of

a PCL scaffold from an upper view.

(a) (b) (c)

P a g e 14 | 78

was seen as an advantage. As it will be explained in the discussion, fewer results are pre-

sented for 704105 sigma PCL since efforts have been focused on PURASORB PC 12 PCL,

explaining why the printing of 704105 Sigma PCL hasn’t been further investigated on blue

tape.

We can see in Fig. 7 that 704105 sigma PCL can be printed into a range of 674,4µm to

323,5µm strands. The feed rate dropping significantly from 7 mm/sec for the biggest nozzle to

0.75 mm/sec for the smallest nozzle, changing the speed from 3x times faster than the Bi-

oscaffolder to 3x times slower. The standard deviation is reasonable for each result. If globally

the printing is satisfying; the strands are not perfectly parallel.

We can see in Fig. 7 a scaffold printed with the small nozzle 3221133, the overall shape and

geometry are satisfying, even if there are noticeable round corners. The first layer (Fig. 7) is

perfectly deposited, but since the layer thickness is not optimized, after a few layers the strand

is not perfectly stacked anymore and starts to overhang as the distance between two layers

Printer Bioscaffol-der

3DDiscovery

Material 704105 Sigma PCL

704105 Sigma PCL PURABSORB PC 12

Support Blue tape Microscope slide Petri dish Micros-cope slide

Nozzle Gauge 23 3221130 3221132 3221133 3221130 3221132 3221133 3221133

Inner Ø nozzle (µm)

337 564 335 234 564 335 234 234

Feedrate (mm/sec

2,5 ≈ 7 2 0,75 2 1 0,8 0,5 0,35 0,25

Measured strand thick-ness d1 (µm)

≈ 465,7 ≈ 674,4 ≈ 572,4 ≈ 323,5 ≈552,9 ≈242,5 ≈0,369 ≈ 186,7

≈ 280,5

≈ 437,3

Standard de-viation d1

14,74 28,68 7,5 12,65 49,37 10,06 42,2 21,73 19,63 22,72

Printer Bioscaffolder

PCL 704105 Sigma PCL

Nozzle 23 Ga

Inner Ø nozzle (µm) 337

Feedrate (mm/sec) 2,5

Strand interspace d3 (mm) 2,25

Layer thickness d2 (mm) 0,1

Number of layer 22

Construction time (min) 7,93

Length (mm) 9

Height (mm) 2,2 Fig. 6. Table of printing settings and pictures of a 704105 PCL scaffold printed with the Bioscaf-

folder from (a) a perspective view, (b) upper view and a lateral view (c)

(a) (b)

(c)

Table 1. Measurement of the dispensed strand thickness (d1) depending on the biofabrication procedure.

P a g e 15 | 78

increases, especially in the single layer zones of the construct described Fig. 5 b. Therefore,

strands start to deform and are not straight anymore, resulting in an altered shape easily visible

on a lateral view (Fig. 7c) and on the upper view (Fig. 7b). In the settings, we can read in the

table “Advised Layer Thickness (mm)”, it is a personal estimation of what should be the correct

“Layer Thickness” that would lead to an optimized construct.

b) PURASORB PC 12 PCL

All the prints of the PURASORB PC 12 have been made on a petri dish, except for one result

(Table 1). Actually, first attempts to print the PURASORB PC 12 were made on a microscope

slide with the small nozzle (3221133), exactly as we did for the 704105 sigma PCL. This ob-

tained a disappointing result since in addition to a feed rate 10 times slower than the Bioscaf-

folder, a slightly better resolution of 437 µm in average was obtained (Table 1), the strands

are not perfectly parallel (annex 1). However, when the nozzle was switched to the medium or

big size, the PCL was coming off the microscope slide, even when changing all the different

printing parameters. The conclusion was reached that the PURASORB PC 12 was not adher-

ing properly to the glass surface. After several unsuccessful attempts to increase adherence

by varying the temperature of the microscope slide, two different support were trialed, 2090

blue tape scotch 3M and a petri dish. PURASORB PC 12 shows the same printing performance

on a petri dish than on the 2090 blue tape scotch 3M, at the only difference that the PCL is

adhering so strongly on the

Printer 3D Discovery

PCL 704105 Sigma PCL

Nozzle 3221133

Inner Ø nozzle (µm) 234

Feedrate (mm/sec) 0,75

Strand interspace d3(mm) 2,25

Layer thickness d2 (mm) 0,18

Advised Layer thickness d2(mm) 0,12-0,15

Number of layer 12

Construction time (min) 14,4

Length (mm) 9

Height (mm) 2,16

Fig. 7. Table of printing settings and pictures of a 704105 PCL scaffold printed with the

3DDiscovery from (a) a perspective view, (b) upper view and a lateral view (c)

(a) (b)

(c)

P a g e 16 | 78

petri dish that, it can be difficult to detach it from the petri dish with no harm. Results shown

here were obtained on petri dish only because it was easier to image than on blue tape. For-

tunately, by increasing adhesion to the surface, 2090 blue tape scotch/petri dish improved the

deposition of PURASORB PC 12. We can see in Fig. 8 that PURASORB PC 12 PCL can be

printed in a range of 552.9µm to 186µm parallel strands depending on the nozzle and the feed

rate. Yet, even printed on blue tape/petri dish, feed rates used to obtain these resolutions are

significantly lower when working with PURASORB PC 12 PCL (from 2mm/sec to 0.35 mm/sec)

than with 704105 sigma PCL or the Bioscaffolder. The standard deviation is especially high for

the big nozzle (3221130) since it was hard to find optimized printing setting resulting in strands

with non-uniform width, and alteration of the overall shape. On the other hand, the standard

deviation of the sample printed with the medium nozzle (3221132) at 0.8 mm/sec is high, only

because the strand thickness is measured after imaging, and for unclear reasons the image

quality of this sample was poor, resulting in a less accurate measurement.

The printed scaffold obtained with PURASORB PC 12 PCL shows the importance of optimizing

printing settings, especially the layer thickness. For example, scaffolds printed with the small

nozzle (32211333), stacking of layers is influenced by layer thickness. In Fig. 8a, b & c the

settings are optimized, showing how well the layers are stacked, either on the single layer

zones or the double layer zones described Fig. 5b. While on Fig. 8 d, e & f, we can see

Printer 3D Discovery

PCL PURASORB 12

Nozzle 3221133

Inner Ø nozzle (µm) 234

Strand interspace d3 (mm) 2,25

Feedrate (mm/sec) ≈0,35

Support Blue tape

Layer thickness d2 (mm) 0,06 0,15

Number of layer 40 16

Construction time (min) 102,86 41,14

Length (mm) 9 9

Height (mm) 2,4 2,4

B

Fig. 8. Table of printing settings and pictures of two PURASORB PC 12 PCL scaffolds printed with

the small nozzle of the EX cartridge of the 3DDiscovery from (a)(d) a perspective view, (b)(e) upper

view and a lateral view (c)(f)

(a) (b) (e) (d)

(c) (f)

P a g e 17 | 78

overhanging between layers in the single layer zones as already described earlier for the

704105 Sigma PCL scaffold, but still correctly stacked in the double layer zones. In addition,

we can notice the impact on the overall geometry in the upper view of Fig. 9 and Fig. 9, the

first one has a perfect square shape while the second has round corners and strands are not

perfectly parallel. Furthermore, due to layers not being correctly stacked on the single layer

zones, we can reasonably suppose that the mechanical properties are strongly impeded as

well. However, it is important to point out that in addition of the printing speed being really slow,

the more layer thickness is reduced, the more layers are required to reach the same height,

and therefore the longer the printing time will be. Thus, for a simple square of 2.4 mm and

9mm of length, it required approximately 1h 43 min, so approximately 14 times longer than the

Bioscaffolder. The results concerning scaffolds printed with the medium nozzle (32211332)

don’t really provide more information and are given in annex 1. During experiments, it appeared

that in absence of warming plate, PURASORB PC 12 constructs deform on blue tape (Fig. 9)

but not on a petri dish.

2.4 Discussion

Throughout the results, we can understand how the printing output is impacted by numerous

factors, such as the support, the properties of the PCL, the nozzle or even the HMI settings. It

is hard to foresee, prior to testing, their respective degree of impact on the printing output.

Clearly shown is the importance of properly characterizing a material, especially its molecular

weight, since even though the 704105 Sigma and PURASORB PC 12 are both PCL, their

adhesion to the support and their feed rate are different. Furthermore, during research in liter-

ature we found out that in addition to changing the printing output, PCL with different molecular

weight will show different stiffness, that could have an impact on cell differentiation and poten-

tially on other factors [7]. Secondly, we observed the impact of settings such as layer thickness

on the output, from the shape of the scaffold to its mechanical integrity. Even though the over-

hanging layers described in the single layer zones could potentially be used to promote diffu-

sion at the expense of weakening the scaffold.it is important to note that if the 2090 blue tape

scotch 3M is necessary to correctly print with the PURASORB PC 12, it is an additional hurdle

for sterility in a perspective of bioprinting. Concerning the 704105 sigma PCL It is very im-

portant to take a step back, it would be a mistake to assume that the PURASORB PC 12 PCL

Fig. 9. Picture of a deformed PURASORB PC 12 PCL scaffold due to the absence of warm-

ing plate while printed on blue tape

P a g e 18 | 78

is printed with a better resolution than the 704105 Sigma PCL. It is important to keep in mind

that, by lack of time, printing settings are not optimized and the printing was studied only on a

microscope slide. Yet we know that printing on microscope slide was possible with the PURA-

SORB PC 12 only at a very slow speed, therefore, by printing the 704105 sigma PCL on blue

tape or a petri dish, theoretically it should be possible to increase the printing speed, and con-

sequently reach a resolution as good as or better than with the PURASORB PC 12.

Finally, if the PURASORB PC 12 is printed at a very slow feed rate by the 3DDiscovery, it

doesn’t necessarily mean that the 3DDiscovery is slower than the bioscaffolder, since the print-

ing of PURASORB PC 12 hasn’t been experimented with the bioscaffolder. On the other hand,

if the printing of the 704105 sigma PCL on blue tape with the 3DDiscovery is optimized, it

should be possible to increase the printing speed and consequently increase the resolution.

Furthermore, when printed on microscope slide and with the medium nozzle (32211332), that

has a comparable diameter than the bioscaffolder nozzle (335µm vs 337µm), the 704105

sigma PCL currently has a resolution only 100µm (572µm vs 465µm) higher than the bioscaf-

folder and is printed only 0.5 mm/sec slower (2mm/sec vs 2.5mm/sec) (Table 1). Therefore,

the 3DDiscovery could be as performant as the bioscaffolder.

2.5 Conclusion

In this first chapter, we saw that despite a pneumatic based extrusion the 3D Discovery is able

to properly print PCL with the EX cartridge, however, the performance thereof may highly vary

depending on the properties of the polymer itself (molecular weight) and the optimization of

printing settings and conditions.

Finally, even if this work was able to implement and optimize the printing of PCL with the 3D

Discovery and its pneumatic-based extrusion, it is important to note that the feed rate and layer

thickness necessaries to obtain good results highly vary between the different nozzles and the

PCL used. Therefore, the choice of the nozzle and of the material will condition the potential

of the 3D Discovery. With the smallest nozzle and the PURASORB PC12, the 3D Discovery

can print biocompatible PCL scaffold with precision but at the expense of a very slow printing

time, making the realization of big scaffold impossible or extremely time-consuming.

3. Chapter 2: Spraying vs Deposition

3.1 Introduction

After implementing and optimizing the printing of PCL with the 3D Discovery, the next step of

the project began: dual printing of PCL and cell-laden GelMA. However, two different

P a g e 19 | 78

approaches were possible to dispense cell-laden GelMA; spray with CF-300H (Fig. 3 C) print-

head, or deposition with DD-135N (Fig. 3 C) printhead. For better clarity, this short chapter will

be dedicated to the description of these two approaches, while the results of the experiment

are in Chapter 3.

3.1.1 Jetting of cell-laden GelMA with CF-300

Spraying liquid GelMA in between the PCL strands with the CF-300H through a microvalve is

the first method that was trialed. Firstly, the CF-300H is designed for accurate jetting or contact

dispensing, and CF stands for Cell Friendly. This method is extensively used

by several projects in the UMC laboratory, therefore it is already known as a

reliable method that can be optimized quite quickly. Since the GelMA is dis-

pensed as a liquid, optimizing the printing can be reduced to the optimization

of the quantity of material dispensed and can be done in a relatively short pe-

riod of time.

However, there are a few issues to address when printing cells with this dis-

pensing method. First, since the GelMA is liquefied at 37°C, during long printing

sessions cells slowly fall and accumulate at the bottom of the cartridge with

time (Fig. 10). As a consequence, it is really difficult to determine if the ho-

mogeneity of the cell concentration in the dispensed GelMA is preserved

throughout the printing. If gently shaking the cartridge could solve this prob-

lem, it requires unscrewing the cartridge and increase chances of breaking

sterility.

Secondly, since the GelMA is being jetted in a liquid form, it can easily be dispensed homoge-

neously in a closed volume, however it doesn’t have any shape fidelity and therefore presents

a limited scope of use as a filling, while confronting the user to issues such as leaking of GelMA

out of the printed construct.

3.1.2 Deposition of GelMA with DD-135N

Using the other approach, with the DD-135N of the 3D Discovery it is technically possible to

print GelMA in a gel form. In addition to solving issues related to the printing of liquid GelMA

mentioned earlier, it would allow new perspectives.

In the first place, cells being immobilized in the GelMA, their distribution stays homogeneous

across the cartridge and throughout the printing. Secondly and most importantly, if GelMA

could be printed with shape fidelity, it would imply the ability to precisely dispense GelMA in

complexed patterns, highly increasing the scope of use [4]. However, GelMA poor printability

Fig. 10. Picture of

the cell pellet at the

bottom of the car-

tridge in 37°C

GelMA

P a g e 20 | 78

causes difficulty for the fabrication of complex porous 3D scaffolds, and as a response, re-

searches were dedicated on its rheological properties and its dispensing. Billiet et al. showed

in their work the impact of numerous factors such as the temperature, the nozzle shape, or the

inlet pressure on the rheological properties of GelMA, and the necessity to control and monitor

these factors for precise dispensing of GelMA [8].

However, although the perspective of precisely patterning GelMA is truly interesting, it can be

really challenging to implement it, and facing these difficulties some research teams preferred

to investigate the creation of new bio inks based on GelMA that would have a better manufac-

turability [9][10].

3.1.3 Cell viability: Spraying versus Deposition

If implementing the dispensing of cell-laden GelMA in spray or deposition are two different

challenges from a technical point of view, they have a different impact too on a primary aspect

of bioprinting that is the cell viability. Preserving the cell-viability is primary in bioprinting, and

almost all the possible factors involved in the printing process were reviewed to determine how

they are impacting cell viability. Factors such as the inlet pressure, the material temperature

(directly influencing the viscosity), or the nozzle used (shape, length, inner diameter). However,

without prior knowledge, it is impossible to foresee the influence of each of these factors, and

consequently to determine which approach between spraying or deposition is likely to be the

most successful when it comes to preserving the highest cell viability. Luckily, the impact of

these factors on cell viability is well documented in the literature. Billiet et al. concluded that

the highest cell viability is obtained at a low inlet pressure (<2 bar) with conical nozzle [8]. At

high inlet pressure (>3 bar) cylindrical and conical nozzle of same inner diameter haven’t

shown any real difference. They created a heat map of the shear stress in different nozzle

(Fig. 11 B), and found that higher peak of shear stress can be found in the conical nozzle but

only at the very end of the tip, while even if shear stress is lower in the cylindrical it is present

Fig. 11. (A) Range of nozzle, (1) 300µm cylindrical nozzle for GelMA spraying, (2)(3) respectively 330µm

and 200µm cylindrical nozzle for GelMA deposition, and (4) 200µm conical nozzle for GelMA deposition.

(B) Cell-gel flow during syringe needle deposition. Heat map of the shear stress at 1 bar inlet pressure

for a conical needle (a) and cylindrical (b) needle, obtained by finite element modelling of non-cross-

linked cell-gel (10 w/v%) mixture. Fig.s are to scale for needle internal diameter of 200 µm. Source: [8]

1

2 3

4 A

A

P a g e 21 | 78

all along the nozzle [8]. This observation could explain the results obtained by Jones et al. that

concluded that cell viability diminishes with the length of a cylindrical nozzle since then cells

would be exposed for a longer time to the shear stress [11]. Finally, Yan et al. and Nair et al.

published two papers about the influence of inlet pressure and nozzle inner diameter on cell

viability, and came to the conclusion that the cell viability decreases as the pressure increases

and the nozzle diameter decreases; the effect of pressure being significantly larger than the

nozzle inner diameter. Furthermore, their surface-fitting model along with their shear stress

model shows the correlation between cell viability and shear stress induced by the process

parameters (inlet pressure and nozzle inner diameter)[12][13].

Therefore, we can try to predict the viability of our two approaches according to literature:

• In the spray approach, a 15 · 106 cells/ml cell-laden 10%(w/v) GelMA is heated at 37°C,

hence, is liquefied, and sprayed through a short cylindrical nozzle with an inner diam-

eter of 300µm (Fig. 11 A 1) and a low inlet pressure of 0.5 bar/7.25 psi.

• In the deposition approach, a 15 · 106 cells/ml cell-laden 10%(w/v) GelMA is kept at

18-24°C, hence, is viscous, and deposited through conical or cylindrical nozzle of dif-

ferent range of inner diameter (Fig. 11 A 2 3 4), with low to medium inlet pressure (1

to 3 bar).

3.2 Conclusion

Consequently, according to literature, it was expected the sprayed cells approach a higher cell

viability, since GelMA viscosity is lower, printed with a bigger nozzle and a low pressure. Even

if working with a range of different nozzles, or different temperatures, with the deposition ap-

proach, it is likely that the cell viability will be lower since the viscosity of GelMA will require

higher inlet pressure for extrusion.

4. Chapter 3: Cell viability and printing

After reviewing the main differences and features of two possible approaches for dual printing,

below outlines the printing procedure and obtained results.

4.1 Introduction

As explained in the general introduction, for dual printing a cell-laden GelMA hydrogel solution

is deposited between strands of PCL with the 3D Discovery using layer-by-layer deposition

according to a computer-aided design (Fig. 12 a). Even though the final goal of the current

work is to dual print our material in a complex ear shape, chapter 1 demonstrated that such a

print would take several hours, consequently, it would be nearly impossible to print several

P a g e 22 | 78

scaffolds in one work day. Therefore, it is difficult to optimize efficiently the deposition and is

impossible to undertake assays. Thus, as a first step, a simple design was chosen: a simple

square shape (Fig. 12 b) which consistently reduced the printing time. The GelMA will be

dispensed between strands of PCL, either as a liquid form and sprayed with a CF-300H print-

head or deposited as a gel with a DD-135N printhead (Fig. 3c). PCL is deposited with the EX

cartridge in a DD-135N printhead and with the settings used to obtain results of chapter 1. Our

first concern was to study if cell viability is preserved after these printing conditions, at which

extent and how to improve it. After facing several technical issues, a molded cell-laden GelMA

experiment that doesn’t require the correct printing of scaffolds was implemented in order to

have an insight of the impact of the printing process on cell viability despite issues related to

the material.

4.2 Material and method

4.2.1 Sterility

To assure sterile printing conditions every component in contact with printed tissue must be

sterilized, and the same guidelines used by Rimann et al. were incorporated [14]. Equipment

(thermobase, microscope slides, cartridge heater/cooler...etcetera) needed for the experiment

are installed prior to the sterilization. Special attention is given to microvalves that are cleaned

prior to sterilization by ultrasonication for 15 min at 40°C, to ensure it is not clogged. Then, the

flow box of the bioprinter was cleaned with 75% ethanol and further sterilized by UV light for

at least 30 min. Rimann et al, as well as our laboratory protocol, recommend to sterilize by

autoclave all the components to be, However, the manufacturer Nordson EFD warns on every

package to not heat the syringe barrels higher than 38°C, with no further instructions for pis-

tons, blue caps or tip caps. Thus, it was preferable to immerse every component in 75% etha-

nol and further sterilized by UV light in the flow box of the printer for at least 30 minutes.

4.2.2 Inks

Dual printing was achieved with 704105 Polycaprolactone (Sigma Aldrich) (average Mn

45,000) (Mw 48,000-90,000) and 80% DOF cell-laden GelMA. The photoinitiator combined

with GelMA in this experiment is 2-hydroxy-1-[4-(2-hydroxyethoxy) phenyl]-2-methyl-1-pro-

panone (Irgacure 2959, Ciba, Basel, Switzerland), stock solutions in PBS were filter sterilized

and mixed with GelMA to obtain a final concentration of 0.05% (w/v). The GelMA was then

dissolved in filter-sterilized PBS at the concentration of 12.5%(w/v) and mixed on a roller for 1

hour in at 37°C incubator. Confluent cells after 1 or 2 weeks of culture were detached with

trypsin to be collected, re-suspended in DMEM, centrifuged, and resuspended in PBS to be

P a g e 23 | 78

evenly mixed with GelMA (8:2, GelMA volume: cell- suspension volume) on a roller for 10 min

at 37°C, to finally obtain a final cell-laden 10% GelMA solution with a cell concentration of 15

x 106 cells/ml. ATDC5 cells were used only for training whilst equine auricular chondrocyte

cells were used for the final experiment of both printed hybrid construct and molded gels.

4.2.3 Bioprinting procedure hybrid constructs

Manufacturing was performed at room temperature, 704105 PCL was loaded and printed with

the EX cartridge placed in a DD-135N printhead and heated up to 80°C. Cell-laden (ATDC or

equine auricular chondrocyte) GelMA (see ink section) is loaded in a 10ml disposable cartridge

in a sterile manner (Nordson EFD, Ohio, USA) and co-printed either by spraying or deposition.

Printing settings can be found in Table 2. The Computer-aided design (CAD) file was designed

in BioCAD and imported in the HMI of the 3D Discovery. The two materials are co-printed on

sterile microscope slides placed on a warming plate (Thermobase platform heater, regenHU)

heated to 32°C. Once finalized, constructs were crosslinked by 10 min irradiation with a UV

Superlite (Lumatec, Munchen, Germany) (S-UV 201 A, 220 volts, 50 Hz) with an intensity of ≈

13.3 mW/cm² measured at 365 nm, a total UV irradiation of 7980 mJ/cm². Finally, constructs

were moved to 12-well culture plates containing medium and maintained in culture (see cell

culture section).

4.2.4 Cell-laden molded gel

Printing settings

PCL

Printing settings

GelMA

(sprayed)

Printing settings

GelMA

(deposited)

Material 704105 sigma PCL 10 % GelMA 10 % GelMA

Printhead 1 (DD-135N + EX) 4 (CF-300H) 2 (DD-135N)

Nozzle 3221133 cylindrical 300µm cylindrical 200µm

Pressure (bar) 4,4 0,5 3,3

Feedrate (mm/sec) 0,75 30 20

Temperature (°C) 80 37 24

Layer thickness (mm) 0,12-0,15 0,12 0,12

Strand interspace

(mm)

2,25 2,25 2,25

Cellularity (cells/ml) - 15*10^6 15*10^6

Valve opening time

(µs)

- ≈1950 -

Dosing distance (mm) - 0,7 -

Table 2. Printing settings of dual printed 704105 PCL scaffold with either sprayed GelMA or

deposited GelMA

P a g e 24 | 78

Hydrogel precursor solutions were prepared in PBS, with a final concentration of the photoin-

itiator Irgacure 2959 (Ciba, Basel, Switzerland) of 0.05% (w/v). GelMA has been dissolved on

a roller for 1 hour in a 37°C incubator. Confluent equine auricular chondrocytes after 1 or 2

weeks of culture were detached with trypsin to be collected, then re-suspended in DMEM,

centrifuged, and resuspended in PBS to be evenly mixed with 12.5 %(w/v) gelMA (8:2, gelMA

volume: cell- suspension volume) on a roller for 10 min at 37°C to reach a final concentration

of 15 · 106 cells/ml. In a sterile manner, cell-laden gelMA was either printed in deposition con-

ditions (cylindrical nozzle ø 200µm, 2.2 bar/31.9psi, 24°C) or spraying conditions (cylindrical

nozzle ø 300µm, 0.5 bar/7.25 psi, 37°C) on a petri dish, and collected to be injected in a Teflon

Mould producing gels of 6mm*2mm (disc: diameter*height) (≈56.55mm3). The gels were photo-

crosslinked using 365nm UV-light Superlite (Lumatec, Munchen, Germany) (S-UV 201 A, 220

volts, 50 Hz) with an intensity of ≈ 13.3 mW/cm² for 10 min (7980 mJ/cm²). Finally, gels were

moved from the mold to 12-well culture plates containing medium and maintained in culture

(Refer to cell culture section).

4.2.5 Cell culture and constructs culture

ATDC 5 and equine auricular chondrocyte cells were cultured in a flask maintained in a hu-

midified 5% CO2-containing atmosphere (37°C), cultivation mediums were changed every 2

days. Cultivation medium for equine auricular chondrocyte was consisting of DMEM + Gluta-

Max-I, +4,5g/L D-glucose, + pyruvate (Life Technologies), supplemented with 10% FBS, pen-

icillin (100 U/mL), streptomycin (100 µg/ mL). While ATDC5 cultivation medium was consisting

of DMEM/F-12 HAM’s medium + GlutaMax-I (Life Technologies), supplemented with 5% FBS,

penicillin (100 U/mL), streptomycin (100 µg/ mL), for ATDC 5. Confluent cells after 1 or 2

weeks of culture were collected and mixed with gelMA to be used in the biofabrication proce-

dure described earlier in this chapter. The hybrid constructs and molded gels were then placed

in 12-well plates and maintained at 37°C in 5% CO2 condition with fresh medium change every

Fig. 12. Design and three- dimensional (3D) bioprinting of hybrid constructs with structural and biolog-

ical features. (a) hybrid biofabrication using pneumatic dispenser heads. (b) Schematic of designed 3D

hybrid construct with alternating strands of polycaprolactone (PCL) and chondrocyte-impregnated

GelMA in each layer. Source [3]

(a) (b)

P a g e 25 | 78

two days. Whilst the same medium as the flask culture was used for auricular chondrocyte

constructs and moulded gel, a medium consisting of DMEM/F-12 HAM’s medium + GlutaMax-

I (Life Technologies), supplemented with 5% FBS, penicillin (10D 0 U/mL), streptomycin (100

µg/ mL), dexamethasone (0.1 mM), ascorbate-2-phosphate (0.4 mM), and ITS X (1X) was

used for ATDC 5 constructs.

4.2.6 Live/Dead assays of ATDC5 constructs and molded gels

To qualitatively determine the cell viability in the hybrid constructs and molded gels, a cell

viability assay was conducted using a LIVE/DEAD Kit (ThermoFisher scientific) and fluores-

cence microscopy. Living cells with normal intracellular esterase activity are stained by the

green fluorescent Calcein-AM dye. On the other hand, Ethidium Homodimer-1 (EthD-1) dye

enters cells with damaged membranes and undergoes an enhancement of fluorescence upon

binding to nucleic acids, thereby producing a bright red fluorescence in dead cells. The viability

of cells in the hybrid constructs and molded gels was assayed immediately after biofabrication

(day 0) and at days 1, 3, and 7 of subsequent in vitro culture. At each time point, the con-

structs/gels (n=1 per condition) were removed from the culture, washed with plain DMEM, and

stained in 2 µM calcein- AM and 2 µM EthD-1 solution in DMEM for 20 min in a 37°C, 5% CO2

incubator. The constructs were washed with PBS thrice for 5 min and imaged using an upright

fluorescence microscope (Olympus BX51) under 40X magnification. Pictures could not be

treated with ImageJ due to the high cell density and the quality of the signal, therefore quanti-

tative results couldn’t be obtained.

4.2.7 AlamarBlue® of molded gels

To quantitatively determine the cell viability in the molded gels, a cell viability assay was con-

ducted using AlamarBlue® DAL1025(Invitrogen). AlamarBlue® (resazurin) is a proven non-

toxic cell viability indicator that uses the natural reducing power of living cells to convert resaz-

urin into resorufin, therefore, the amount of fluorescence produced is proportional to the num-

ber of cells and to the metabolic activity of living cells. The viability of cells in molded gels was

assayed immediately after biofabrication (day 0) and at days 1, 3, and 7 of subsequent in vitro

culture. At each time point, the medium is changed and 1/10th volume (200µl) of AlamarBlue®

reagent is directly added to the culture media. The reaction was incubated in darkness for

16h40 at 37°C, 5%. After incubation, three 200 μl replicates were taken from each well and

transferred into a 96-well plate with a flat bottom. Finally, fluorescence is measured with a

microplate fluorometer, Fluoroskan Ascent FL (Thermo Fisher Scientific, MA, USA) (excitation

544 nm, emission 572 nm). After removing all assay solution and washing the samples twice

P a g e 26 | 78

with PBS, the samples were cultured further for use in the next experiment. The incubation

time was overnight, it is important to note that even if the assays were started at day 0-1-3-7,

results are labeled as the day after (1-2-4-8) on Fig.16 since the measures correspond to the

end of the incubation.

4.2.8 Histology staining

ATDC5 laden gelMA and PCL dual printed constructs were prepared for histological analysis,

first fixated in 4% formalin at room temperature for 24 h, then dehydrated with baths of increas-

ing concentration ethanol (70%-100%), then is subsequently be submerged in a histo-clear

bath. Finally, the samples were embedded in paraffin and cut in a 5µm slice with a microtome.

4.3 Results

4.3.1 Dual printing optimization

Unfortunately, the results are very limited as dual printing could not be optimized due to tech-

nical issues (Refer to troubleshooting). The spraying approach with 704105 PCL and sprayed

ATDC5-laden gelMA was successful and a first draft of the settings is available in table 2.

Several sprayed ATDC5 cell-laden constructs were successfully printed and were used initially

for live/dead training and optimization of the protocol. Even though the signal is not very clear

as can be seen in Fig. 13, ATDC5 can be observed on PCL strands and overall the scaffold.

Even though higher numbers of dual printed scaffolds were successfully printed, more

live/dead data is not available since for a short period the lab changed the calcein-AM (Refer

to section: troubleshooting). For the deposition approach, it is very hard to draw any conclu-

sions. Indeed, some constructs were printed with the settings provided in table 2, but the print-

ing was hardly reproducible one day to another even within the same settings and conditions,

illustrating the challenging aspect of this approach as described in the previous chapter. As-

says could not be achieved on these samples due to fungal infection. Sadly, no results of dual

printing with PURASORB PC12 PCL or with equine auricular chondrocyte are available due to

the timeline of the project and technical issue with the printer (Refer to section: troubleshoot-

ing).

Fig. 13. Live/dead picture at day 7 of dual printed construct of 704108 PCL and ATDC5 impregnated

gelMA. In red are delimited the PCL strands. Magnification x100

P a g e 27 | 78

4.3.2 Live/Dead of cell-laden molded gels

Fig. 14 demonstrates the live/dead result of the molded cell-laden gelMA pucks. At day 0, we

can observe that after printing the green signal is clear with a sharp definition of cells, there is

hardly any significant difference between each condition, it is noticeable that red and green

signals don’t overlap. The darker crack with a higher red signal on the “before printing” condi-

tion, is simply a defect in the gel due to manipulation. After day 1, in every condition the red

and green signals start to overlap, some cells presenting both red and green fluorescence.

The result of deposited gelMA at day 1 appeared to be, after further experimentation, charac-

teristic of a gel being upside down. Indeed, during incubation the gel rests on the bottom of the

Day 0

Day 1

Day 3

Day 7

Unprinted cells Sprayed cells Deposited cells

Fig. 14. Live/dead pictures of moulded gels with unprinted, sprayed or deposited chondrocyte

impregnated GelMA, from day 0 to day 7. Magnification X40.

P a g e 28 | 78

well-plate, consequently, reagents are not diffusing properly through the gel from the thereof

base. Due to a lower fluorescence signal, it was necessary at day 3 or 7 to increase the exci-

tation time, increasing at the same time the background signal coming from gelMA

autofluorescence and introducing noise into the signal. However, even if once more, it is hard

to see a significant difference between the conditions, a yellow tone can be observed but due

to the noise and the different focal planes, it is difficult to say if more red and green signal are

superposed of if there is only an increase of the red signal. We can see the same results at

the extremity of the gels Fig. 14, with a signal mostly yellow on day 7. Finally, Fig. 15 demon-

strates that in a gel cut in a half, the fluorescent signal is only present at the surface of the gel.

4.3.3 AlamarBlue® of cell-laden molded gels

Fig. 16 demonstrates the evolution of the fluorescence intensity as a function of the number

of days, the fluorescence intensity is directly related to the number of cells and their metabolic

activity. At day 1 the non-printed cells show the highest fluorescence, while the signal of

sprayed and deposited cells have a signal 3 times and 6 times lower, respectively. If we look

at numbers, sprayed cell signal is almost twice that of the deposited cells signal. Between day

1 and day 2, all the signals dropped and are at their lowest point. However, while unprinted

cells and sprayed cells signals are respectively at 62% and 50% of their initial value, deposited

cells signal dramatically dropped to 10% of its initial value. From day 4 to day 8, the signal of

unprinted cells and deposited cells are slowly rising again to reach respectively 73% and 88%

of their value on day 1. The signal of sprayed cells is more unpredictable, it suddenly increases

between day 2 and 4, to reach a value of 237% of the initial one, and finally dramatically de-

creases between day 4 and 8. Despite this second drop, the sprayed cell signal increases

again to reach a final value of 122% of the initial value (day 1). It is plausible that values meas-

ured at day 4 or 7 for sprayed cells are biased for unknown reasons, resulting in this curious

fluctuation of the fluorescence signal. Finally, on day 8 the unprinted cells signal is still the

highest with a value twice higher than the sprayed cells signal and 5 times higher than the

deposited cells. The signal of the sprayed cells was determined to be 2.5 times higher than the

deposited cells.

Fig. 15. Live/dead pictures of moulded gels with sprayed chondrocyte impregnated gelMA

at day 3 and cut in half. Magnification 40X

P a g e 29 | 78

4.4 Troubleshooting

Unfortunately, only a few results are presented here, it is important to peruse the reasons that

hindered the correct progression of the project. First, the printing of PURASORB® PC 12 is

described in the first chapter and does not appear here. It is simply because PURASORB®

PC 12 was made available in the laboratory only 2 months before the end of this internship.

Therefore, the few results presented here for the dual printing only concerns the 704105 Sigma

PCL and were obtained prior to the arrival of PURASORB PC 12 PCL in the laboratory. The

limited quantity of results with the PURASORB PC 12 PCL can be explained by the fact that

the second optimization of printing settings was necessary, retarding the implementation of

dual printing. Secondly, the cell concentration used for the project is 15.106cells/ml, however,

in order to have a reasonable amount of material (4-5 ml), it requires a substantial number of

cells and consequently a considerable culture time. This explains why the training portion of

the project was realized with ATDC5, due to their fast growth. It is essential to note that reach-

ing such a large number of cells with auricular chondrocyte can take up to 2 weeks, limiting

the number of attempts to print and obtain results at 2 or 3 print sessions only per month.

Furthermore, when the experiment was finally ready to be carried out, the microvalve was not

functional for a month. Thirdly, it is fortunate that it was possible to properly prepare the exper-

iment while cells are in culture with cell-free gelMA for the spraying approach, since at 37°C

the gelMA has a low viscosity, subsequently cells have only a small impact on the rheological

properties. Otherwise for the deposition approach where gelMA is very viscous due to a lower

manufacturing temperature, cells have a big impact on rheological properties as shown by

Billiet et al.[8], and the experiment needs to be prepared and optimized with cells. Therefore,

attempts were even more limited due to cell speed of growth for the deposition approach.

More results were also expected from the live/dead assays, unfortunately during an

experiment, the calcein regent of the laboratory was changed (aliquots labeled “Calcein-AM, 1

mg/ml, MB 21-04-2016”), this new calcein was not compatible for these live/dead assays.

When treated with the same protocol, gelMA presented an intense auto-fluorescence covering

all cells signals and made any observations impossible. The protocol was changed with differ-

ent dilutions, from ½ to 1/16 of the normal concentration of calcein, different incubation times,

trials on cell-free gels and gels with lower cell concentration, but every time the output was the

same with an intense auto-fluorescence. Eventually, a new stock of calcein from Life technol-

ogies was available in the lab allowing correctly performance of the live/dead of the last exper-

iment.

P a g e 30 | 78

As outlined in the material and methodology section, the incubation time used for AlamarBlue®

was long (16h40). Even though AlamarBlue® incubation time can go up to 18h, there might

be a more efficient incubation time. However, as printing cells is extremely time-consuming,

and AlamarBlue® incubation time takes at least 2 hours, this forced the AlamarBlue® to be

conducted overnight. The UV exposure used in the project is so high simply because the UV

intensity could be measured only at the end of the experiment when the optometer X9-2 (Gi-

gahertz-Optik, Germany) was available in the lab (Refer to annex 3).

4.5 Discussion

Henceforth, only the results obtained with the molded gels will be discussed as, with the ex-

ception of a few live/dead pictures, the dual printed samples couldn’t be used for further ex-

periments. It must be noted that due to the low number of samples and the technical issues

faced throughout this project, these results are not statistically significant, and therefore must

be interpreted with hindsight. It would be interesting to reproduce experiments a second time

with a higher sample size. The results from the live/dead assays are hard to interpret. At the

first glance, we could think that cell viability is decreasing over time and independently of the

printing process due to the yellow tone of the pictures that indicates an increase of the red

signal. However, the hypothesis of a decrease of cell viability over time would be is not corre-

lating with our AlamarBlue® results and would result in pictures with a reddish tone. If we look

closely at the pictures, we can observe that red and green signal are superposing, resulting in

this yellow tone. We know that Ethidium Homodimer-1 (EthD-1) dye produces a bright red

162,35

101,49111,18

119,31

48,49

24,62

115,18

59,64

27,83 2,9821,70 24,64

0,00

20,00

40,00

60,00

80,00

100,00

120,00

140,00

160,00

180,00

0 1 2 3 4 5 6 7 8 9

Flu

ore

sce

nce

inte

nsi

ty (

arb

itra

ry u

nit

)

Days after printing

un-printed

spraying

deposition

Fig. 16. Evolution of the fluorescent intensity of moulded gels made of chondrocyte impregnated gelMA

in unprinted, sprayed or deposited condition, as a function of time.

as function of time

P a g e 31 | 78

fluorescence upon binding to nucleic acids, but instead of signalling a damaged membrane of

a dying cell, it could indicate that cells have porous membranes. Live/Dead assay was initially

designed for two-dimensional culture, yet we know that cells can behave differently in three-

dimensional conditions. Thus, in addition of explaining the superposition of red and green sig-

nal resulting in the yellow tone of pictures, this hypothesis would be consistent with the Ala-

marBlue® results that indicate an increase of fluorescence, and consequently an increase of

cell-proliferation or cell-viability. However, the sharpness of the signal diminishing with time for

unknown reasons, and in absence of triplicate, conclusions cannot be drawn from these pic-

tures. The Live/dead assay might not be suitable for our work. The cell density is too important

and the structure too thick to allow any valuable information to be gathered in this way. Indeed,

with this number of cells, a lot of signal from different focal planes overlap and treating the

image with ImageJ is impossible. Moreover, when working with thick constructs as in this study,

it is impossible to observe all the focal planes with a fluorescence microscope. Therefore, it

would be interesting to use a confocal microscope and see if the signal is clearer when com-

piling the different focal planes. Secondly, we saw Fig. 15. that the fluorescent signal is only

present at the surface of the gel, therefore either reagent might not diffuse correctly through

the whole sample or the signal exists but can’t be observed because the light can’t diffuse

properly in the gel. Finally, the live/dead technique is a good qualitative indicator, but output

involving thick constructs with high cell density is limited.

The AlamarBlue® is a great alternative or complementary technique to the live/dead, as it gives

quantitative information on cell viability and cell proliferation despite the thickness of the

construct or a high cell density. However, it can be challenging to use it for printed scaffolds

as, to accurately compare different conditions it is important to start with an equal number of

cells between each scaffold, therefore it is of prime importance that the same volume of mate-

rial is dispensed in each scaffold and that the cells concentration is homogeneous in the car-

tridge.

In our case, the AlamarBlue® results do not support the live/dead results but matches the

literature since the spraying approach led to a higher fluorescence signal than the deposition

approach, meaning that more cells are alive or are more metabolically active. The fluorescence

signal of unprinted cells is higher than the printed cells, proving an impact of the printing pro-

cess on cell viability. All the conditions show the same tendency, the fluorescence signal of

AlamarBlue® drops in the first 48h and then recovers after the day 2. The drop of fluorescence

could be explained by the fact that apoptosis might take some time, therefore immediately after

printing cells still show an active metabolism while they are actually dying, leading to a lower

fluorescence signal 24h later. Literature has shown that if the printing process is harmless then

this drop in cell viability won’t be observed, and the cell proliferation will lead to an increase of

the fluorescence over time [4]. On the other hand, if the printing process negatively affects cell

P a g e 32 | 78

viability, we might observe a drop in cell viability and a recovery over time [16][2]. The curious

thing about these presented AlamarBlue® results is that the drop of the fluorescence signal is

present 48h after printing and then shows a recovery. Furthermore, even cells that did not go

through the printing process show a drop of the fluorescent signal. How to explain this drop for

48h and not only 12h for all the samples?

Since the drop in viability is affecting all the samples including the unprinted cells, it could be

related to the photo polymerization process that all samples underwent. The impact of a con-

centration of 0.05%(w/v) of Irgacure 2959 on cell viability is well documented in literature, and

has been shown compatible with acceptable level of cell viability in 3D construct after reason-

able UV exposure, therefore [17][18] it is unlikely that Irgacure 2959 is causing the drop of the

fluorescence signal. On the other hand, the total intensity of UV exposition in our experiment

(7980 mJ/cm²) is important, while in the literature values of UV exposition are generally around

1300-1800 mJ/cm² to preserve a good cell viability. The work of BIlliet et al. showed a cell

viability of only 55.72±7.26% for UV-A irradiation doses of 5400 mJ/cm² [8], leading to the

conclusion that an exposition of 7980 mJ/cm² must severely hinder cell viability and thus the

fluorescent signal.

Why are the fluorescence signals of the sprayed and deposited cells so low compared to the

unprinted cells signal? The answer is in the question, it may be reasonably asserted that the

printing process is the cause of the lower fluorescent signal, and even though our results are

not in triplicate they seem to concur with literature. Indeed, chapter 2 demonstrated that with

the chosen printing conditions, sprayed cells should have a higher viability than the cell depos-

ited that undergo higher shear stress during printing. However, a smaller drop of the fluores-

cent signal of the sprayed cells would have expected since the conditions and settings used

should preserve a high cell viability. The important UV exposition combined with the printing

stress could increase the adverse impact on cells. Once more, cautiousness is required re-

garding conclusions since the experiment currently has only been completed once.

Finally, in the material and method appears a histology section, even though no results are

presented in the result section. Several dual printed samples went through the described pro-

tocol, and histo-clear was used instead of xylene since xylene is a solvent of PCL and would

partly damage our samples. However, with the PCL present throughout the construct it was

impossible to properly cut a slice for staining, it could be suggested to use cryostat and see if

the slicing process is easier. On the other hand, if we would use xylene it might be possible to

slice our sample but depending on the manipulation skill of the user the sample integrity will

be affected by the empty space left by the PCL, and potentially information on the interaction

between the gelMA and the PCL would be lost. We can reasonably assume that the team of

Kang et al. that published in the magazine Nature have high-level skills, however, obtaining

P a g e 33 | 78

perfectly sliced histologic slide is very difficult when PCL is involved [15]. Therefore, the choice

between xylene and histo-clear might depend on the architecture of the PCL scaffold of the

experiment and of the skill of those involved in the project.

4.6 Conclusion and Further Experiments

The material and methods have shown the importance of correctly choosing the reagents,

even if the molecule is the same the results might be different. For instance, the calcein from

life technologies worked with gelMA constructs, however, the calcein aliquots in the lab never

gave proper results due to an intense gelMA autofluorescence that was covering any signals,

despite modification of the protocol. For any further work, it is clear that 7980 mJ/cm² UV is

harmful to cell viability and it seems logical to reduce the total UV exposition to a value closer

of 1800 mJ/cm². The next experimenter can refer to annex 3 to calculate the time of exposition

and distance from the UV-light depending on the one he will choose. Even though it is not

really necessary, potentially changing the photoinitiator would improve results, indeed Rouil-

lard et al. and Billiet et al. have both shown higher cell viability using the photoinitiator VA-086

compared to Irgacure2959[8][19].

Also discussed was the difficulty to create histology slices using the 3D printed construct partly

made of PCL, and the solvent used (xylene vs histo-clear) to prepare the histology slide must

be determined depending on experimenter’s skill and the PCL scaffold.

Secondly, related to our results, even if by lack of sample and statistical analysis our results

need to be confirmed by further experimentation, it seems to confirm that the sprayed cell

preserves a higher cell viability/proliferation than cells printed in the deposition conditions. It

was expected that sprayed cells would preserve a higher cell viability/proliferation, even though

the intense UV exposure might have biased our results. The live/dead is difficult to interpret by

lack of sharpness of the signal over time, but the superposition of the red and green signals

over time is truly interesting and could suggest a change of the membrane porosity, not related

with apoptosis. Therefore, it is recommended to reproduce the Moulded gel experimentation

with more samples to confirm the first results obtained in this study, and to use a calibration

curve for AlamarBlue® to provide better insight.

Dual printed scaffold results are discussed in Chapter 5 conclusion.

5. Chapter 4: Auricular shape and biofabrication

5.1 Introduction

The goal of this work is to dual print PCL and gelMA in the shape of auricular cartilage. How-

ever, even after going through all the previous steps we are unable to correctly dual print PCL

P a g e 34 | 78

and gelMA in a morphological shape due to software limitations. Indeed, the 3D Discovery

being a printer sold by regenHU, regenHU is in charge of providing adapted software. Never-

theless, regenHU is currently providing two software for our lab, specifically “BioCAD” and

“MMconverter”. To precisely understand this issue, it is necessary to describe these software

in more depth. Even though the Bioscaffolder is no longer a candidate for the realization of the

project, its biofabrication procedure will be described for information.

5.2 Material and method

5.2.1 Bioscaffolder and PrimCAM

The Bioscaffolder is controlled through a software combining the HMI and a drawing interface

called PrimCAM. PrimCAM provides very poor drawing tools, but its main usefulness is its

ability to import STL files and precisely control layer by layer the printing settings. On the other

hand, its interface is not perfectly adapted for dual printing and performs poorly. We can see

in Fig. 17 the typical translation of a file from its creation in a CAD software to its printing.

5.2.2 BioCAD

BioCAD is a layer by layer type of drawing software, thus it is designed to allow the creation of

simple shape with basic tools. It is possible with some expertise to create layers with a complex

internal architecture combining different material, but it is not made to design a complex shape

such as the ear. Therefore, BioCad was used in all the previous work to obtain a square-

shaped scaffold, but cannot be used for an auricular shape. We can see Fig. 18, drawings

made in BioCAD and the printed result with gelatin dyed orange.

5.2.3 MMconverter

Print

Fig. 17. Biofabrication process of the Bioscaffolder with a STL file of the auricular shape, through

the drawing software PrimCAM, and the HMI that controls the print.

3D model (.STL) Drawing PrimCAM (.CAM)

HMI (NC code)

P a g e 35 | 78

MMconverter allows the user to import STL files and convert them into .iso files importable in

the HMI. Therefore, it is possible to work with all sort of complex shapes since the source is

an STL file that can be designed with some expertise in any 3D modeling software. The soft-

ware provides the possibility to work with different materials, but not simultaneously. “Simulta-

neously” meaning that the two materials can be printed in the same construct but not in the

same layer, which constitutes an issue. Consequently, it is possible to pattern PCL in an au-

ricular shape but impossible to precisely dispense gelMA in between the PCL strands. Further-

more, if MMconverter can import STL file like PrimCAM, it does not provide a control over the

printing settings layer-by-layer, therefore the output is considerably limited. We can see in Fig.

19 the typical translation of a file from its creation in a CAD software through to its conversion

in a .iso file understood by the 3D Discovery.

5.3 Results and troubleshooting

One issue encountered when working with BioCAD and MMconverter is the difficult translation

between the two software. BioCAD offers total control over settings such as the strand inter-

space layer-by-layer, whilst MMconverter only provides limited control options such as “filling

pattern type” and “density”. It is not clear how MMconverter calculates the strand interspace

from a filling density, and consequently, it is impossible to precisely control the strand inter-

space through MMconverter. The problem being that the printing settings controlling the quan-

tity of gelMA that needs to be dispensed mainly depends on the strand interspace, thus print

settings that have been optimized in BioCAD cannot be used directly in MMconverter until the

filling density in percentages is not directly linked to a strand interspace distance in millimeters.

For this purpose, it is necessary to choose a “density”, print a scaffold, and measure the strand

Print

Fig. 18. Biofabrication process of the 3DDiscovery with a drawing in bioCAD to the HMI that con-

trols the print and a picture of a dual printed construct of PCL and orange dyed gelMA

Drawing BioCAD (.BCD) HMI (.iso)

Fig. 19. Biofabrication process of the 3DDiscovery with an STL file from the modelling in the CAD software

(Rhinoceros 4.0), through MMconverter, and the HMI that control the print

Print

3D model (.STL) MM converter (.iso)

HMI (.iso)

P a g e 36 | 78

interspace manually corresponding to this filling density. An attempt was made to work around

the problem in MMconverter, since the jetting of gelMA does not necessarily require a perfect

pattern. As the jetting is not consistent, the gelMA is not homogeneous in the construct and

the PCL is not deposited correctly. This is due to the gelMA being jetted on the strands and

thus impedes adhesion on the previous layer, as the gelMA is not dyed it can be difficult to

make a sharp observation. Another foreseen problem is the printing time. Indeed, in Chapter

1 it was shown that for a simple PURASORB PC 12 PCL square of 9mm*2mm (length*height),

it took up to 1h 40 min with the smallest nozzle. However, with the platform heating the support

at 32°C, in our few attempts It has been noticed that during long prints (>1 hour) the gelMA is

drying over time. If partially crosslinked gelMA in between each layer could be a solution, it

would be much easier to simply increase the feed rate of PCL and greatly reduce the printing

time.

5.4 Discussion and Conclusion

In response to this software limitation, RegenHU developed “Object pattern”, a plugin for Bio-

CAD. It can be assumed that “Object pattern” would allow the import of an STL file while

combining the layer-by-layer precise control of BioCAD. Nonetheless, it is in beta version and

still not available in our lab. Until a new software or plugin is obtained it will be impossible to

properly dual print two materials on the same layer in a complex shape.

6. Chapter 5: General conclusion and prospective work

To end this report below is a review of the main points of each chapter in a final conclusion.

First this work demonstrated that the 3D Discovery with pneumatic based extrusion and new

EX cartridge can reliably print PCL, however, depending on the materials (material’s molecular

weight, nozzle Ø) its potential highly varies. If the 3D Discovery can print PURASORB PC 12

strand up to 186Ø it is at the expense of an extremely slow feed rate. Therefore, the 3D Dis-

covery cannot combine a high resolution and a high printing speed with its pneumatic extrusion

but remains a powerful tool. This is described as well as the importance of characterizing a

polymer, since its properties, especially its molecular weight will impact all the aspects of a

biofabrication project - from its printing to the mechanical properties of the scaffold and its

influence on cells as mentioned by Hendrikson et al.[7]Overall, the first 3D Discovery combined

with the EX cartridge remains a powerful tool despite a significantly slow feed rate as it can

combine PCL scaffolds with one material deposited with the DD-135N left and two other inks

sprayed with the two CF-300H, while the new 3DDiscovery can combine high-resolution PCL

scaffold but only with one sprayed material.

P a g e 37 | 78

This report has shown how gelMA could be dispensed in a scaffold either as liquid filling or a

shaped gel strand, and the impact of each of these approach on cell viability. The conclusion

was that the sprayed approach showed a higher preservation of cell viability after printing,

whilst being easily implementable and therefore should appear as a first choice even if as a

liquid its potential is limited. On the other hand, the deposition approach of gelMA represents

a great potential but remains highly challenging and requires the user to undertake a time-

consuming work of optimization before being efficient. This is before mentioning a bigger im-

pact on cell viability due to the higher pressure required for extrusion, and therefore should be

considered only if essential for the correct progression of a project. In the case of the gelMA

deposition, it can be interesting to consider the modification of the ink, for example as men-

tioned earlier with hyaluronic acid or gellan to increase the printability of the material.

Another aspect of this work is the importance of choosing and adapting assays depending on

the project, especially in biofabrication where depending on the sample, this may not occur.

Indeed, the materials used might not be compatible with the assay’s protocol, for example in

histology where a sample with a PCL scaffold will dissolve in xylene providing the necessity of

changing the solvent to histo-clear, which then results in samples that are hard to process with

a microtome depending on the PCL architecture of the scaffold. Beyond materials, the sample

might not be suitable for an assay due to its dimensions, this work showed a 2D sample or thin

samples where live/dead provides a lot of information. Conversely, in thick constructs, the out-

put of the assays can be greatly limited or require adaptation of the protocol conditions and

collection of the data, for example by changing the microscope. Being confronted by these

limitations can lead to rethinking the analysis of a project, in this case, for example, opting for

alamarBlue that gives quantitative results independently of the dimension of the sample and

without cytotoxicity, allowing us to experiment at different time point despite a poor number of

samples is a highly useful tool.

Finally, even when the printing process is mastered, it is important to keep in mind that the

potential of a printer directly depends on its software, in this case even if theoretically the printer

can dual print an ear shape, as the software can’t code this shape while combining two mate-

rials, it entirely hinders the progression of the project.

The conclusion of this work is that, when starting a biofabrication project, every detail of its

framework must be defined. Biofabrication provides us with powerful tools, however, the po-

tential of these tools can be fully exploited only if every aspect of the project is cautiously

studied. The numerous factors from the material itself to the software, the biofabrication pro-

cess or the assays and how they interact (Fig. 20) and interfere in the printing process can be

P a g e 38 | 78

very disconcerting for the user and lead to time-consuming and fruitless work, therefore it is

essential for the experimenter to:

- Characterize properly the materials/inks; an example provided in this report is the im-

pact of the molecular weight of two different PCL types or the physical of gelMA on the

biofabrication process

- Optimize the printing process, such as the used tools; this work showed that a different

printer or different printhead will be used depending on the resolution needed for the

PCL scaffold, the materials it must be combined with, and to the method for combining

them (sprayed or deposited). Take into account the limitations imposed by the software

that is at the origin of the biofabrication process

- Printer calen-dar

- Assay calen-dar

P a g e 39 | 78

Fig. 20. (B) Concept map of variables and relations critical to biofabrication of an hydrogel, adapted

from:[20]. (A) Simplified concept map of variables and relations critical to biofabrication of a thermo poly-

mer, (C) complex interactions between the thermo polymer and the hydrogel during dual printing, mainly

directed by the software. (D) Elements not directly impacting the biofabrication process but orbiting around

it and necessary to consider.

Thermo polymer - Polymer - Molecular weight

Printing fidelity

- Complexity

- Resolution

- Construct size

Viscosity/adhe-

rence on sup-

port

Nozzle

gauge

Fabrication

time

Software

A

C

B

Experimental assays

Planning: - Cell culture - Printer calendar - Assay calendar

D

P a g e 40 | 78

REFERENCES [1] I. a Otto, F. P. W. Melchels, X. Zhao, M. a Randolph, M. Kon, C. C. Breugem, and J. Malda, “Auricular

reconstruction using biofabrication-based tissue engineering strategies,” Biofabrication, vol. 7, no. 3, p. 032001, 2015.

[2] W. Schuurman, V. Khristov, M. W. Pot, P. R. van Weeren, W. J. a Dhert, and J. Malda, “Bioprinting of hybrid tissue constructs with tailorable mechanical properties.,” Biofabrication, vol. 3, no. 2, p. 021001, 2011.

[3] Z. Izadifar, T. Chang, W. Kulyk, X. Chen, and B. F. Eames, “Analyzing Biological Performance of 3D-Printed, Cell-Impregnated Hybrid Constructs for Cartilage Tissue Engineering,” Tissue Eng. Part C Methods, vol. 22, no. 3, p. ten.tec.2015.0307, 2016.

[4] H.-W. Kang, S. J. Lee, I. K. Ko, C. Kengla, J. J. Yoo, and A. Atala, “A 3D bioprinting system to produce human-scale tissue constructs with structural integrity,” Nat. Biotechnol., vol. 34, no. 3, pp. 312–319, 2016.

[5] L. A. Bosworth, S. R. Rathbone, R. S. Bradley, and S. H. Cartmell, “Dynamic loading of electrospun yarns guides mesenchymal stem cells towards a tendon lineage,” J. Mech. Behav. Biomed. Mater., vol. 39, pp. 175–183, 2014.

[6] E. Díaz, I. Sandonis, and M. B. Valle, “In Vitro Degradation of Poly ( caprolactone )/ nHA Composites,” vol. 2014, 2014.

[7] W. J. Hendrikson, J. Rouwkema, C. a. van Blitterswijk, and L. Moroni, “Influence of PCL molecular weight on mesenchymal stromal cell differentiation,” R. Soc. Chem. Adv., vol. 5, no. 67, pp. 54510–54516, 2015.

[8] T. Billiet, E. Gevaert, T. De Schryver, M. Cornelissen, and P. Dubruel, “The 3D printing of gelatin methacrylamide cell-laden tissue-engineered constructs with high cell viability,” Biomaterials, vol. 35, no. 1, pp. 49–62, 2014.

[9] F. P. W. Melchels, W. J. a. Dhert, D. W. Hutmacher, and J. Malda, “Development and characterisation of a new bioink for additive tissue manufacturing,” J. Mater. Chem. B, vol. 2, no. 16, p. 2282, 2014.

[10] W. Schuurman, P. a. Levett, M. W. Pot, P. R. van Weeren, W. J. a Dhert, D. W. Hutmacher, F. P. W. Melchels, T. J. Klein, and J. Malda, “Gelatin-methacrylamide hydrogels as potential biomaterials for fabrication of tissue-engineered cartilage constructs,” Macromol. Biosci., vol. 13, no. 5, pp. 551–561, 2013.

[11] A. Faulkner-Jones, C. Fyfe, D.-J. Cornelissen, J. Gardner, J. King, A. Courtney, and W. Shu, “Bioprinting of human pluripotent stem cells and their directed differentiation into hepatocyte-like cells for the generation of mini-livers in 3D.,” Biofabrication, vol. 7, no. 4, p. 044102, 2015.

[12] K. Nair, M. Gandhi, S. Khalil, K. C. Yan, M. Marcolongo, K. Barbee, and W. Sun, “Characterization of cell viability during bioprinting processes,” Biotechnol. J., vol. 4, no. 8, pp. 1168–1177, 2009.

[13] K. C. Yan, K. Paluch, K. Nair, and W. Sun, “Effects of Process Parameters on Cell Damage in a 3d Cell Printing Process,” Imece2009 Proc. Asme Int. Mech. Eng. Congr. Expo. Vol 2, pp. 75–81\n525, 2010.

[14] M. Rimann, E. Bono, H. Annaheim, M. Bleisch, and U. Graf-Hausner, “Standardized 3D Bioprinting of Soft Tissue Models with Human Primary Cells.,” J. Lab. Autom., p. 2211068214567146–, 2015.

[15] H.-W. Kang, S. J. Lee, I. K. Ko, C. Kengla, J. J. Yoo, and A. Atala, “A 3D bioprinting system to produce human-scale tissue constructs with structural integrity,” Nat. Biotechnol., vol. 34, no. 3, pp. 312–319, 2016.

[16] Y. Yu, Y. Zhang, J. a Martin, and I. T. Ozbolat, “Evaluation of cell viability and functionality in vessel-like bioprintable cell-laden tubular channels.,” J. Biomech. Eng., vol. 135, no. 9, p. 91011, 2013.

[17] N. E. Fedorovich, M. H. Oudshoorn, D. van Geemen, W. E. Hennink, J. Alblas, and W. J. a Dhert, “The effect of photopolymerization on stem cells embedded in hydrogels,” Biomaterials, vol. 30, no. 3, pp. 344–353, 2009.

[18] J. Jung and J. Oh, “Influence of photo-initiator concentration on the viability of cells encapsulated in photo-crosslinked microgels fabricated by microfluidics,” Dig. J. Nanomater. Biostructures, vol. 9, no. 2, pp. 503–509, 2014.

[19] A. D. Rouillard, C. M. Berglund, J. Y. Lee, W. J. Polacheck, Y. Tsui, L. J. Bonassar, and B. J. Kirby, “Methods for photocrosslinking alginate hydrogel scaffolds with high cell viability.,” Tissue Eng. Part C. Methods, vol. 17, no. 2, pp. 173–179, 2011.

[20] J. Malda, J. Visser, F. P. Melchels, T. Jüngst, W. E. Hennink, W. J. a Dhert, J. Groll, and D. W. Hutmacher, “25th anniversary article: Engineering hydrogels for biofabrication,” Adv. Mater., vol. 25, no. 36, pp. 5011–5028, 2013.

P a g e 41 | 78

ANNEX 1

Printer Bioscaffolder

PCL 704105 Sigma PCL

Needle gauge 23 Ga

Inner Ø nozzle (µm) 337

Feed rate (mm/s) 2,5

Feed XY 150

Feed Z 400

Spindle speed 200

Material Temperature (°C)

80

Support Blue tape

Printer 3D Discovery

PCL 704105 Sigma PCL

Nozzle 3221133

Inner Ø nozzle (µm) 234 Feedrate (mm/sec) 0,75

Strand interspace d3(mm) 2,25

Layer thickness d2 (mm) 0,18 Advised Layer thickness d2(mm) 0,12-0,15

Number of layers 12

Construction time (min) 14,4

Length (mm) 9

Height (mm) 2,16

Table S2. Bioscaffolder processing conditions for the deposition of 704105 Sigma

PCL.

Table S1. Technical description provided by the manufacturer of the metal nozzle used for PCL printing

with the EX cartridge of the 3DDiscovery

Fig. S1. 3DDiscovery processing conditions and pictures of scaffold printed with 704105

Sigma PCL.

P a g e 42 | 78

ANNEX 2

Fig

. S

2. S

ett

ing

s a

nd

mea

sure

d s

tra

nd th

ickne

ss o

f b

oth

th

e B

ioscaff

old

er

and

th

e 3

DD

isco

ve

ry w

ith

th

e t

wo

diffe

rent P

CL

an

d w

ith

pic

ture

of th

e 2

laye

r co

nstr

ucts

used

fo

r m

easu

rem

ents

.

P a g e 43 | 78

ANNEX 3

Table 1.

Bioscaffolder processing conditions for the 3D fiber deposition of 704105 Sigma PCL.

P a g e 44 | 78

FROM TISSUE ENGINEERING TO DRUG DELIVERY

We have seen that despite manufacturing challenges, the fabrication of complex scaffolds is

very promising for tissue engineering. However, tissue engineering is not the only field of re-

search that benefited from the development of additive manufacturing techniques together with

biomaterials compatible with those techniques. Thus, scaffolds also appeared to be potential

tools for drug delivery, and more precisely for local drug delivery. The following work explores

the usage of scaffolds as a drug carrier for local drug delivery; for this purpose, polycaprolac-

tone scaffolds were manufactured. Moreover, a novel composite was fabricated and charac-

terized to improve drug release.

P a g e 45 | 78

.

Quentin Clément PEIFFER

QUT student #: n9827102

MSc Biofabrication

Microporous polycaprolactone

scaffolds for local drug delivery

Daily supervisor: Dr.Phong Tran

Examiner: Dr.Ir.J.Malda

D/Prof.Dietmar W. HUTMACHER

Major Research Project

IHBI QUT

03/10/2016 – 03/07/2017

P a g e 46 | 78

LAYMAN’S SUMMARY

Drugs administered in the blood are effective but can also be very toxic as the entire body is

exposed. If delivered directly to the diseased site, drugs which are poorly soluble in the blood

can be administered in higher doses and so be more effective while being less toxic. However,

a local approach can be complicated as it implies to have access to the diseased site and as

such is especially relevant if surgery is required. As an example, local drug delivery would be

adapted for the treatment of cancer recurrence after tumor removal surgery or could be used

to reduce the risk of bacterial infections along with inflammatory reactions from the body. The

development of tissue engineering along with 3D manufacturing techniques led to the fabrica-

tion of implants with controlled features which have appeared to be promising tools for local

drug delivery. This work focus on the development of a new composite material which would

be compatible with 3D printing and that could sustain the delivery of drugs, notably by creating

small pores on the implant surface. The new composite material was successfully fabricated

in porous structures that could release drugs. Nonetheless, porous structures have shown no-

table benefits only for soluble drugs compared to the non-modified material. Finally, more work

is required to have a better understanding of how the creation of pores affects the material and

the release of the drug.

ABSTRACT

Challenges to developing efficacious systemic drug delivery systems remain, notably concern-

ing the administration of poorly soluble drugs. Local drug delivery appeared as an alternative

to avoid systemic toxicity while targeting delivery site. However, as an invasive approach, local

delivery is especially relevant in a post-operative context such as the treatment of cancer re-

currence, implant-related infections or foreign body reactions. Stimulated by tissue engineering

and the development of additive manufacturing, the fabrication of scaffolds with controlled fea-

tures have appeared to be valuable tools for local delivery. With viscoelastic properties

favorable with 3D manufacturing techniques along with a good biocompatibility, polycaprolac-

tone (PCL) is a biodegradable polymer that became prevalent in tissue engineering. The cur-

rent work aimed to characterize the printability of a novel PCL/PBS composite and investigate

its potential as a drug delivery carrier when manufactured in porous scaffolds. Rheological

measurements have shown that the addition of PBS microparticle to PCL changed its viscoe-

lastic properties in addition to modifying its sensitivity to temperature increase. Secondly, dif-

ferent release kinetics was only observed for soluble drugs while burst effect remained im-

portant despite porosity. Therefore, PCL porous scaffolds have shown limited benefits over

non-porous scaffolds, but more work is required as results suggest they could be biased by

lack of control over drug loading and drug release.

P a g e 47 | 78

ABBREVIATION DDS: Drug delivery system

FTIR: Fourier Transform Infrared Spectroscopy

hOB: human osteoblast

LVR: Linear viscoelastic Region

MRSA: Methicillin-resistant Staphylococcus aureus

PCL: Polycaprolactone

nPCL: non-porous PCL

pPCL: porous PCL

SSI: Surgical Site Implant

TERM: Tissue engineering and Regenerative medicine

ACKNOWLEDGMENT D/Prof. Dietmar W. Hutmacher

Dr.Phong Tran

Dr.Christina Theodoropoulos

Hoang Phuc Dang

Margaux Vigata

Tara Shabab

CARF

KEYWORDS Local drug delivery; PCL; 3D printing; Paclitaxel; Dexamethasone; Cefazolin; Vancomycin;

Porous; Scaffold; controlled release

P a g e 48 | 78

TABLE OF CONTENTS ABSTRACT .......................................................................................................................................... 46

ABBREVIATION ................................................................................................................................. 47

ACKNOWLEDGMENT ....................................................................................................................... 47

KEYWORDS ........................................................................................................................................ 47

1. BACKGROUND AND LITERATURE REVIEW ................................................................. 49

1.1 DDS and local drug delivery ................................................................................49

1.2 Local drug delivery of paclitaxel in adjuvant cancer treatment .............................49

1.3 Surgical site infection and implant-related infection: cefazolin and vancomycin ..50

1.4 Dexamethasone: from biocompatibility to bone regeneration ..............................51

1.5 Local drug delivery and tissue engineering: scaffolding approach .......................51

1.6 Research question ..............................................................................................52

2. MATERIAL AND METHODS ............................................................................................... 52

2.1 Material ...............................................................................................................52

2.2 Preparation of PCL/Porogen composite ..............................................................52

2.3 Rheological study ................................................................................................53

2.4 Scaffold Manufacturing .......................................................................................53

2.5 Loading efficiency and in-vitro drug release studies (Figure 2) ............................54

2.6 Normalization of drug release data......................................................................55

2.7 Fourier Transform Infrared Spectroscopy analysis of paclitaxel-loaded films ......56

2.8 Cell culture ..........................................................................................................56

2.9 Dexamethasone bioactivity studies .....................................................................56

3. RESULTS AND DISCUSSION ............................................................................................ 57

3.1 Rheology.............................................................................................................57

3.1.1 Amplitude sweep analysis: linear viscoelastic region of nPCL and pPCL .......... 57

3.1.2 Temperature sweep analysis ..................................................................................... 59

3.2 Paclitaxel FTIR and Stereomicroscopy ...............................................................61

3.3 Drug release .......................................................................................................64

3.3.1 Porous and non-porous scaffolds ............................................................................. 64

3.3.2 Drug loss and loading efficiency ................................................................................ 65

3.3.3 Paclitaxel drug release................................................................................................ 65

3.3.3 Dexamethasone: drug release .................................................................................. 68

3.3.4 Dexamethasone: bioactivity ....................................................................................... 68

3.3.5 Cefazolin drug release ................................................................................................ 72

3.3.6 Vancomycin drug release ........................................................................................... 72

4. CONCLUSION AND FUTURE WORK ............................................................................... 75

REFERENCES .................................................................................................................................... 76

P a g e 49 | 78

1. BACKGROUND AND LITERATURE REVIEW

1.1 DDS and local drug delivery

Drug delivery corresponds to approaches, systems, technologies, and formulations designed

to administer a pharmaceutical compound to achieve a desired therapeutic effect. Since the

introduction of the first sustained release formulation in the 1950s[1], a significant amount of

drug delivery systems (DDS) were developed and successfully translated into clinics. But de-

spite a lot of efforts, the DDS developed in the past three decades haven’t known the same

success, notably concerning their application to clinics[1] (e.g. nanoparticles)[2]. This limited

success is mainly associated with the challenges of systemic administration[3], such as the

formulations of poorly soluble drugs [1] or targeted delivery[4]. In addition to have low oral

bioavailability, poorly soluble drugs are a problem for parenteral injection as they often require

excipient which can be very toxic (e.g. Paclitaxel and Cremaphor[5]) [1] [6]. However, it is

estimated that more than 40% of new chemical entities developed in the pharmaceutical in-

dustry are practically insoluble in water [6], which make them really difficult to develop into

clinically useful formulations (e.g. paclitaxel[5]). Secondly, in absence of targeting, a drug ad-

ministered parenterally can have severe adverse effects as the drug is delivered to the entire

body (e.g. cytotoxic drug), often resulting in low therapeutic index [7] [8].

Local drug delivery was rationalized as an alternative to the systemic administration as early

as in the 1980s, notably for chemotherapy treatments [9] [10] [11]. The benefits of a local

delivery over a systemic delivery have been rationalized several times[3] [7] [12]. By being

delivered directly to the targeted site, drugs can be administered at higher doses, and so, be

more potent while reducing the whole body toxicity [12]. Besides increasing the therapeutic

index, local drug delivery systems are also associated with longer exposure time as the drug

release is sustained for a prolonged period of time[3]. Nonetheless, except for parts of the body

which are easily accessible from the outside, e.g. skin, local delivery implies an invasive inter-

vention. However, the development of tissue engineering and the increasing use of medical

devices tend to indicate that local delivery is very promising. Below are presented 3 clinical

issues that could be addressed with local delivery.

1.2 Local drug delivery of paclitaxel in adjuvant cancer treatment

Paclitaxel is a chemotherapy drug from the Taxane family that stabilizes the microtubule poly-

mer and protects it from disassembly, ultimately leading to cell apoptosis during mitotic cellular

function [13]. Paclitaxel has a proven potency and is used in clinics to treat various cancers

such as ovarian cancer [14], metastatic breast cancer[15], or anthracycline-resistant breast

cancer[16]. Yet, paclitaxel is poorly soluble and its systemic administration is problematic and

P a g e 50 | 78

is associated with a low therapeutic index and an important toxicity[17]. The need for new

formulations of paclitaxel has been rapidly acknowledged [1][18][19], and similarly to other

toxic chemotherapy agents, the localized delivery of paclitaxel was suggested as a possible

solution [5] [20]. Local delivery would be extremely relevant in the adjuvant treatment of breast

cancer since an implant could be placed during surgery. Paclitaxel has already shown some

efficacy in the adjuvant treatment of breast cancer to prevent local recurrence which is a major

cause of death [21] [22]. An implant could deliver a high dose of paclitaxel increasing its bioa-

vailability to the tumor site while minimizing systemic side effects and improve the long-term

survival and quality of life of patients [7]. Furthermore, this implant could also deliver other

therapeutic agents to prevent bacterial infection or improve healing as explained below, nota-

bly when the risk of infection could possibly influence cancer recurrence[23].

1.3 Surgical site infection and implant-related infection: cefazolin and vancomy-

cin

With an incidence of 38%, SSI is the most common nosocomial infection among the surgical

patient, while accounting for approximately 15%[24] (up to 20% depending on the study[25])

among hospitalized patients, making them the third most frequently reported nosocomial in-

fections[25]. Thus, SSI and its complications represent an important clinical and economic

burden[24][25]. The risk of SSI can persist for up to 30 days after a surgical operation or even

up to one year if the patient is given an implant. Thus, the use of an implant and medical device

suffer from additional risks of ‘‘device-related’’ or ‘‘implant-associated’’ infection[3]. During

surgery, pathogens can colonize an implant surface which can result in host patient morbidity,

device removal and mortality in some cases. It appeared that preventing bacterial adhesion to

an implant in the next few hours following surgery, notably to avoid the formation of biofilms,

is decisive to achieve a long-term success of an implant [26] [27]. Hence, it was postulated

that the presence or release of antibiotics from the implant surface could possibly prevent such

colonization and significantly reduce the risk of implant-related infection[3].Cefazolin and van-

comycin are antibiotics commonly used in clinics as perioperative prophylaxis treatment to

prevent surgical site infection. Cefazolin is known for its efficacy against gram-positive bacteria

such as Staphylococcus aureus responsible for a large amount of SSI [25][28][29], while van-

comycin is meant to treat or prevent severe infection and is notably used against Methicillin-

resistant Staphylococcus aureus (MRSA) infections.[29] [30]. As such, the local delivery of

vancomycin and cefazolin to prevent SSI and implant-related infection, and more precisely

their efficacy to inhibit biofilm formation were studied and have shown promising results [3] [26]

[31] [32]. Venkata P. Mantripragada et al. even combined cefazolin and vancomycin to be

released together from microparticle for a better antibacterial effect [33].

P a g e 51 | 78

1.4 Dexamethasone: from biocompatibility to bone regeneration

When it comes to the clinical applications of biomaterials and medical devices, one major chal-

lenge remains beyond the risk of infection; the inflammatory reaction of the host, and notably

the foreign body reaction [34]. Any implant can trigger a non-specific host response leading to

the formation of a fibrous capsule surrounding the implanted object, which can lead to failure

of the implant [35]. As a result lot of efforts were done to have a better understanding of the

mechanism regulating biomaterial biocompatibility [36]. To improve the overall host response

and implant integration it has been proposed to reduce the inflammatory reaction. Dexame-

thasone is a corticosteroid medication used in diverse conditions to treat inflammation and

have shown promising results for such application. Hence, incorporation of controlled-release

systems, notably PLGA microsphere, of anti-inflammatory drugs such as dexamethasone was

developed and are very promising [3] [37] [38] [39] [40]. But local delivery of dexamethasone

could also potentially serve another purpose in TERM. In addition to anti-inflammatory proper-

ties, dexamethasone is known to be used in cell culture for its potent osteogenic differentiation

effect [41] [42]. Therefore, it has been suggested to combine bone implant with the controlled-

release system to deliver dexamethasone, and promising results seem to indicate that dexa-

methasone can improve overall bone regeneration [43].

1.5 Local drug delivery and tissue engineering: scaffolding approach

If local delivery from an implant is promising, it is a challenge to get precise control over the

kinetics of the release. The local release will depend on the implant and drug properties and

how they are combined, in this way tissue engineering highly stimulated drug delivery research

[44]. The development of biodegradable polymers [45] [46] [47] together with the emergence

of 3D manufacturing techniques led to the fabrication of scaffolds as novel therapeutic tools

for local drug delivery [48][49][50]. Polycaprolactone (PCL) is one of the biodegradable

synthetic polymers which has been increasingly used in tissue engineering and drug delivery

research. In addition, to be biocompatible, its viscoelastic properties and low melting point are

compatible with various manufacturing techniques which makes it an ideal polymer[51][52].

The combination of PCL with different manufacturing techniques to fabricate new PCL-based

delivery systems of therapeutic agents (drugs, DNA, and siRNA, proteins, growth factors) have

been extensively studied [53];[54], and summarised by Debasish Mondal et al.[55]. Several

strategies have been developed to load therapeutic agents on such DDS. On one hand,

therapeutic agents are encapsulated and/or blended in the matrix of the scaffold[55][56], the

release kinetics is then depending on the degradation rate of the polymer[57]. This approach

has been shown to be promising and is notably used with PCL to achieve a prolonged release

and avoid a burst effect by using PCL slow biodegradability. Still, this approach can require

P a g e 52 | 78

sophisticated manufacturing techniques as it can potentially highly decrease drug bioactivity

by exposing it to the manufacturing process (e.g. high temperature, organic solvent etc..)[56].

On the other hand, therapeutic agents can be deposited on the surface of scaffolds as a

coating, in this case, the release kinetic is depending on the degradation of the scaffold

together with a diffusion process[57]. This approach is favorable to the release of a higher dose

of therapeutic agents in a shorter period of time. The deposition of therapeutic agents on the

scaffolds can be seen as more versatile and can preserve higher rate of bioactivity for unstable

drugs as the process is independent of the manufacturing of the scaffold. Yet, this approach is

also known to be associated with an important burst release of therapeutic agents and a poor

ability to sustain the release over a short period of time [57]. Previous studies have shown that

manufacturing scaffolds with micropores was a promising technique[56]. By increasing the

surface area, a higher amount of therapeutic agents can be loaded and released in more

sustainable ways, it also seems that porous surfaces also to favor initial cell adhesion by

stimulating cell anchorage [56].

1.6 Research question

This work aimed to investigate the potential of porous PCL scaffolds as drug carrier compared

to non-porous scaffolds. To this end, a novel PCL/Porogen composite was fabricated and char-

acterized with rheology to determine its viscoelastic properties and printability. The composite

was then manufactured with 3D printing technique into porous scaffolds with an increased

surface area with the aim to improve drug loading efficacy and reduce burst release. Finally,

the drug release kinetics of 4 different therapeutics agents was studied; Paclitaxel, Dexame-

thasone, Cefazolin, and Vancomycin.

2. MATERIAL AND METHODS

2.1 Material

- Medical grade polycaprolactone Purac Purasorb PC12 (Mw 120,000 g/mol)

- PBS (Oxoid, BR0014 Dulbecco `A’ Tablets)

- Paclitaxel powder from Taxus brevifolia, ≥95% (HPLC) (T7402-5MG, Sigma-Aldrich)

- Dexamethasone powder, ≥98% (HPLC) (D1756-100MG, Sigma-Aldrich)

- Vancomycin hydrochloride from Streptomyces orientalis (V2002-1G, Sigma-Aldrich)

- Cefazolin sodium salt 89.1-110.1% (C5020-100MG, Sigma-Aldrich)

2.2 Preparation of PCL/Porogen composite

To prepare the composites, PBS pellets were crushed in a powder with a pestle and mortar.

The PBS powder was then sieved to obtain microparticles with Ø≤ 38µm. Three blends of PCL

P a g e 53 | 78

and porogen of different ratios (17/33/44 w%) are dissolved together in chloroform to be sol-

vent-casted in a beaker. After cooling down, the films are peeled and placed in zip bags to-

gether with autoclave bag containing silica gel beads. The bags are left open in the fume hood

for 2 days after what they are stored in a low humidity chamber.

2.3 Rheological study

It is important to precise that rheological measurements were done on PCL/PBS 17/33/44 w%

composites films which are not leached, as our focus is on how PBS microparticles are influ-

encing the printability of PCL. Rheological measurements were carried out on a rheometer

(Anton Parr M302 Rheometer) equipped with a cylindrical, parallel plate geometry using 25

mm diameter plate and 1mm gap. The different composite films were melted at 110°C for 5

min and underwent amplitude sweep tests to investigate their linear viscoelastic region. The

shear strain was ranging from 0.01% to 150%, and an angular frequency of 10 rad/sec was

kept constant during the analysis. Temperature sweep tests were also carried out on the dif-

ferent composite films to have a better understanding of their viscoelastic behavior in function

of the temperature. Composite films were melted at 125°C and the temperature was gradually

decreased by 5°C between each measurement with a 5min delay until reaching the final tem-

perature of 40°C. The angular frequency was maintained to 10 rad/sec and the shear strain at

1% during the analysis. For both temperature and amplitude sweeps, G' and G" were plotted

in two different figures for a better visibility, but to be able to compare them properly, a third

figure plotting tan δ was realized. Tan δ (also called loss tangent) corresponds to the ratio of

loss modulus to the storage modulus.

𝑇𝑎𝑛 𝛿 = 𝐺′′

𝐺′

2.4 Scaffold Manufacturing

Manufacturing was performed at room temperature. Medical grade PC 12 PCL (nPCL) and the

PCL/PBS composites films were respectively heated up to 100°C and 110°C for 30 min and

printed with a screw-driven extrusion-based printer (Fig. 1 a, b) through a ≈413μm ID (inner

diameter) nozzle on a stationary platform. Printed scaffolds measure 40mm2 and possess 6

layers, each layer being printed orthogonally to the previous one. The scaffolds were then

leached for 14 days in 0.01M NaOH to remove the PBS microparticles, creating different po-

rosities depending on the porogen ratio that was used. We can see (Fig. 1 c, d) a pPCL 44w%

scaffold after leaching. For all the experiments involving scaffolds, only scaffolds made of nPCL

and pPCL 44w% will be used. For more clarity, leached porous scaffolds and films will be

P a g e 54 | 78

designated as pPCL 17/33/44 w% for the different composites, while nPCL will stand for non-

porous PCL.

2.5 Loading efficiency and in-vitro drug release studies (Figure 2)

To carry out drug release experiments, scaffolds are cut with a scalpel in small scaffolds of

approximately 3 mm x 3 mm x 3mm (Fig.1 c, d). Those scaffolds are then weighted and placed

in 2ml microcentrifuge tube. The microcentrifuge tubes containing the scaffolds are sterilized

with 100% ethanol and dried for 24h in a flow cabinet. Release studies were carried out for 4

different drugs: Paclitaxel, Dexamethasone, Cefazolin, and Vancomycin. The quantity of drug

pipetted for loading is normalized by the weight of each scaffold. Therefore, Paclitaxel,

Cefazolin, and Vancomycin are loaded with three different doses; low dose: 0.4µg/mg of

scaffold, medium dose: 2µg/mg of scaffold, high dose: 10µg/mg of scaffold. On the other hand,

dexamethasone is loaded with three different doses: low dose: 1µg/mg of scaffold, medium

dose: 5µg/mg of scaffold, high dose: 25µg/mg of scaffold. Paclitaxel and dexamethasone were

dissolved in 100% ethanol while cefazolin was dissolved in 90% ethanol, and vancomycin in

70% ethanol, drugs were then pipetted in the 2ml microcentrifuge tubes containing the scaf-

folds and dried for 48h in a flow cabinet. Before carrying on the release studies, the scaffolds

were transferred to new sterile microcentrifuge tubes. The first tubes were stored at 4°C and

are used to measure the drug lost during loading. A summary of the protocol is schematized

Fig 1. (a) Schematic representation of screw-driven extrusion-based print head reprinted from

Valkenaers et al. [65]. (b) Screw driven Extrusion-based printer of IHBI biofabrication facility used

to fabricate PCL scaffolds. (c) (d) Top and lateral view of a PCL/PBS 44w% scaffold after leaching,

red bars correspond to 1 mm.

(b)

(c) (d)

(a)

P a g e 55 | 78

Fig. 2 To measure dexamethasone and paclitaxel drug loss, 100% ethanol is added to micro-

centrifuge tubes, then vortexed and sonicated for 5min, ethanol is then diluted to 50% with

PBS to avoid a rapid evaporation. For Vancomycin and cefazolin, PBS is directly added to the

microcentrifuge tubes which are then vortexed and sonicated for 5min. Solutions are then col-

lected in UV-vis 96-well plate and measured with spectrophotometry (Bio-Rad, xMark). Dual

wavelength was used to measure drug concentration: Paclitaxel (230-228 nm), Vancomycin

(282-232 nm), Cefazolin (272-254 nm), Dexamethasone (235-241 nm).

Concerning the drug release, release solutions are pipetted in the tubes containing the scaf-

folds and placed in an orbital shaker incubator at 37°C (New Brunswick™ I26, Eppendorf).

Paclitaxel is released in 200µl of a PBS/Tween 20 0.1% mix due to its very poor solubility.

Cefazolin and Vancomycin are released in 200µl of PBS, and Dexamethasone is release in

2ml of PBS to avoid sink condition. The drug release is measured in UV-vis 96-well plates with

spectrophotometry similarly to the drug loss.

2.6 Normalization of drug release data

Three experimental repeats were done for each drug release study. Despite trying to be as

consistent as possible during each repeat, it appeared that the absolute quantity of drug re-

leased between experimental repeats can importantly vary depending on the drug. Those dif-

ferences between experimental repeats could be due to small variations of drug concentration

while preparing the stock solution for drug loading. But it is also possible that, for unknown

Fig 2. (a) Schematic representation of the loading process, (b) the drug loss measurement, (c)

drug release experiment and (d) the drug-scaffold bond characterization.

P a g e 56 | 78

reasons, the drug loading efficiency vary between experimental repeats. Hence, in absence of

precise control over the amount of drug released by the scaffolds no conclusions can be made

from the absolute value. Consequently, it was decided to focus on the release kinetics of each

drug rather than absolute values. To do so, data have been normalized by the overall quantity

of drug released.

2.7 Fourier Transform Infrared Spectroscopy (FTIR) analysis of paclitaxel-

loaded films

An FT-IR Spectroscopy analysis was performed to have a better understanding of the interac-

tion between paclitaxel and PCL. The FTIR analysis was carried out on both scaffolds and

films by nPCL and pPCL 44% with a Nicolet iS50 ATR-FTIR. Both scaffolds and films were

leached for 14 days in 0.01M NaOH before being loaded with paclitaxel, similarly to the drug

release experiments. The FTIR analysis was performed over the region 4000-400cm-1, each

spectrum is the average of 64 scans; results are presented in Fig. 6. The nPCL and pPCL

films used for the FTIR analysis were also observed with a stereomicroscope. Table 1 is based

on literature [58] [59] [60] and summarizes all the vibrational bands which could be potentially

associated with paclitaxel.

2.8 Cell culture

A primary cell line of human osteoblast (hOB) isolated from human's tibia bone were cultured

in flasks maintained in a humidified 5% CO2-containing atmosphere (37°C). The culture me-

dium was changed every 2 to 3 days and consisted of MEM-α with nucleotides and nucleosides

(Gibco, catalog number 12571), supplemented with 10% FBS, penicillin (100 U/mL), strepto-

mycin (100 μg/mL).

2.9 Dexamethasone bioactivity studies

Two cell experiments were carried out to determine the bioactivity of dexamethasone. The first

experiment was performed with free dexamethasone, while the second experiment used dex-

amethasone loaded scaffolds. In both experiment, 10^3 cells per well were seeded in 96 well

plate. In the first experiment, 24h after the cells were seeded, different doses of free dexame-

thasone were directly added to the culture media (from 260µg to 0.25µg). After 24 hours of

treatment, cell viability was assessed by Alamar blue and light microscopy pictures were taken.

In the second experiment, cell viability was assessed when exposed to pPCL 44w% and nPCL

scaffolds loaded with three doses of dexamethasone (150µg/scaffold; 30µg/scaffold; 6µg/scaf-

fold). Unlike drug release studies, the quantity of drug loaded is not normalized by the weight

of the scaffold, as it is important to have the same concentration of drug between samples.

P a g e 57 | 78

Cell viability was then measured by Alamar blue at D1, D3, and D5. Values measured with

Alamar blue were reported to an approximate cell number by using a standard curve.

3. RESULTS AND DISCUSSION

3.1 Rheology

Rheology is a powerful tool to study the viscoelastic behavior of a polymer. In this study, the

rheological analysis was carried out on the different composites to determine how the PBS

microparticles content affects the viscoelastic behavior of PCL. Bear in mind that G’’ and G’

respectively describes the viscous and elastic properties of a material. Tan (δ) corresponds

to the ratio of G’’ to G’, and as a result, depending on if tan (δ) tends to 0 or infinity, the closer

the samples resemble the properties of a pure solid or a pure fluid, respectively. In Fig. 3 & 4,

G’’ always has superior values than G’. This can be explained by the fact that measurements

are carried out at temperatures above the melting point of our polymers. Consequently, sam-

ples are melted and have viscous-dominant (liquid-like) properties, rather than elastic-domi-

nant (solid-like) properties, resulting in G’’ being superior to G’ and Tan (δ) being ≥1.

3.1.1 Amplitude sweep analysis: linear viscoelastic region of nPCL and pPCL

Amplitude sweep measurements were conducted to determine how the PBS microparticles

influence the linear viscoelastic region of neat PCL. Fig. 3 (a) & (b) respectively show the

storage modulus (G') and loss modulus (G") in function of the strain. Similarly to literature a

linear viscoelastic region (LVR) can be observed for neat PCL[61]. While this LVR is relatively

conserved for pPCL17%, this is not the case for PCL/Porogen 33% and 44%. The more the

content of PBS microparticles increases the less PCL shows linear viscoelastic properties. We

can also observe Fig. 3 & 4 that G’’ and G’ values increase with the amount of PCL micropar-

ticles. This trend has also already been observed in previous works with PCL/Silica nanocom-

posites [61]. However, G’ and G’’ are influenced differently depending on the amount of PBS

microparticles and the shear strain. Hence, it is more accurate to interpret the variations of G’’

and G’ together by plotting Tan (δ). First, we can see Fig. 3 (c) that tan (δ) is linear for neat

PCL, which means that G’’ and G’ stay relatively constant independently of the shear strain.

This linear behavior suggests that the internal friction of neat PCL is independent of the strain.

On the other hand, Tan (δ) loses its linear nature with higher content of porogen microparticles,

indicating that G’’ and G’ are influenced differently by the shear strain. Tan (δ) gets closer to 1

at low shear strain, meaning that G’ knows a bigger increase of its value relatively to G’’. Hence,

at low strain and at the same temperature, composites with higher PBS microparticles are

featuring more solid-like properties compared to neat PCL. And the fact that Tan (δ) increases

P a g e 58 | 78

Fig 3. Amplitude sweep tests results. G’, G’’ and Tan (δ) are respectively plotted in (a), (b) and (c) in function

of the shear strain (%) from 0.01% to 150%. Temperature was kept constant at 110°C and angular frequency

at 10 rad/sec. Each point is the average of three experimental repeats (N=3); the bars correspond to SD.

(a)

(c)

(b)

P a g e 59 | 78

with the shear strain, suggests that the properties of PCL/PBS composites get more and more

liquid-like as the strain increases until they finally reach the same state than neat PCL as the

structure starts to break down.

3.1.2 Temperature sweep analysis

In Fig. 4, G", G' and tan(δ) are plotted against the temperature from 125°C to 40°C, while shear

strain is kept constant at 1%. This analysis was carried out to determine how the PBS micro-

particles content influences G", G' or Tan (δ) depending on the temperature. When looking at

temperature sweep results, the first noticeable thing is that G’ and G’’ are linearly decreasing

with higher temperatures. These results can be rationalized by the fact that, as the temperature

gets higher the polymer get progressively closer and closer to a liquid-like state. Hence, vis-

cosity (G’’) and elasticity (G’) diminish as less force is required for the deformation of the ma-

terial. Similarly, to amplitude sweep results, G’ and G’’ have higher values as the amount of

PBS microparticles increases, yet again plotting Tan (δ) is necessary for further interpretation.

Thus, for a constant strain of 1% and between 55°C to 120°C Tan (δ) is lower for composites

with higher PBS microparticles content, which comparably to amplitude sweep results suggest

that those composites feature more solid-like properties. Interestingly, tan (δ) of PCL/PBS 33%

& 44% is linear and less steep than for neat PCL or PCL/PBS 17%, which on the other hand,

seem to be exponentials. These results suggest that neat PCL and PCL/PBS 17% viscoelastic

behaviors get exponentially close to a liquid as temperature rise. On the other hand, higher

PBS microparticles contents seem to make PCL less sensitive to temperature changes; and

at 110°C Tan (δ) of PCL/PBS 44% is almost twice lower than neat PCL. This result is compat-

ible with the fact that higher temperature is needed to print composites with higher microparti-

cles content. Finally, at low temperatures, a G' and G" crossover can be observed Fig. 4. (c).

This crossover means the material starts to behave like a viscoelastic solid and not like a melt

anymore. For this reason, measurements below 50°C are erratic because the material fully

cooled down and that the rheometer is unable to take a proper measurement. Interestingly,

the G'/G" crossover takes place between 55-50°C for the PCL/PBS 33% and 44%, while it

takes place between 50-45°C for neat PCL and PCL/PBS 17%. The temperature sweep anal-

ysis is achieved with decreasing temperatures as mentioned in material and methods. Thus, a

crossover point of G’/G’’ at a higher temperature would suggest that PCL cools down faster

with a high content of PBS microparticles. This behavior could be rationalized by the fact that

PBS microparticles reduce PCL particles freedom of movement which consequently cools

down faster.

P a g e 60 | 78

Fig 4. Temperature sweep results. G’, G’’ and Tan (δ) are respectively plotted in (a), (b) and (c) in

function of the temperature from 125°C to 35°C. Strain was kept constant at 1% and angular frequency

at 10 rad/sec. Each point is the average of three experimental repeats (N=3); the bars correspond to SD.

(c)

(b)

(a)

P a g e 61 | 78

3.2 Paclitaxel FTIR and Stereomicroscopy

The infrared spectrum of a compound is formed by the superposition of the absorption bands

of its specific functional groups and as such can be used as a fingerprint. Measurements were

first taken on porous paclitaxel-loaded scaffolds to be as close as possible of the experimental

conditions of the drug release experiment. However, results weren’t reproducible, which could

be linked to the architecture of scaffolds reducing the contact surface with the FTIR probe.

When using films rather than scaffolds, no paclitaxel signal could be detected on porous scaf-

folds either, even though the deposition of paclitaxel was confirmed by a drug loss measure-

ment (data not shown). We can see on stereomicroscope results Fig. 5 (c) & (d) that no dif-

ferences can be observed between the paclitaxel-loaded and the non-loaded porous film. On

the other hand, the deposition of paclitaxel is easily observable on non-porous film Fig. 5 (c)

& (d). This absence of a signal on porous films could be due to the paclitaxel being deposited

more homogeneously in the pores of the scaffold, resulting in a low drug signal on the surface

of the film. Paclitaxel signal was detectable on non-porous films only when the FTIR probe was

placed on the white drug clusters Fig. 5 (b). Consequently, the results presented are coming

from measurements taken on those white clusters of paclitaxel present on non-porous films.

The spectra presented in Fig 6. are discussed as follow. The spectra of PCL Fig. 6 (a) is

similar to literature [62] [63]. Thus, we can observe the peaks commonly identified such as the

methyl/methylene C-H saturated aliphatic groups asymmetric and symmetric stretching re-

spectively found at 2943 cm-1 and at 2864 cm-1. The strong absorption peak at 1720 cm-1 is

Fig 5. Stereomicroscopy pictures of non-porous (a) (b) films and porous (c) (d) films before (a) (c) and

after (b) (d) loading with paclitaxel. The red bars correspond to 1 mm.

(c)

(a) (b)

(d)

P a g e 62 | 78

assigned to the stretching vibration of the carbonyl compounds -C=O, and the signal at

1470.92 cm-1 seems to correspond to the bend of the methyl/methylene C-H saturated ali-

phatic. Finally, the signal at 1162 cm-1 and 730 cm-1 can be respectively assigned to the alkyl-

substituted ether -C-O-C- stretching and the rocking motion of the methylene. Fig. 6 (e) corre-

spond to the Paclitaxel spectra and is also comparable to what can be found in the literature

[59] [60]. Nonetheless, some bands can be attributed to different functional groups and differ-

ent interpretations were given. The wide vibrational band spreading from 3600cm-1 to 3200cm-

1 correspond to different functional groups such as the O-H hydroxyl stretching and the N-H

amine stretching. Some peaks in the 3130 cm-1 – 3070 cm-1 range could correspond to the C-

H aromatic stretching. Comparably to PCL, the two peaks at 2962 cm-1 – 2888 cm-1 are re-

spectively associated with the saturated aliphatic CH asymmetric and symmetric stretching

vibrations, while the peaks found at 1730 cm-1 and 1702 cm-1 are associated to the -C=O car-

bonyl groups (ester, ketone). The band at 1635 cm-1 is assigned to the amide C=O-N bound.

Literature has previously assigned the 1602 cm-1 band specifically to the aromatic C=C stretch-

ing vibration [60], but the next peaks at 1578 cm-1 and 1533 cm-1 are harder to identify and can

correspond to the amine N-H bends but also to the C=C-C aromatic stretching. The band at

1491 cm-1 is usually assigned to the bend of the methyl/methylene C-H saturated aliphatic

bend like for PCL. The signal at 1273 cm-1 has been previously attributed to the amine C-N

stretching [60] [59] and the strong band at 1241 cm-1 to the C-O ester bond stretching vibration

[60], which could be compatible with the peak found at 1251 cm-1 in the current work. Several

Table 1. Vibrational band assignments of Paclitaxel drug

P a g e 63 | 78

Figure 6. FTIR spectra of PCL (a) Paclitaxel (e) and non-porous PCL films loaded

with low dose (b) medium dose (c) and high dose (d) of paclitaxel

(a)

(b)

(c)

(d)

(e)

P a g e 64 | 78

functional groups can correspond to the band found at 1072 cm-1 and different interpretations

were given in the literature. Thus, it was assigned to the secondary alcohol C-O stretching [59]

or to the C-H out of plane deformation bands [60]. But the peak might be more likely to corre-

sponds to the secondary alcohol C-O stretching which can be assigned to a wide range of

absorption around 1150 cm-1 and 1000 cm-1 [58], while the C-H out of plane deformation

bands are assigned to the range of 900 – 670 cm-1 in general [58]. Nonetheless, it could cor-

respond to the C-H aromatic in-plane bend which is found in the range of 1225 cm1 to 950 cm-

1[58], and that can be assigned to several peaks, such as the band at 981 cm-1. Finally, the

band at 707 cm-1 and the surroundings ones are believed to correspond to the C-H aromatic

out of plane bend [58] [60], but could also possibly correspond to the O-H alcohol out-of-plane

bend[58] The peaks 707 cm-1/981 cm-1/1635 cm-1/1578 cm-1/1533 cm-1 will be specifically

kept as fingerprint of paclitaxel for our work. Fig 6. (b), (c) & (d) correspond to non-porous

PCL films respectively loaded with a low, medium, and a high dose of paclitaxel. The first thing

to notice is that no paclitaxel signal is detected for the low dose as the spectra is identical to

the one of PCL. On the other hand, we can clearly see that the bands associated with paclitaxel

and circled in red are becoming sharper as the dose increase. However, the bands visible in

Fig 6. (b), (c) & (d) are at the same positions than on the spectra of PCL and paclitaxel ob-

served separately. This absence of a shift is generally associated with an absence of

interaction between the two materials as no functional groups bonded between paclitaxel and

PCL.

3.3 Drug release

3.3.1 Porous and non-porous scaffolds

As mentioned in material and methods, all the scaffolds used for drug release experiments are

first leached for 14 days and then cut by hand with a scalpel to obtain small scaffolds of ap-

proximately 3mmx3mmx3mm. The quantity of drug loaded in each release study is normalized

Figure 7. Bar diagram representing the difference of weight between porous and non-

porous scaffolds of similar size after leaching. The bars correspond to SD (n=146)

P a g e 65 | 78

by the weight of each scaffold. For this reason, each scaffold is weight before the sterilization

and loading process. If more specific techniques are required to study the porosity of porous

scaffolds in details, we can already see Fig. 7 the significant difference in weight between

porous and non-porous scaffolds of the same size. Thus, porous scaffolds and non-porous

scaffolds respectively weight 5.6 mg and 9mg in average. Furthermore, it has been observed

during experimentation that porous scaffolds are floating in PBS

3.3.2 Drug loss and loading efficiency

The drug loss and loading efficiency were measured and calculated for all drug release exper-

iments as explained in material and methods. Unfortunately, the results can’t be shown as they

weren’t reproducible. It is not clear yet if the lack of reproducibility is simply due to variations

between experimental repeats or due to the experimental protocol. Thus, the experimental

protocol didn’t seem to work for cefazolin experiments. On the other hand, drug loss results

measured for vancomycin experiments seemed to match with the quantity of drug released by

the scaffolds. Yet again, the loading efficiency of vancomycin calculated for each experimental

repeat was very different. For dexamethasone, depending on the experiment repeat drug loss

results were either matching relatively well the drug release results or were incoherent. Finally,

paclitaxel drug loss results are explained in the next section.

3.3.3 Paclitaxel drug release

Fig. 8 corresponds to one of the three experimental repeats of the paclitaxel drug release

studies. This figure is here to draw attention to the amount of drug released by porous scaffolds

compared to non-porous scaffolds. If different amounts of paclitaxel were released depending

Figure 8. Paclitaxel cumulative release results normalised by the weight of scaffolds. High dose = 10

µg/mg scaffold. Medium dose = 2 µg/mg. Low dose = 0.4 µg/mg. The bars correspond to SD (N=1 / n=6)

P a g e 66 | 78

on the experimental repeat, porous scaffolds released significantly more drug than non-porous

scaffolds in all experimental repeats. When looking at the PCL films on Fig.5, we can observe

that the deposition of paclitaxel seems to be different on the porous and non-porous surface.

Thus, the deposition of paclitaxel on non-porous films is very heterogeneous as we can see

on Fig.5(b), therefore, this first result correlates with the idea that a bigger amount of paclitaxel

could be loaded on porous scaffolds, as observed on Fig.8, thanks to the surface porosity that

seems to allow a better paclitaxel deposition. Moreover, as paclitaxel is deposited in the scaf-

folds pores, the contact surface with the release medium is probably lower than on the non-

porous surface. This hypothesis would imply a faster release of paclitaxel for non-porous scaf-

folds onto which paclitaxel seems to form heterogeneous crystalline patches, yet, as explained

below, no real differences were found between the kinetics of porous and non-porous scaf-

folds. It is important to notice that only a very little amount of paclitaxel has been released

compared to the amount loaded. Fig. 8 we can notice that after 500 hours, porous scaffolds

loaded with high doses of paclitaxel (10µg/mg of scaffolds) released only 1.7µg/mg of the

scaffold, which is only equivalent to 17% of the initial amount of paclitaxel-loaded. Neverthe-

less, the amount of paclitaxel lost in the tube during loading for this experimental repeat was

corresponding to only 26% of the initial amount loaded in average. Suggesting that 57% of the

paclitaxel initially loaded is missing. The same phenomenon happened in the two other exper-

imental repeats. After sonication, a signal equivalent to 5% of the initial amount of paclitaxel-

loaded was measured for both porous and non-porous scaffolds, suggesting that a considera-

ble amount of paclitaxel could be potentially remaining on the surface of both porous and non-

scaffolds after 500 hours. The fact that a considerable amount of paclitaxel is not released

could be explained, either by an interaction between paclitaxel and PCL, or by paclitaxel being

too poorly soluble in PBS/tween 0.1%. If FTIR results aren’t sufficient to disprove entirely the

absence of interaction between PCL and Paclitaxel, it might be more likely that the release

medium is the source of the issue. When normalized by the total amount of drug released (Fig.

9), it is difficult to say if there is a difference between the release kinetics of porous and non-

porous scaffolds. It seems that porous scaffold loaded with a medium dose (Fig 9. b) could

have a very slightly slower release compared to non-porous scaffolds, but this difference is too

small to make any conclusion. Moreover, the results obtained for low dose and high dose are

not suggesting any differences between porous and non-porous scaffolds release (Fig 9. a,

c). The burst release is quite acceptable as after the first 24h only 20% to 30% of the drug is

released. Paclitaxel is then released over the course of approximately 20 days, which is a

significant amount of time, but we can see that after 200 hours the paclitaxel release signifi-

cantly slow down as 80% of the drug content has been released. The fact that release experi-

ments were stopped after 500h for practical reasons also need to be taken into account as a

P a g e 67 | 78

Figure 9. Paclitaxel cumulative release results normalised by the total amount of drug released.

High dose = 10 µg/mg scaffold. Medium dose = 2 µg/mg. Low dose = 0.4 µg/mg. Each point is the

average of 6 replicates (N=3).

(a)

(b)

(c)

P a g e 68 | 78

small amount of paclitaxel was still being released, especially for samples loaded with high

doses. Therefore, technically the release of paclitaxel lasts longer than 500 hours.

3.3.3 Dexamethasone: drug release

When normalized by the total amount of dexamethasone released (Fig 10.), no differences

can be observed in the release kinetic between porous and non-porous scaffolds. Unlike

paclitaxel, the burst release in the first 24h is more important and seems to be dose dependent.

Therefore, after 24h scaffolds loaded with a low dose of dexamethasone released 30% to 65%

of their dexamethasone, while scaffolds loaded with a high dose released 70% to 75%. There-

fore, it seems that the burst release of dexamethasone in the first 24h gets more important as

the dose loaded on the scaffold increase. Finally, most of the dexamethasone is getting re-

leased over the first 4 days. Thus, scaffolds loaded with a low dose of dexamethasone released

80% of their content, while scaffolds loaded with a high dose released 95% of

3.3.4 Dexamethasone: bioactivity

The cell viability of hOB exposed to different doses of free dexamethasone for 24h is presented

in Fig 11. A clear drop in cell viability can be observed between 325 µg/ml and 650 µg/ml of

dexamethasone, statistical analysis revealed that this drop is statistically significant. The mi-

croscopy pictures (data not shown) also shown that in addition to having fewer cells, a signifi-

cant amount have a round morphology when exposed to high dose of dexamethasone. Below

325 µg/ml and above 2.5 µg/ml of dexamethasone, it is difficult to make assumptions about

any effect of dexamethasone, as cell viability remain relatively similar to the cell control. No

differences can be observed either on the light microscopy pictures below 325 µg/ml. The

concentration of 2.5 µg/ml 1 .3 µg/ml seems to slightly increase the cell viability, but no statis-

tically significant differences were found. However, literature tends to suggest that dexame-

thasone is still toxic at a concentration of 1.3 µg/ml[64], consequently, no conclusion can be

made. Fig 12. we can see the cell viability of hOB while exposed to porous and non-porous

scaffolds loaded with three doses of dexamethasone (150µg/scaffold, 30 µg/scaffold, 6

µg/scaffold) over 5 days. After 24h of treatment, it appears that hOB cultured with scaffolds

loaded with high doses of dexamethasone have statistically significantly lower cell viabilities,

while the other conditions are comparable to controls. At day 3, the cell viability of hOB cultured

with dexamethasone-loaded scaffolds have increased compared to day 1 and are statistically

significantly higher than the controls. Interestingly, if no differences were found for porous scaf-

folds; the non-porous scaffolds loaded with a medium and low dose of dexamethasone have

a significantly higher cell viability than the one loaded with a high dose. On the other hand, cell

viability of the controls is lower than at day 1. At day 5, the cell viability of hOB cultured with

P a g e 69 | 78

Figure 10. Dexamethasone cumulative release results normalised by the total amount of drug

released. High dose = 25 µg/mg scaffold. Medium dose = 5 µg/mg. Low dose = 1 µg/mg. Each

point is the average of 6 replicates (N=3).

(a)

(b)

(c)

P a g e 70 | 78

loaded scaffolds is lower than at day 3 but remain higher than at day 1. The control conditions

have similar cell viability than at day 3 and remain statistically significantly lower than scaffolds

loaded with dexamethasone. It is interesting to notice that the standard error means of control

conditions are considerable comparably to other conditions. Compared to day 3, non-porous

scaffolds loaded with medium and low doses-maintained similar cell viabilities, while other

conditions seem to have slightly lower values. Moreover, non-porous scaffolds loaded with

medium and low doses have a statistically significantly higher cell viability than for high doses;

again, no differences were found for porous scaffolds. Consequently, the cytotoxicity associ-

ated with the high dose of dexamethasone suggest the dexamethasone released from the

scaffold is still bioactive. Passed day 1, results suggest that scaffolds loaded with dexame-

thasone increase cell viability similarly to the previous assay. Moreover, for unknown reasons,

Figure 11. Human osteoblasts cell viability assay measured by alamar blue after incubation of

24h with free dexamethasone. Bars correspond to SEM (N=1; n=8). A T-test analysis was car-

ried out to verify the statistical difference of the mean values (**=p < 0.01).

**

**

Figure 12. Human osteoblasts cell viability assay measured by alamar blue at day 1, 3 and 5 when

exposed to scaffolds loaded with dexamethasone (High dose = 150µg/scaffold; Medium =

30µg/scaffold; Low dose = 6µg/scaffold). Bars correspond to SEM (N=3; n=4) A one-way analysis

of variance was carried out to verify the statistical difference of the mean values (**=p < 0.01).

P a g e 71 | 78

Figure 13. Cefazolin cumulative release results normalised by the total amount of drug released.

High dose = 10 µg/mg scaffold. Medium dose = 2 µg/mg. Low dose = 0.4 µg/mg. Each point is the

average of 6 replicates (N=2).

(a)

(b)

(c)

P a g e 72 | 78

this phenomenon seems to be more important for cell exposed to low dose of dexamethasone

on non-porous scaffolds. Yet again literature has shown a decrease of cell viability and an

increase of cell apoptosis of hOB when exposed to a concentration of 10-6M [64]. Thus, it is

difficult to draw a conclusion on the ability of dexamethasone to increase cell viability. One

possible explanation could be that, passed the burst release of day 1, very small amount of

dexamethasone below the concentration of 10-6M are released which then stimulate cell

growth. Finally, it can be noticed that the NaOH leaching process applied to porous and non-

porous scaffolds doesn’t seem to have a negative impact on cell viability, as no statistically

significant differences could be found among the controls.

3.3.5 Cefazolin drug release

Unlike paclitaxel and dexamethasone, when looking at the normalized results Fig 13. we can

clearly see a different release kinetic between porous and non-porous scaffolds. Thus, porous

scaffolds release cefazolin more gradually over more than 100h, while non-porous scaffolds

release all the cefazolin in a burst during the first three hours. Like for dexamethasone, the

burst release of porous scaffolds in the first three hours seems to increase with the dose of

cefazolin. We can also observe that for both porous and non-porous scaffolds, cefazolin seems

to be released faster as the dose increases. Consequently, after 24h, low dose porous and

non-porous scaffolds respectively released 68% and 89% while high dose porous and non-

porous scaffolds respectively released 85% and 99% of the total amount of cefazolin. For un-

known reasons, the experimental protocol to measure drug loss is not working for cefazolin,

and very small values of drug loss are obtained while a significant drug loss is expected as

only 40% of the cefazolin initially loaded was released on average. Finally, those first results

are promising as porous scaffolds seem to be able to release cefazolin over a longer span of

time. However, it is important to take into account the fact that the burst effect and release

kinetics get respectively higher and faster as the dose of cefazolin increases. Consequently, it

could be interesting to see if the difference of release kinetic between porous and non-porous

scaffolds still exist at higher doses.

3.3.6 Vancomycin drug release

We can see Fig 14. results of the vancomycin release experiments normalized by the total

amount of vancomycin released. Even though the difference is not as considerable as the one

observed previously for cefazolin, porous scaffolds are clearly releasing vancomycin slightly

slower than non-porous scaffolds. Interestingly, this difference is hardly visible for low dose

samples. Results of low dose samples are often more spread as the measured values are very

small compared to high doses and are more influenced by measurement variation and errors

P a g e 73 | 78

Figure 14. Vancomycin cumulative release results normalised by the total amount of drug re-

leased. High dose = 10 µg/mg scaffold. Medium dose = 2 µg/mg. Low dose = 0.4 µg/mg. Each

point is the average of 6 replicates (N=3).

(a)

(b)

(c)

P a g e 74 | 78

as such. Unfortunately, the burst release of vancomycin is way more important than for

cefazolin, and most of the vancomycin present on the scaffold surface is released in PBS in

the first three hours for both type of scaffold. It is interesting to notice that similarly to cefazolin

and dexamethasone results, the burst release of vancomycin increases with the dose inde-

pendently of the scaffold type. Therefore, 87% of the vancomycin is released by porous scaf-

fold loaded with high dose in the first hour, while porous scaffold loaded with a low dose re-

leased only 67% of the vancomycin.

3.3.7 Final discussion

To get more insight into the factors influencing the drug release studies, it is important to dis-

cuss the results all together besides analyzing them separately. One main question arises from

this work; why the release kinetics is different between porous and non-porous scaffolds only

for cefazolin and vancomycin? In an attempt to explain the results, it is logical to look first at

the properties of the drug. Cefazolin and vancomycin have in common a high solubility in water,

on contrary to paclitaxel and dexamethasone which both have a very poor solubility in water.

This difference of solubility in water is the first element of the answer, even if it doesn’t explain

by itself how it influences the release kinetics of the drug between a porous and non-porous

scaffold. As mentioned earlier for Paclitaxel, it would be interesting to investigate if a different

release kinetics is observed with a release media in which paclitaxel and dexamethasone are

more soluble. Secondly, it appeared during experimentation that both porous and non-porous

scaffolds are sinking when placed in ethanol during drug loading. Yet, non-porous scaffolds

sink while porous scaffolds float when immersed in PBS or in cell media. This phenomenon

could be explained by the lower surface tension of ethanol combined with the hydrophobicity

of PCL. Thus, small air bubbles could be entrapped in the micropores of porous scaffolds,

making the scaffolds float in PBS but sink in liquid with lower surface tension such as ethanol.

The presence of air bubbles on the scaffold surface could also reduce the surface contact

between the release medium and the scaffold, which as a result would reduce the amount of

drug released. Hence, drug release results might be different if porous scaffolds were fully

immersed in the release media. Further experimentations could also help us have a better

understanding of what’s happening. Therefore, it would be important to extend the FTIR anal-

ysis to cefazolin, vancomycin, and dexamethasone as the difference between the results could

also due to a possible interaction between the drug and the scaffold. Indeed, depending on the

type of bonding created between the drug and the scaffold after loading, the release of the

drug could be hindered if immersion in the release media is not sufficient to detach the drug

from the scaffold. Moreover, it could also be interesting to analyze the homogeneity of each

drug deposition on the surface of porous and non-porous scaffolds as the release will directly

depend on the way each drug adheres to the surface. It was clear from Fig.5 that deposition

P a g e 75 | 78

of paclitaxel on non-porous films was not heterogeneous while no conclusion could be made

for the porous film as the limitations of light microscopy, notably the light reflection of PCL,

don’t allow us to visualize drug deposition in the scaffold pores. However, more advanced

imaging technologies such as scanning electronic microscopy might make possible the obser-

vation of the drug deposition in the pores.

4. CONCLUSION AND FUTURE WORK

First, this work investigated the printability of a novel PCL/PBS composite. Rheological meas-

urements have shown that as the mass ratio of PBS microparticles increase PCL is losing its

linear viscoelastic region, in addition, to being less sensitive to temperature increase. The

PCL/PBS composite was then successfully printed into scaffolds with a screw-based extrusion

3D printing technique by increasing the manufacturing temperature. Secondly, this work inves-

tigated the potential of PCL scaffolds as a drug carrier, and more precisely how the scaffold

microporosity is influencing the release kinetics of different drugs. Our results have shown that

porosity seems to decrease the burst effect and prolong the drug release of soluble drugs

compared to non-porous scaffolds. On the other hand, no significant differences could be ob-

served for scaffolds loaded with insoluble drugs. But despite some encouraging results, the

burst release remains important for soluble drugs, and as such, their clinical relevance remains

limited. Concerning the influence of the loading process, results seem to indicate that it is not

affecting the drug bioactivity since a high concentration of dexamethasone has shown a nox-

ious effect on cell viability. Obviously, further work is required to verify the drug bioactivity of

Paclitaxel, cefazolin, and vancomycin. Finally, results of drug release studies have also shown

a lack of control over the drug loading process. Indeed, the total amount of drug released

between experimental repeats was considerably different. As such, there is a strong need to

find an assay that would allow us to measure accurately the drug loss during the loading pro-

cess. Without this data, it is difficult to draw a conclusion as we’re unable to know precisely

how much drug has been released comparatively to the total amount loaded on each scaffold,

which would maybe show different release kinetics. To follow up this work two experiments

could be carried out to shed more light on the results. Firstly, HPLC could maybe be used to

measure the amount of drug left in the tube after loading; secondly, the scaffolds could maybe

be degraded in a solution which doesn’t affect the drug to measure how much drug is left on

the scaffolds at the end of the drug release study. Thus, more work is required to determine if

experimental conditions are optimal to get more insight into how porosity exactly affects drug

loading and drug release.

P a g e 76 | 78

REFERENCES

[1] Y. H. Yun, B. K. Lee, and K. Park, “Controlled Drug Delivery: Historical perspective for the next generation,” J. Control. Release, vol. 219, pp. 2–7, 2015.

[2] K. Park, “Facing the Truth alabout Nanotechnology in Drug Delivery,” no. 9, pp. 7442–7447, 2013.

[3] P. Wu and D. W. G. Ã, “Drug / device combinations for local drug therapies and infection prophylaxis,” vol. 27, pp. 2450–2467, 2006.

[4] F. Yamashita and M. Hashida, “Pharmacokinetic considerations for targeted drug delivery,” Adv. Drug Deliv. Rev., vol. 65, no. 1, pp. 139–147, 2013.

[5] M. S. Surapaneni, S. K. Das, and N. G. Das, “Designing Paclitaxel drug delivery systems aimed at improved patient outcomes: current status and challenges.,” ISRN Pharmacol., vol. 2012, p. 623139, 2012.

[6] K. T. Savjani, A. K. Gajjar, and J. K. Savjani, “Drug Solubility : Importance and Enhancement Techniques,” vol. 2012, no. 100 mL, 2012.

[7] J. Wolinsky, Y. Colson, and M. Grinstaff, “Local Drug Delivery Strategies for Cancer Treatment: Gels, Nanoparticles, Polymeric Films, Rods, and Wafers Public Access,” J. Control. Release, vol. 159, no. 1, pp. 1–34, 2012.

[8] E. Senkus and J. Jassem, “Cardiovascular effects of systemic cancer treatment,” Cancer Treat. Rev., vol. 37, no. 4, pp. 300–311, 2011.

[9] Collins, Jerry M. "Pharmacologic rationale for regional drug delivery." Journal of Clinical Oncology 2.5 (1984): 498-504.

[10] Ensminger, William D., and John W. Gyves. "Regional cancer chemotherapy." Cancer treatment reports 68.1 (1984): 101-115 [11] R. LANGER, “Implantable controlled release systems,” Pharmac. Ther, vol. 21, no. 2, p. 7258, 1983.

[12] K. Krukiewicz and J. K. Zak, “Biomaterial-based regional chemotherapy : Local anticancer drug delivery to enhance chemotherapy and minimize its side-effects,” Mater. Sci. Eng. C, vol. 62, pp. 927–942, 2016.

[13] FDA, “TAXOL® (paclitaxel) INJECTION,” pp. 1–53.

[14] L. T. O. The, “Weekly paclitaxel in ovarian cancer — the latest success story,” pp. 16–17, 2011.

[15] E. Perez, “Paclitaxel in Breast Cancer.,” Oncologist, vol. 3, no. 6, pp. 373–389, 1998.

[16] X. Pivot, L. Asmar, and G. N. Hortobagyi, “The efficacy of chemotherapy with docetaxel and paclitaxel in anthracycline-resistant breast cancer (Review).,” Int. J. Oncol., vol. 15, no. 2, pp. 381–386, 1999.

[17] N. I. Marupudi, J. E. Han, K. W. Li, V. M. Renard, B. M. Tyler, H. Brem, N. I. Marupudi, J. E. Han, K. W. Li, V. M. Renard, and B. M. Tyler, “Paclitaxel : a review of adverse toxicities and novel delivery strategies Paclitaxe,” vol. 0338, no. August, 2017.

[18] J. M. Meerum Terwogt, B. Nuijen, W. W. Ten Bokkel Huinink, and J. H. Beijnen, “Alternative formulations of paclitaxel,” Cancer Treat. Rev., vol. 23, no. 2, pp. 87–95, 1997.

[19] R. Panchagnula, “Pharmaceutical aspects of paclitaxel,” Int. J. Pharm., vol. 172, no. 1–2, pp. 1–15, 1998.

[20] A. B. Dhanikula and R. Panchagnula, “Localized paclitaxel delivery,” Int. J. Pharm., vol. 183, no. 2, pp. 85–100, 1999.

[21] T. Manabe, H. Okino, R. Maeyama, K. Mizumoto, M. Tanaka, and T. Matsuda, “New Infusion Device for Trans-tissue , Sustained Local Delivery of Anticancer Agent to Surgically Resected Tissue : Potential Use for Suppression of Local Recurrence of Pancreatic Cancer,” pp. 203–207, 2004.

[22] D. R. Rittenhouse, S. M. Shortell, and E. S. Fisher, “Weekly Paclitaxel in the Adjuvant Treatment of Breast Cancer,” N. Engl. J. Med., pp. 2301–2303, 2009.

[23] S. M. Beecher, D. P. O. Leary, R. Mclaughlin, K. J. Sweeney, and M. J. Kerin, “Influence of complications following immediate breast reconstruction on breast cancer recurrence rates,” pp. 391–398, 2016.

[24] D. E. Reichman and J. A. Greenberg, “Reducing Surgical Site Infections : A Review,” vol. 2, no. 4, pp. 212–221, 2009.

[25] C. D. Owens and K. Stoessel, “Surgical site infections : epidemiology , microbiology and prevention,” J. Hosp. Infect., vol. 70, pp. 3–10, 2008.

P a g e 77 | 78

[26] M. Gimeno, P. Pinczowski, M. Pérez, A. Giorello, M. Ángel, J. Santamaría, M. Arruebo, and L. Luján, “European Journal of Pharmaceutics and Biopharmaceutics A controlled antibiotic release system to prevent orthopedic-implant associated infections : An in vitro study,” Eur. J. Pharm. Biopharm., vol. 96, pp. 264–271, 2015.

[27] K. A. Poelstra, N. A. Barekzi, A. M. Rediske, A. G. Felts, J. B. Slunt, and D. W. Grainger, “Prophylactic treatment of gram-positive and gram-negative abdominal implant infections using locally delivered polyclonal antibodies,” 2001.

[28] O. Access, “Clinical Medicine : Therapeutics Safety and Efficacy of Cefazolin Sodium in the Management of Bacterial Infection and in Surgical Prophylaxis,” pp. 1607–1615.

[29] D. W. Bratzler, P. M. Houck, and S. Infection, “Antimicrobial prophylaxis for surgery : an advisory statement from the National Surgical Infection Prevention Project,” vol. 189, pp. 395–404, 2005.

[30] P. V Giannoudis, J. Parker, and M. H. Wilcox, “Methicillin-resistant Staphylococcus aureus in trauma and orthopaedic practice,” pp. 749–754, 2005.

[31] E. M. Hetrick, M. H. Schoenfisch, E. M. Hetrick, and E. M. Hetrick, “Reducing implant-related infections : active release strategies,” pp. 780–789, 2006.

[32] Z. Zhang, J. Tang, H. Wang, Q. Xia, S. Xu, and C. C. Han, “Controlled Antibiotics Release System through Simple Blended Electrospun Fibers for Sustained Antibacterial E ff ects,” 2015.

[33] V. P. Mantripragada, A. C. Jayasuriya, and E. Program, “Effect of dual delivery of antibiotics (vancomycin and cefazolin) and BMP-7 from chitosan microparticles on Staphylococcus epidermidis and pre-osteoblasts in vitro,” pp. 1–22, 2016.

[34] H. Daniell, “FOREIGN BODY REACTION TO BIOMATERIALS James,” James M. Anderson, Anal. Rodriguez, David T. Chang, vol. 76, no. October 2009, pp. 211–220, 2012.

[35] J. M. Anderson, A. Rodriguez, and D. T. Chang, “Foreign body reaction to biomaterials,” vol. 20, pp. 86–100, 2008.

[36] D. F. Williams, “Biomaterials On the mechanisms of biocompatibility ” vol. 29, pp. 2941–2953, 2008.

[37] J. M. Morais, F. Papadimitrakopoulos, and D. J. Burgess, “Biomaterials / Tissue Interactions : Possible Solutions to Overcome Foreign Body Response,” vol. 12, no. 2, pp. 188–196, 2010.

[38] T. T. Dang, K. M. Bratlie, S. R. Bogatyrev, X. Y. Chen, R. Langer, and D. G. Anderson, “Biomaterials Spatiotemporal effects of a controlled-release anti-in fl ammatory drug on the cellular dynamics of host response,” Biomaterials, vol. 32, no. 19, pp. 4464–4470, 2011.

[39] B. S. Zolnik and D. J. Burgess, “Evaluation of in vivo – in vitro release of dexamethasone from PLGA microspheres,” vol. 127, pp. 137–145, 2008.

[40] D. Kim and D. C. Martin, “Sustained release of dexamethasone from hydrophilic matrices using PLGA nanoparticles for neural drug delivery,” vol. 27, pp. 3031–3037, 2006.

[41] F. Langenbach and J. Handschel, “Effects of dexamethasone , ascorbic acid and β -glycerophosphate on the osteogenic differentiation of stem cells in vitro,” Stem Cell Res. Ther., p. 1, 2013.

[42] O. Ghali, O. Broux, G. Falgayrac, N. Haren, J. P. T. M. Van Leeuwen, G. Penel, P. Hardouin, and C. Chauveau, “Dexamethasone in osteogenic medium strongly induces adipocyte differentiation of mouse bone marrow stromal cells and increases osteoblast differentiation,” pp. 1–15, 2015.

[43] A. El-fiqi, J. Kim, and H. Kim, “Osteoinductive Fibrous Sca ff olds of Biopolymer / Mesoporous Bioactive Glass Nanocarriers with Excellent Bioactivity and Long- Term Delivery of Osteogenic Drug,” 2015.

[44] A. K. Dash and G. C. C. Ii, “Therapeutic Applications of Implantable Drug Delivery Systems,” vol. 12, no. 1998, pp. 1–12, 1999.

[45] R. De Souza, P. Zahedi, C. J. Allen, and M. Piquette-miller, “Polymeric drug delivery systems for localized cancer chemotherapy Polymeric drug delivery systems for localized cancer chemotherapy,” vol. 7544, no. August, 2017.

[46] M. Sokolsky-papkov, K. Agashi, A. Olaye, K. Shakesheff, and A. J. Domb, “Polymer carriers for drug

delivery in tissue engineering ☆,” vol. 59, pp. 187–206, 2007.

[47] J. A. Hubbell, “Synthetic biodegradable polymers for tissue engineering and drug delivery,” pp. 246–251.

[48] T. J. Sill and H. A. Von Recum, “Electrospinning : Applications in drug delivery and tissue engineering,” no. June 2008, 2016.

P a g e 78 | 78

[49] G. Jonathan and A. Karim, “3D printing in pharmaceutics: A new tool for designing customized drug delivery systems,” Int. J. Pharm., vol. 499, no. 1–2, pp. 376–394, 2016.

[50] T. Garg, O. Singh, S. Arora, and M. R.S.R, “Scaffold: A Novel Carrier for Cell and Drug Delivery,” Crit. Rev. Ther. Drug Carr. Syst., vol. 29, no. 1, pp. 1–63, 2012.

[51] M. A. Woodruff and D. W. Hutmacher, “The return of a forgotten polymer - Polycaprolactone in the 21st century,” Prog. Polym. Sci., vol. 35, no. 10, pp. 1217–1256, 2010.

[52] D. Mondal, M. Griffith, and S. S. Venkatraman, “Polycaprolactone-based biomaterials for tissue engineering and drug delivery : Current scenario and challenges,” GPOM, vol. 65, no. 5, pp. 255–265, 2016.

[53] B. Rai, S. H. Teoh, D. W. Hutmacher, T. Cao, and K. H. Ho, “Novel PCL-based honeycomb scaffolds as drug delivery systems for rhBMP-2,” Biomaterials, vol. 26, no. 17, pp. 3739–3748, 2005.

[54] M. S. Kamath, S. S. S. J. Ahmed, M. Dhanasekaran, and S. Winkins Santosh, “Polycaprolactone scaffold engineered for sustained release of resveratrol: Therapeutic enhancement in bone tissue engineering,” Int. J. Nanomedicine, vol. 9, no. 1, pp. 183–195, 2013.

[55] D. Mondal, M. Griffith, and S. S. Venkatraman, “Current scenario and challenges,” GPOM, vol. 65, no. 5, pp. 255–265, 2016.

[56] B. Dorj, J. Won, O. Purevdorj, K. D. Patel, and J. Kim, “A novel therapeutic design of microporous-structured biopolymer scaffolds for drug loading and delivery,” Acta Biomater., vol. 10, no. 3, pp. 1238–1250, 2014.

[57] K. J. Rambhia and P. X. Ma, “Controlled drug release for tissue engineering,” J. Control. Release, vol. 219, pp. 119–128, 2015.

[58] J. Coates, R. A. M. Ed, and J. Coates, “Interpretation of Infrared Spectra , A Practical Approach Interpretation of Infrared Spectra , A Practical Approach,” pp. 10815–10837, 2000.

[59] T. S. R. Devi and S. Gayathri, “FTIR AND FT-RAMAN SPECTRAL ANALYSIS OF PACLITAXEL DRUGS,” vol. 2, no. 2, 2010.

[60] S. A. Garea and A. D. I. Ghebaur, “FT-IR Spectroscopy and Thermogravimetrical Characterization of Prodrugs Based on Different Dendritic Polymers and Antitumoral Drug,” pp. 1–4, 1992.

[61] Y. Li, C. Han, X. Zhang, J. Bian, and L. Han, “Rheology , Mechanical Properties , and Biodegradation of Poly ( e -caprolactone )/ Silica Nanocomposites,” 2013.

[62] B. Abderrahim, E. Abderrahman, A. Mohamed, T. Fatima, and T. Abdesselam, “Kinetic Thermal Degradation of Cellulose , Polybutylene Succinate and a Green Composite : Comparative Study,” vol. 3, no. 4, pp. 95–110, 2015.

[63] A. R. Hernández, O. C. Contreras, and J. C. Acevedo, “Poly ( ε -caprolactone ) Degradation Under Acidic and Alkaline Conditions,” vol. 3, no. 4, pp. 70–75, 2013.

[64] M. A. Naves, R. M. R. Pereira, and N. Comodo, “Effect of dexamethasone on human osteoblasts in culture : involvement of b 1 integrin and integrin-linked kinase,” vol. 35, pp. 1147–1151, 2011.

[65] H. Valkenaers, F. Vogeler, A. Voet, and J. Kruth, “Screw extrusion based 3D printing, a novel additive manufacturing technology,” Coma’13, 2013.