ALS) with Mutations in Angiogenin and Superoxide Dismutase 1

Therapeutic strategies for the treatment of Amyotrophic Lateral Sclerosis (ALS) with mutations in Angiogenin and Superoxide Dismutase 1

by Krishna Chaitanya Aluri

Bachelor of Pharmacy, Rajiv Gandhi University of Health Sciences Master of Science in Biopharmaceutical Science, Northeastern University

A dissertation submitted to

The Faculty of the College of Science of Northeastern University in partial fulfillment of the requirements for the degree of Doctor of Philosophy

April, 16, 2020

Dissertation directed by

Jeffrey N. Agar Professor of Chemistry and Chemical Biology

Dedication

“When I walk, I walk with you. Where I go, you're with me always.”

― Alice Hoffman, The Story Sisters, 2009.

I dedicate this work to my family and friends. A special thanks to my parents Aluri Gopal Rao and

Mallela Visalakshi; brother Venkata Vishnuvardan Aluri and wife Prathyusha Gundlapally for their inspiration and words of encouragement.

I also dedicate this work to my friends Husain Attarwala, Arnik Shah, Aatman Doshi, Kirtika

Asrani, Smith Patel and Ranjitha Gaddipati for their support.

Acknowledgments

I would like to express my deep and sincere gratitude to Prof. Jeffrey N. Agar for continuous

support and guidance.

I would like to thank my fellow labmates Dr. Joseph P. Salisbury, Dr. Daniel P. Donnelly, Md.

Amin Hossain, Durgalakshmi Sivasankar, and Nicholas D. Schmitt for their contributions and thoughtful discussions.

I thank my thesis committee Prof. Alexander Ivanov, Prof. Ke Zhang, Dr. Jared R. Auclair, Dr.

Roman Manetsch, and Dr. Saeho Chong for their insightful comments, time, and encouragement.

I would like to thank our collaborators Dr. Jochen H.M. Prehn, Dr. Roman Manetsch, Dr. Adam

Ekenseair, Matthew G. Dowgiallo, Brandon C. Miller, and Ninad Kanetkar for their contributions.

Finally, I would like to thank my colleagues from Alnylam Pharmaceuticals for their support and

encouragement throughout the process of completing my dissertation and organization Alnylam

Pharmaceuticals for providing tuition assistance.

Abstract of Dissertation

Amyotrophic Lateral Sclerosis (ALS) is a fatal progressive neurodegenerative disorder. The cause of ALS is not completely understood. About 0.5–1% of ALS cases are associated with mutations in the angiogenin (ANG). These mutations are thought to cause disease through a loss of ANG function, but this hypothesis has not been evaluated statistically. In addition, the potential for ANG to promote disease has not been considered. With the goal of better defining the etiology of ANG-

ALS, we assembled all clinical onset and disease duration data and determined if these were

correlated with biochemical properties of ANG variants. Loss of ANG stability and ribonuclease

activity were found to correlate with early ALS onset, confirming an aspect of the prevailing model

of ANG-ALS. Conversely, loss of ANG stability and ribonuclease activity correlated with longer

survival following diagnosis, which is inconsistent with the prevailing model. These results

indicate that functional ANG appears to decrease the risk of developing ALS but exacerbate ALS

once in progress. These findings are rationalized in terms of studies demonstrating that distinct

mechanisms contribute to ALS onset and progression and propose that ANG replacement or

stabilization would benefit pre-symptomatic ANG-ALS patients. However, this study challenges

the prevailing hypothesis that augmenting ANG will benefit symptomatic ANG-ALS patients.

Instead, our results suggest that the silencing of ANG activity may be beneficial for symptomatic

ALS patients.

ANG has an in vivo half-life of less than 2 hours, to use ANG as prophylactic for ANG-ALS

multiple doses of ANG are required daily. To mitigate this problem, we evaluated cyclic

thiosulfinates as a novel class of compounds for hydrogel synthesis to encapsulate ANG. We used

cyclic thiosulfinates to cross-link PEG-thiol monomers and evaluated the safety of these hydrogels

in vitro. Using alkylated bovine serum albumin as a surrogate to ANG (which has no free cysteines

to cross-link) we demonstrated diffusion mediated sustained protein release.

Finally, 2% of ALS cases were associated with mutations in superoxide dismutase 1 (SOD1).

SOD1 is a homodimeric protein, in ALS-SOD1 variants the dimer destabilizes forming monomers which are prone to aggregation and are associated with toxicity. Previous studies demonstrated dimer stabilization as a viable therapeutic strategy but the cross-linkers used often form dead-end modifications with lone thiols making them harmful. Using alpha and beta lipoic acid (routinely used as dietary supplements) we demonstrated cyclic disulfides and cyclic thiosulfinates can efficiently cross-link SOD1 monomers while avoiding dead-end modifications of lone thiols.

Table of Contents

Dedication ...... 2

Acknowledgments...... 3

Abstract of Dissertation ...... 4

Table of Contents ...... 6

List of Tables ...... 9

Table of Figures ...... 10

1. Introduction ...... 12

1.1 Amyotrophic Lateral Sclerosis ...... 12

1.2 Epidemiology in ALS ...... 18

1.3 Role of angiogenin in ALS ...... 19

1.4 Controlled protein delivery and cross-linkers for hydrogel synthesis ...... 20

2. Loss of angiogenin function is related to earlier ALS onset and paradoxical increase in ALS duration

...... 22

2.1 Statement of Contribution ...... 23

2.2 Introduction ...... 24

2.2.1 Structure and function of angiogenin ...... 24

2.2.2 Identification of angiogenin as ALS risk factor and current hypothesis ...... 28

2.2.3 Physicochemical properties of proteins and role in disease pathophysiology ...... 29

2.2.4 Statistical methods...... 30

2.3 Results ...... 38

2.3.1 Loss of ANG stability correlates with faster ALS onset ...... 42

2.3.2 Loss of ANG stability and ribonuclease activity correlate with longer ALS duration ...... 45

2.3.3 Additional hypothesis testing ...... 49

2.3.4 Preclinical validation of a possible deleterious effect of ANG post-ALS onset...... 59

2.4 Discussion ...... 60

3. Cyclic Thiosulfinates as a Novel Class of Disulfide Cleavable Cross-linkers for Facile Hydrogel

Synthesis...... 67

3.1 Statement of contributions ...... 68

3.2 Introduction ...... 69

3.2.1 Hydrogels ...... 69

3.2.2 Crosslinkers in Hydrogel synthesis ...... 70

3.2.3 Cyclic Thiosulfinates...... 71

3.3 Results and Discussion ...... 72

3.3.1 Synthesis of 1,2-dithiane-1-oxide ...... 72

3.3.2 Cross-linking of 4-arm PEG thiol (PEG-4SH) ...... 75

3.3.3 Rheological characterization and Swelling of the hydrogel ...... 78

3.3.4 Determination of mesh size ...... 80

3.3.5 Degradation of the hydrogel ...... 82

3.3.6 Cytotoxicity and biocompatibility ...... 84

3.3.7 Protein encapsulation and release ...... 92

4. Cross-linking Superoxide dismutase 1 using cyclic thiosulfinates ...... 106

4.1 Statement of Contribution ...... 107

4.2 Introduction ...... 108

4.2.1 Structure of superoxide dismutase 1 and role in ALS ...... 108

4.2.2 Stabilizing SOD1 monomers using chemical cross-linking ...... 110

4.3 Results and Discussion ...... 111

4.3.1 Cyclic thiosulfinates for SOD1 stabilization ...... 111

4.3.2 Synthesis of β-lipoic acid ...... 114

4.3.3 Crosslinking of SOD1 monomers ...... 118

5. Conclusions of the Dissertation and Future Directions ...... 120

6. References ...... 122

List of Tables

Table 1. Genes and proteins implicated in ALS pathogenesis, their function and a possible mechanism

for ALS causation...... 13

Table 2. ALS Patient onset and disease duration ...... 35

Table 3. Stability of ALS associated ANG variants ...... 37

Table 4. Calculation of aggregation propensities for ANG variants associated with ALS...... 39

Table 5. Non-parametric tests Spearman’s Rho and Kendall’s Tau to evaluate the correlation of ALS

onset and disease duration to the physicochemical properties of ANG...... 49

Table 6. Kaplan-Meier curves’ log-rank, Breslow and Tarone-ware tests are used to evaluate statistical

equivalence in ALS onset, disease duration and lifespan of ANG variants with ∆∆G less than or

equal to -1 and variants with ∆∆G greater than -1 or relative (WT) ribonuclease activity less than

or equal to 10% or ribonuclease activity greater than 10%...... 53

Table 7. Adjustment of false discovery rate using the Benjamini-Hochberg method. The only p value

lost significance upon Benjamini-Hochberg adjustment was highlighted in bold...... 56

Table 8. Survival data of tgSOD1G93A-ALS mice dosed with 10 μg ANG...... 63

Table of Figures

Figure 1. Functional groups commonly used to cross-link thiols...... 21

Figure 2: Structure of human Angiogenin. Created using PYMOL. Three disulfide bonds were

represented by spheres...... 26

Figure 3: Role of angiogenin in various compartments of the cell...... 27

Figure 4: Lack of correlation between ANG stability and ANG ribonuclease activity and ANG

aggregation propensity...... 40

Figure 5: Destabilization of ANG variants correlates with faster ALS onset ...... 43

Figure 6: Loss of ribonuclease activity of ANG variants does not correlate with ALS onset...... 44

Figure 7: Destabilization of ANG variants correlates with longer ALS duration ...... 46

Figure 8: Loss of ANG ribonuclease activity correlates with longer ALS duration...... 48

Figure 9. ALS onset, disease duration, and lifespan do not correlate with aggregation propensity...... 52

Figure 10. PD onset does not correlate with ANG ribonuclease activity...... 54

Figure 11. ALS lifespan does not correlate with stability, loss of ribonuclease activity of ANG variants.

...... 55

Figure 12: Test for Cox proportionality assumption...... 58

Figure 13: Intraperitoneal treatment of tgSOD1G93A-ALS mice with 10 μg recombinant huANG post

disease onset did not prolong survival...... 60

Figure 14. Schematic representing changes in properties of hydrogels that can be altered to change the

drug release properties from hydrogels...... 70

Figure 15. Synthesis of 1,2-dithiane-1-oxide...... 73

Figure 16. NMR of 1,2-dithiane-1-oxide...... 74

Figure 17. Mechanism of hydrogel formation using PEG-4SH and 1,2-dithiane-1-oxide...... 76

Figure 18. Inverted test tubes demonstrating the formation of hydrogel using PEG-4SH and 1,2-dithiane-

1-oxide...... 77

Figure 19. Effect of pH on gelation time...... 78

Figure 20. Determination of equilibrium swelling of PEG-4SH-1,2-dithiane -1-oxide hydrogel...... 81

Figure 21. Characterization of hydrogel degradation in the presence of glutathione...... 83

Figure 22. Loss of storage modulus of hydrogel when incubated with 0.1,1 and 10 mM glutathione was

monitored using rheology...... 84

Figure 23. Cells are cultured as a monolayer in presence of 1 mM hydrogel in a culture plate...... 86

Figure 24. Cell viability in the presence of hydrogel cross-linked using 1,2-dithiane-1-oxide...... 88

Figure 25. Cells are cultured as a monolayer in presence of 1 mM hydrogel in a culture plate...... 89

Figure 26. Live/Dead cell assay of Hep G2 cells encapsulated in hydrogel synthesized by mixing PEG-

4SH to 1,2-dithiane 1-oxide at 1:2 molar ratio...... 91

Figure 27. In vitro release of BSA from the hydrogel...... 94

Figure 28. In vitro of release of BSA without alkylation ...... 95

Figure 29. Optimization of IPTG incubation time ...... 97

Figure 30. Detection of ANG expression in cell pellets and supernatant media...... 98

Figure 31. WT-ANG retention at various steps of expression, solubilization and refolding was confirmed

by SDS-PAGE...... 99

Figure 32. Purification of WT-ANG using Hydrophobic interaction chromatography...... 100

Figure 33. SDS-PAGE of HIC purification. Fractions 36-46 demonstrate elution of WT-ANG...... 101

Figure 34. Representative total ion chromatogram of WT-ANG trypsin digest - purified from E.Coli. 101

Figure 35. Sequence coverage of LC-MS/MS of tryptic digest of WT-ANG purified from E.coli...... 102

Figure 36. Schematic demonstrating encapsulation of protein in the hydrogel...... 104

Figure 37. In vitro release of ANG...... 105

Figure 38. Structure of human SOD1...... 109

Figure 39. Proposed mechanism of cross-linking by cyclic thiosulfinates...... 113

Figure 40. Synthesis of β-lipoic acid...... 114

Figure 41. 1H NMR of β-lipoic acid...... 116

Figure 42. High-resolution mass spectrum of β-lipoic acid...... 117

Figure 43. Mass spectrometry assay of α-lipoic acid and β-lipoic acid cross-linking SOD1...... 119

1. Introduction

1.1 Amyotrophic Lateral Sclerosis

Amyotrophic lateral sclerosis (ALS) also known as Lou Gehrig’s disease is a fatal and complex neurodegenerative disorder, first described by Charcot et al. in 1869 (Charcot, 1869). ALS is

characterized by rapid progressive loss of motor neurons resulting in muscle weakness, wasting,

and eventually death due to respiratory failure. ALS has an annual mean incidence of 1.89 per

100,000 people and a mean prevalence of 5.2 per 100000 people. Men have a 1.5-2 fold higher

prevalence of ALS than females (Wijesekera & Leigh, 2009). The overall lifetime risk of

developing ALS is 1 in ca. 400 (Ingre, Roos, Piehl, Kamel, & Fang, 2015). The mean onset of

ALS is about 60 years and disease prognosis varies based on the symptoms at the onset of disease;

Bulbar form of ALS relates to symptoms initiating with head and neck area, initial symptoms

include difficulties in speech and swallowing. Bulbar onset is generally considered the most severe

form of ALS and observed slightly more prevalent in women than men. The lumbar form of ALS,

also known as spinal ALS, has symptoms originating in extremities, mainly muscle weakness and fasciculations in limbs. The lumbar form of ALS is associated with two-thirds of ALS patients.

Patients with some forms of ALS also demonstrate loss of cognitive function and the symptoms closely resemble frontotemporal dementia (Lomen-Hoerth, Anderson, & Miller, 2002); most recent studies confirmed the involvement of mutations in C9ORF72 gene in both diseases

(Lattante, Ciura, Rouleau, & Kabashi, 2015). ALS has a median disease duration (onset of the first symptoms until the patient’s death or when respiratory assistance was required for patient’s

survival) between 1 - 2 years for bulbar onset and 3 - 5 years for lumbar onset (Wijesekera &

Leigh, 2009).

The majority of ALS cases are considered to be sporadic (sALS) and about 5 - 10% of cases are

associated with familial history (familial) (Rowland & Shneider, 2001). About 20% of familial

and 5% of sporadic ALS cases are associated with mutations in protein superoxide dismutase 1

(SOD1). Over a long period, SOD1 was the only known genetic cause for ALS. Rapid advances

in genomic sequencing technologies such as next-generation sequencing (NGS) enabled the

identification of more than 20 novel genes associated with ALS. Table 1 shows the genes

associated with ALS, their function, prevalence, and possible mode of pathogenicity. The most prevalent mutations associated with ALS include the genes SOD1, TARDP, C9orf72, ANG, FUS,

OPTN, and UBQLN2 (Morgan & Orrell, 2016).

Table 1. Genes and proteins implicated in ALS pathogenesis, their function and a possible mechanism for ALS causation.

Gene Protein Protein Prevalence Possible Pathogenicity References

Function (fALS) Mechanism

ALS2 Alsin Cellular <1% Oxidative stress (X. W. Su,

transport Broach,

Connor,

Gerhard, &

Simmons,

2014; Yang et

al., 2001)

ANG Angiogenin Angiogensis, <1% Loss of function, (Greenway et

RNA Aberrant RNA al., 2004; X. W.

metabolism processing Su et al., 2014)

APEX1 Apurinic Nuclease NA Oxidative stress (S. Chen,

endonuclease Sayana, Zhang,

& Le, 2013;

Hayward,

Colville,

Swingler, &

Brock, 1999)

ATXN2 Ataxin-2 RNA NA Stress-induced apoptosis (Elden et al.,

metabolism 2010; Hart &

Gitler, 2012)

C9ORF72 Chromosome 9 RNA 40% Gain or loss of function, (Gendron,

open reading frame metabolism RNA -mediated toxicity Belzil, Zhang,

72 & Petrucelli,

2014; Gurney

et al., 1994)

CHCHD10 Coiled-coil helix Unknown NA Mitochondrial toxicity (Bannwarth et

coiled-coil helix al., 2014)

protein

DAO D-amino acid Glutamatergic NA Protein Aggregation (X. W. Su et

oxidase signaling al., 2014)

CHMP2B Charged Cellular NA Protein Aggregation (Parkinson et

multivesicular body transport al., 2006; X. W.

protein 2B Su et al., 2014)

DCTN1 Dynactin subunit1 Axonal 1% Loss of function (Puls et al.,

transport 2003; Taylor,

Brown Jr, &

Cleveland,

2016)

FUS Fused in DNA/RNA 5% Aberrant RNA (X. W. Su et

sarcoma metabolism processing al., 2014;

Vance et al.,

2009)

HFE Human Iron NA Oxidative stress

hemochromatosis metabolism

protein hnRNP heterogeneous RNA NA Aggregation (H. J. Kim et

nuclear metabolism al., 2013; X. W.

ribonucleoproteins Su et al., 2014)

MATR3 Matrin 3 DNA/RNA NA Nuclear transport (J. O. Johnson

metabolism alteration et al., 2014)

OPTN Optineurin-1 Protein <1% Disruption (X. W. Su et

metabolism autophagosome al., 2014;

recruitment Wong &

Holzbaur,

2014; C.-H.

Wu et al.,

2012)

PFN1 Profilin 1 Axonal <1% Cytoskeletal alterations (X. W. Su et

outgrowth al., 2014; C.-H.

Wu et al.,

2012)

SETX Senaraxin DNA/RNA < 1% (X. W. Su et

metabolism al., 2014)

SOD1 Superoxide Oxidative 20% Oxidative stress, (Rosen et al.,

Dismutase 1 stress mitochondrial 1993; Sheng &

reduction dysregulation, microglial Xu, 2016)

activation, metabolic

stress, toxic gain of

function, protein

aggregation

SPG11 Spatacsin Neuronal NA Axonal dysfunction (Orlacchio et

maturation al., 2010; Sheng

& Xu, 2016; X.

W. Su et al.,

2014)

SQSTM1 Sequestosome 1 Protein <2% Autophagy impairment (Deng et al.,

metabolism 2011; Goode et

al., 2016; X. W.

Su et al., 2014)

TARDBP TAR DNA binding DNA/RNA 3% protein aggregation (Sreedharan et

protein metabolism al., 2008; X. W.

Su et al., 2014)

TBK1 Serine/threonine- Cell signaling NA Neuroinflammation, (Freischmidt et

proteinkinase impaired autophagy al., 2015;

TANK-binding Oakes, Davies,

kinase 1 & Collins,

2017; Taylor et

al., 2016)

TUBA4A Tubulin α-4A chain Microtubule <1% Cytoskeletal damage (Smith et al.,

subunit 2014; Taylor et

al., 2016)

UBQLN2 Ubiquilin2 Protein <2% Impairment of protein (Deng et al.,

metabolism degradation pathway, 2011; Gorrie et

protein aggregation al., 2014; X. W.

Su et al., 2014)

VCP Valosin- Protein 1-2% NA (J. O. Johnson

containing protein metabolism et al., 2010; X.

W. Su et al.,

2014)

Diagnosis of ALS is currently based on an assessment of clinical symptoms and physical

examinations standardized as the El Escorial criteria (Brooks, Miller, Swash, & Munsat, 2000).

The diagnosis of dominantly inherited fALS (e.g. SOD-1 FALS) is relatively straightforward based on familial history, because the disease is highly penetrant and kills roughly half of an affected family.

There is currently no cure for ALS andonly two drugs are approved for treatment. Riluzole, approved in 1995, is only able to prolong survival by up to three months (Zoccolella et al., 2007) and Radicava (edaravone), approved in 2017, improves symptoms but has not been shown to improve survival time (Robinson, 2017).

1.2 Epidemiology in ALS

Epidemiology uses a multidisciplinary statistical framework to understand the association of

various risk factors such as age, smoking, genetic mutations, diet, etc. to the time of onset and

progression of the disease (etiology)(Kuh & Shlomo, 2004). There has been a tremendous increase

in the application of epidemiology to analyze the risk factors in neurologic disorders in the last

decade. Genetic defects play a crucial role in neurological diseases. The emergence of genetic

epidemiology (the combination of traditional epidemiological approaches with various risk factors

associated with mutations in genes) in recent years significantly improved our understanding of

these complex disorders. In fact, these studies guided studies elucidating the molecular

mechanisms of diseases such as Alzheimer's disease, Parkinson’s disease, ALS, multiple sclerosis

etc.(Bertram & Tanzi, 2005; Mayeux, 2003).

In ALS, various studies correlating the physicochemical properties of disease-associated protein

variants laid the foundation and emphasized the need for these epidemiological studies. Wang et

al. applied epidemiology methods to correlate stability as well as aggregation propensity

(likelihood of proteins to aggregate based on mutations to primary structure) of ALS-SOD1

variants to disease duration. In this study they demonstrated that there is a synergistic effect of

aggregation propensity and stability of ALS-SOD1 variants on disease duration, and concluded that when added together these physicochemical risk factors account for an increased hazard > 300

(Q. Wang, J. L. Johnson, N. Y. Agar, & J. N. Agar, 2008). This is an exemplary study that paved the path for other studies looking into these risk factors in relation to ALS etiology. For example,

Abdolvahabi et al. studied the stochastic aggregation rates using mean lag time, propagation rate and mean maximal Thioflavin T fluorescence (amyloid dye) of these ALS-SOD1 variants in vitro,

18 and showed these correlate with disease duration (Abdolvahabi et al., 2017). Alemasov et al. developed a predictive model for disease duration (with accuracy ~ 4 years) using changes in the stability of hydrogen bonding of these variants (Alemasov et al., 2019). Finally, Schmitt et al. using their mass spectrometry experiments studying the structure of these ALS-SOD1 variants concluded certain post-translational modifications produce conformational changes to the electrostatic loop and cause destabilization and induce toxicity (Schmitt & Agar, 2017).

1.3 Role of angiogenin in ALS

Angiogenin is expressed widely in the human body. Angiogenesis is the canonical role of ANG.

In healthy neuronal cells, ANG is known to be important for neurite growth and pathfinding

(Subramanian & Feng, 2007). As a response to cellular stress, motor neurons secrete ANG, which is taken up by the astrocytes and acts as a neuroprotective agent in a paracrine manner

(Subramanian & Feng, 2007). Mutant ANG variant K40I was also taken up by the astrocytes but lacks neuroprotective function (Skorupa et al., 2012) indicating the ANG function is important during disease onset. ANG knockdown in using siRNA in motor neurons increased cell susceptibility to excitotoxic injury indicating the neuroprotective function of ANG (Kieran et al.,

2008). Under stress ANG is known to trigger multiple cell signaling pathways, altering cellular transcription and cleaving tRNA to form secretory granules. The effect of these changes in ANG functions on the metabolic support andmicroglial and astrocytes activation is not completely understood, in particular in cells expressing ALS-ANG variants. McLaughlin et al. have shown plasma and cerebrospinal fluid concentrations of ANG correlate well in healthy individuals,

whereas the correlation was lost in ALS patients indicating tissue-specific regulation of ANG

(McLaughlin et al., 2010).

1.4 Controlled protein delivery and cross-linkers for hydrogel synthesis

Therapeutic delivery of proteins is considered as a valuable tool for treating many disorders.

Delivery of therapeutic proteins such as ANG faces substantial challenges due to 1) in vivo

stability 2) immunogenicity and 3) short half-life. To avoid these challenges the proteins require

modifications or require encapsulation in a delivery vehicle. Nanoparticles and liposomes are

widely employed for encapsulation of the proteins. But the hydrophobicity of these delivery

mechanisms cause challenges for encapsulating hydrophilic proteins (Estrada & Champion, 2015;

Kobsa & Saltzman, 2008; Pisal, Kosloski, & Balu-Iyer, 2010). Hydrogels evolved as a major

alternative for the controlled delivery of proteins (Vermonden, Censi, & Hennink, 2012).

Physicochemical properties of hydrogels can be altered to optimize the delivery.

Hydrogel synthesis is performed by cross-linking functionalized synthetic macromers such a PEG

(poly(ethylene glycol)), PLGA (poly(lactic-co-hydroxymethyl glycolic acid) or natural monomers

such as chitosan, hyaluronic acid, keratin, etc. (Bhattarai, Gunn, & Zhang, 2010; Burdick &

Prestwich, 2011; C.-C. Lin & K. S. J. P. r. Anseth, 2009; S. Wang, Taraballi, Tan, Ng, & research,

2012). Thiol based cross-linking is especially of interest due to the abundance of thiol groups in

proteins. Various chemistries (Figure 3) such as thiol-epoxy, thiol-isocyanate, thiol-Michael

addition (thiol-ene), thiol-radical click reactions (thiol-acrylamide), etc. are widely explored for the synthesis of hydrogels (Hoyle, Lowe, & Bowman, 2010).

Figure 1. Functional groups commonly used to cross-link thiols.

In the subsequent chapters, we apply physicochemical epidemiology to the study effect of

physicochemical properties of ANG-ALS variants to ALS onset and disease duration. We introduce cyclic thiosulfinates as cross-linkers for facile synthesis of hydrogels and demonstrate their application towards protein delivery and finally, we introduce cyclic disulfides and cyclic thiosulfinates as cross-linkers for SOD1 monomers with an aim of applying these towards stabilizing ALS-SOD1 variants.

2. Loss of angiogenin function is related to earlier ALS onset and paradoxical increase

in ALS duration

Krishna C.Aluri 1,2, Joseph P. Salisbury 1,2, Jochen H. M. Prehn4 & Jeffrey N. Agar1,2,3

1 Barnett Institute of Chemical and Biological Analysis, Northeastern University, Boston, MA,

02115, USA.

2 Department of Chemistry and Chemical Biology, Northeastern University, 360 Huntington

Avenue, Boston, Massachusetts, 02115, United States.

3 Department of Pharmaceutical Sciences, Northeastern University, 360 Huntington Avenue,

Boston, Massachusetts, 02115, United States.

4 Department of Physiology and Medical Physics, SFI Future-Neuro Centre, Royal College of

Surgeons in Ireland, Dublin, 2, Ireland.

Reproduced (or 'Reproduced in part') with permission Aluri, K. C., Salisbury, J. P., Prehn, J. H.,

& Agar, J. N. (2020). Loss of angiogenin function is related to earlier ALS onset and a paradoxical increase in ALS duration. Scientific Reports, 10(1), 1-13. Copyright [2020] Scientific Reports.

2.1 Statement of Contribution

Experimental contributions to this chapter by Krishna C. Aluri are as follows: Krishna C.Aluri collected and organized data, performed all statistical analysis. Animal studies were performed in

Prof. Prehn’s lab. The manuscript was written by Krishna C. Aluri and Jeffrey N. Agar with contributions Joseph P. Salisbury and Jochen H. M. Prehn. All figures were prepared by Krishna

C. Aluri.

2.2 Introduction

2.2.1 Structure and function of angiogenin

ANG is a vertebrate-specific 14.1 kDa protein (Figure 2) belonging to the Ribonuclease A (RNase

A) superfamily. ANG is expressed by a wide variety of tissues and is a secreted protein. Compared

to human RNase A, human ANG has a 35% sequence similarity and a significantly slower and

more selective ribonuclease activity, which acts upon ribosomal and messenger RNAs and is

required for its powerful angiogenic activity. ANG has a kidney-shaped tertiary fold made of seven

antiparallel beta-strand and two alpha-helices connected trough loops. ANG acts by

transphosphorylation and hydrolysis for cleaving RNA and preferentially cleave next to

pyrimidines (Acharya, Shapiro, Allen, Riordan, & Vallee, 1994). The active site of ANG includes

various sub-sites that are involved in its function, a substrate-binding site that cleaves

phosphodiester bonds containing amino acids H13, K40 and H144 (highly conserved), purine

binding site, and an additional site for binding to peripheral bases holding mRNA in position. ANG

consists of three disulfide bonds in the hydrophobic core (Acharya et al., 1994; Palmer, Scheraga,

Riordan, & Vallee, 1986). The canonical role of Ang is the induction of angiogenesis by binding

to alpha-actin and initiates basement membrane degradation by activating plasminogen activator.

It further binds to 170 kDa protein to initiate signal transduction in the cytoplasm (Gao & Xu,

2008). ANG also undergoes leader-sequence dependent nuclear translocation and activates transcription of ribosomal RNA transcription (Tsuji et al., 2005) through interaction with ribosomal encoding DNA containing an Angiogenin Binding Element (ABE). Its ability to induce tumor neovascularization through a potent angiogenic activity is ANG’s best-characterized function and provided the first major support for Folkman’s tumor growth hypothesis. (Sheng &

Xu, 2016; Zuazo-Gaztelu & Casanovas, 2018) ANG was found involved in initiating cell

proliferation and also overexpressed in multiple types of cancers (Chao et al., 2003; Etoh, Shibuta,

Barnard, Kitano, & Mori, 2000; Shimoyama et al., 1996; Yoshioka, Wang, Kishimoto, Tsuji, &

Hu, 2006).

ANG performs diverse functions (Figure 3) in addition to its canonical angiogenic activity, often through interactions with proteins crucial to cell survival and growth. These interactions regulate cascades that mediate cell proliferation, apoptosis, and stress response, including AKT extracellular signal-related kinase 1/2 (ERK1/2); protein kinase B/Akt; and stress-associated protein kinase/c-Jun N-terminal kinase (SAPK/JNK). ANG interaction with NF-kB results in suppression of NF-kB nuclear translocation and inflammatory response (Bradshaw et al., 2017;

Crabtree et al., 2007; Sheng & Xu, 2016). Upon stress, in the cytoplasm, ANG performs regulated cleavage of tRNA into smaller fragments termed as tiRNA, which in turn regulates protein translation and generally promotes cell survival. ANG can promote neurite growth and pathfinding

(Subramanian & Feng, 2007); act as a neuroprotective and cytoprotective agent; activate microglia; and modulate astrocyte function (Kieran et al., 2008; Subramanian, Crabtree, &

Acharya, 2008). The mechanisms responsible for many of ANG’s non-canonical biological activities are not completely understood.

Figure 2: Structure of human Angiogenin. Created using PYMOL. Three disulfide bonds were represented by spheres.

Figure 3: Role of angiogenin in various compartments of the cell. Activation or upregulation of pathway indicated by . Inhibition indicated by X and, downregulation indicated by

2.2.2 Identification of angiogenin as ALS risk factor and current hypothesis

Greenway et al. first reported ANG mutations in ALS patients in Irish patients (Greenway et al.,

2004), Since the initial report, 24 mutations in ANG associated with ALS, and 12 mutations

associated with Parkinson’s have been reported, with these mutations being responsible for ca.

0.5-1% of these disorders (M.-L. Chen, Wu, Tai, & Lin, 2014; Kiernan et al., 2011). Wu et al. identified mutations of ANG in the American population with ALS and concluded that the loss of ribonuclease activity, nuclear translocation, or both, leads to loss of ANG function, which in turn leads to ALS (D. Wu et al., 2007). Crabtree et al. further solidified the importance of ribonuclease activity by showing six out of seven ALS associated ANG variants studied lost ribonuclease activity (Crabtree et al., 2007). Kishikawa et al. suggested angiogenin replacement as a potential therapy (Kishikawa, Wu, & Hu, 2008). However, two variants, R121H and R121C, have 156% and 131% of WT ribonuclease activity, respectively, indicating that loss of ribonuclease activity cannot be solely responsible for ANG-ALS. Previous studies used binary metrics, i.e. “yes” or

“no” with respect to having ALS or RNase activity, yielding a useful qualitative model. The goal of this study is a quantitative model of ANG-ALS; in particular, how biochemical changes relate to both age-of-ALS-onset and disease duration. Such a model offers the benefit of statistically evaluating prevailing hypotheses. In addition, such a model could inform treatment strategy, by separately evaluating pre-symptomatic and symptomatic risks.

2.2.3 Physicochemical properties of proteins and role in disease pathophysiology

Correlating biochemical and molecular aspects of disease-associated variants of proteins with their

respective phenotypes has aided in the understanding and diagnoses of many genetic disorders,

including Huntington’s disease (Duyao et al., 1993), Alzheimer’s disease (Luheshi et al., 2007),

Glaucoma (J. N. Burns, Turnage, Walker, & Lieberman, 2011), lysosomal storage diseases (Duyao

et al., 1993; Gieselmann, 2005), and ALS (Q. Wang et al., 2008). As previously reported

(Abdolvahabi et al., 2017; Q. Wang et al., 2008) in SOD1-associated fALS, there is a synergistic correlation between variant SOD1 instability and aggregation propensity and duration of survival after ALS onset. To our knowledge, this is the first report studying the physicochemical epidemiology of ANG associated with ALS.

Herein, we employ a technique we term “physicochemical epidemiology,” which previously demonstrated that the length of ALS patient’s disease duration (i.e. survival time after onset) depends upon the aggregation propensity and stability of fALS SOD1 variants (Q. Wang et al.,

2008). Specifically, the combined hazard ratio of the SOD1 variant’s loss-of-thermodynamic stability and gain-of-aggregation propensity was >300 and could account for 78% of the variability in patient’s survival after onset. For comparison, the hazard ratio for smoking with respect to lung cancer is ca. 12. The same study failed to identify any factor that affected the age-of-ALS-onset

(Q. Wang et al., 2008). These findings have been corroborated in numerous studies (Abdolvahabi et al., 2017; Austin et al., 2014; Belli, Ramazzotti, & Chiti, 2011; Borchelt et al., 1994; F. Chiti &

Dobson, 2006; Pratt et al., 2014; Ratovitski et al., 1999), and physiochemical epidemiology has since been applied to multiple diseases (Belli et al., 2011; F. Chiti & Dobson, 2006; Conway et al.,

2000; de Groot, Aviles, Vendrell, & Ventura, 2006; Fedde, Michell, Henthorn, & Whyte, 1996;

Watson et al., 2005). For example, ALS-associated TAR DNA-binding protein 43 (TDP-43)

variants with increased stabilities (i.e. variants with longer half-lives compared to wild-type TDP-

43), significantly expedite ALS onset (Watanabe, Kaneko, & Yamanaka, 2013), and protein

stability and activity predict the disease clinical phenotype in glucose-6-phosphate dehydrogenase

deficiency (Cunningham, Colavin, Huang, & Mochly-Rosen, 2017).

Using well-established epidemiological techniques, ribonuclease activity, aggregation propensity,

and thermodynamic stability are assessed with respect to ALS onset and disease duration. The

resulting models confirm that the loss of ANG variant’s ribonuclease activity and stability

contributes to ALS pathogenesis and confirm that ANG replacement is a valid pre-symptomatic

treatment strategy, but do not support the use of ANG replacement following the clinical diagnosis

of ALS symptoms.

2.2.4 Statistical methods

Advancement in the development of sophisticated mathematical models and complex computational methodologies enabled the widespread use of statistics in medicine. Statistical methods are routinely used in various applications in basic medical research such as identifying disease risk factors, correlating molecular changes with disease epidemiology, designing efficient clinical trials, the understanding effect of drugs on disease prognosis, etc. Continuous change in

covariates under analysis are routinely analyzed using methods such as analysis of variance

(ANOVA). For analysis of binary (yes or no) events such as the effect of treatment on survival of

patients, disease duration and survival of patients Kaplan-Meier analysis and Cox proportional hazards model were widely employed. In medical research, it is common to have patients lose follow-up during the study duration due to various reasons leads to partial data collection. In these

cases, we don’t have the time of event data but we have data to confirm that the event did not occur

until the last follow up, these events are termed as censored events and add valuable information

to the dataset. Censored events add further complexity, Kaplan-Meier analysis and the Cox proportional model are the most commonly employed statistical methods to analyze these data.

Both Kaplan-Meier and Cox analysis include familiar aspects of statistical testing such as the use of dependent and independent variables, testing the null hypothesis, determination of confidence intervals, etc.(Rich et al., 2010)

2.2.4.1 Kaplan-Meier Analysis

Kaplan-Meier Analysis also is known as the product-limit estimate is a nonparametric test described by Kaplan and Meier in 1958. (Kaplan & Meier, 1958) Kaplan-Meier curve estimates the probability of an event to occur in a given duration of time. The underlying assumptions in

Kaplan-Meier analysis include the censored patients have the same survival probability as those who continue in the study following the censored, survival probabilities are same for subjects recruited at any point of the study and finally, the event happens at the specified time.

Equation 1:

𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝑜𝑜𝑜𝑜 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑎𝑎𝑎𝑎 𝑡𝑡ℎ𝑒𝑒 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑜𝑜𝑜𝑜 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 − 𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 𝑜𝑜𝑜𝑜 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑤𝑤𝑤𝑤𝑤𝑤ℎ 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑆𝑆𝑡𝑡 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝑜𝑜𝑜𝑜 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑎𝑎𝑎𝑎 𝑡𝑡ℎ𝑒𝑒 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑜𝑜𝑜𝑜 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠

For each time point probability of an event to occur (Survival probability) can be calculated using

equation 1. The cumulative survival probability can be calculated by multiplying the probability

of the timepoint with the probability of all the preceding time points. To understand the difference

in survival of two groups, survival curves of two or more groups are tested for null-hypothesis

using various tests such as Log-rank, Breslow, and Tarone-ware tests (Clark, Bradburn, Love, &

Altman, 2003; Goel, Khanna, & Kishore, 2010).

2.2.4.2 Cox proportional hazards model

Kaplan-Meier analysis provides valuable information about the survival probabilities of two

groups and helps to differentiate survival probability between multiple groups, but it does not

provide any information on effect size i.e. it doesn’t provide an estimation of covariates impact on

the event to occur (Bradburn, Clark, Love, & Altman, 2003; Clark et al., 2003). Cox proportional

hazards model is a statistical method employed to infer effect size. The hazard function is defined

as the probability of an event to occur at an at a given time in the study. Cox proportional hazard

is a multivariate, semi-parametric method proposed by Cox in 1978. Hazard function ( ) at a

given time of study can be calculated using equation 2 according to the Cox proportionalℎ hazards𝑡𝑡

model. is baseline hazard function (calculated as the intercept of multiple linear regression), x1,

0 x2,.. xp areℎ covariates and b1, b2,.. bp are the size of the respective coefficients.

Equation 2:

( ) = ( ) ( ) 𝑏𝑏1𝑥𝑥1+𝑏𝑏2𝑥𝑥2+⋯+𝑏𝑏𝑝𝑝𝑥𝑥𝑝𝑝 0 ℎ 𝑡𝑡 ℎ 𝑡𝑡 ∗ 𝑒𝑒𝑒𝑒𝑒𝑒

Its underlying assumptions include the survival times are independent of each other and hazard curves are proportional which should be verified for a given data set. The hazard ratio can be described as exp(bi) (Cox, 1972).

Identification of ANG-ALS variants and relevant physicochemical properties

A comprehensive literature search retrieved 13 reports, including 48 patients representing 15 ANG

variants that met our inclusion criteria (see methods section for details). In addition, the

Amyotrophic Lateral Sclerosis Online Genetics Database (ALSOD) (Wroe, Wai-Ling Butler,

Andersen, Powell, & Al-Chalabi, 2008) and the ALS mutation database (Yoshida et al., 2010) were searched and did not produce additional data that met inclusion criteria. ALS onset data were available for all 48 ALS patients, and disease duration data were available for 30 of these 48 patients (Table 1). ANG ribonuclease activity and stability are hypothesized to affect clinical outcomes (Bradshaw et al., 2017; Thiyagarajan, Ferguson, Subramanian, & Acharya, 2012; D. Wu et al., 2007) and were therefore culled from the scientific literature. Ribonuclease activity data were available for 15 variants (Table 2) and stability data were available for 7 variants (Table 3).

Inclusion of clinical data in our study required that: 1) an ALS patient had been identified and reported in the literature to be carrying a specific ANG variant; 2) ALS-associated ANG variant had published thermal stability and/or ribonuclease activity data, and 3) individual ALS onset and/or duration of survival for that was explicitly stated that could be matched with the specific

ANG variant. In order to identify publications with clinical data of patients who met our inclusion criteria, PubMed was searched as late as December 2018 for all articles mentioning a combination of either “ANG” or “angiogenin” with either “ALS” or “amyotrophic lateral sclerosis” using the string (ANG and ALS) OR (angiogenin and ALS) OR (ANG and amyotrophic lateral sclerosis)

OR (angiogenin AND amyotrophic lateral sclerosis). The resulting articles were evaluated individually for ALS patients with a mutation in ANG. Publications reporting ANG mutations in

ALS patients were then further evaluated to identify publications that specifically include the ALS onset data and disease duration of individual patients that had a specifically identified ANG variant

that has been characterized for stability or ribonuclease activity. In extracting data from these

articles, disease duration was initiated with the onset of the first symptoms until the patient’s death

or when respiratory assistance was required for the patient’s survival. The resulting dataset was

checked against the Amyotrophic Lateral Sclerosis Database (http://www.alsod.org/) (Wroe et al.,

2008) and the ALS mutation database (Yoshida et al., 2010) to identify any additional data that

met our inclusion criteria (Table 2).

A single study obtained quantitative stability of ANG variants Q12L, K17E, K17I, R31K, C39W,

K40I, and I46V (Table 3) (Crabtree et al., 2007). No other stability data for ALS-associated ANG variants are available in the literature. The involvement of mutations such as K17I and I46V in causing ALS was questioned by previous reports (Michael A Van Es et al., 2011). We included these mutations in our analysis due to lack of comprehensive epidemiology of these rare variants and the presence of reports suggesting their benign prevalence (Padhi, Jayaram, & Gomes, 2013) in subsets of populations such as Caucasians (Pan et al., 2015), and Italians (Gellera et al., 2008).

The mean of ribonuclease activity was used where multiple ribonuclease activities were reported

(Table 2). Aggregation propensities of mutant ANG variants were calculated from the equation described by Wang et al (Q. Wang et al., 2008) ( Table 4). Patient with the concomitant SOD1 mutation was excluded from analysis (ANG R121C). Statistical analyses were performed using R

3.4.4 (Team, 2013), IBM SPSS Statistics 18 (IBM Inc., Armonk, NY, USA) and Stata SE 15 (Stata

Corp LLC. College Station, TX, USA).

Table 2. ALS Patient onset and disease duration

ANG Variant Disease duration ALS Reference (months) Onset (years)

Q12L >0 48 (Greenway et al., 2006)

Q12L >84 75 (Greenway et al., 2006)

K17I >0 53 (Greenway et al., 2006)

K17I 6 61 (M. A. van Es et al., 2009)

K17I 42 70 (M. A. van Es et al., 2009)

K17I >24 72 (M. A. van Es et al., 2009)

K17I 34 68 (M. A. van Es et al., 2009)

K17I >26 55 (M. A. van Es et al., 2009)

K17I 9 46 (Seilhean et al., 2009)

K17I 27 47 (Millecamps et al., 2010)

K17I 18 68 (Millecamps et al., 2010)

K17I 29 70 (Michael A Van Es et al., 2011)

K17I 16 62 (Michael A Van Es et al., 2011)

K17I 52 55 (Michael A Van Es et al., 2011)

K17I 20 77 (Michael A Van Es et al., 2011)

K17I 66 52 (van Blitterswijk et al., 2012)

K17E >36 66 (Greenway et al., 2006)

K17E 9.6 83 (Greenway et al., 2006)

R31K 12 66 (Greenway et al., 2006)

C39W 48 45 (Greenway et al., 2006)

C39W 84 47 (Greenway et al., 2006)

K40I 48 45 (Greenway et al., 2006)

K40I 120 27 (Greenway et al., 2006)

K40I 36 70 (Greenway et al., 2006)

I46V 18 76 (Greenway et al., 2006)

I46V 18 41 (Greenway et al., 2006)

I46V 144 45 (Greenway et al., 2006)

I46V 114 46 (Paubel et al., 2008)

I46V 25.2 73 (Paubel et al., 2008)

I46V 54 (Gellera et al., 2008)

I46V 51 (Gellera et al., 2008)

I46V 45 (Gellera et al., 2008)

I46V 64 (Gellera et al., 2008)

I46V 55 (Gellera et al., 2008)

I46V 31 (Gellera et al., 2008)

I46V 60 (Michael A Van Es et al., 2011)

I46V >60 68 (Conforti et al., 2008)

K54E 24 28 (Fernandez-Santiago et al., 2009)

K54E 60 49 (Kirby et al., 2013)

T80S 18 72 (Michael A Van Es et al., 2011)

F100I 8 72 (Michael A Van Es et al., 2011)

V103I 40 55 (Zou et al., 2012)

V113I 51 (Gellera et al., 2008)

V113I 63 (Gellera et al., 2008)

H114R 68 (Gellera et al., 2008)

R121H 31 31 (Paubel et al., 2008)

R121C* 30 72 (Luigetti et al., 2011) * Concomitant G93D SOD1 mutation

Table 3. Stability of ALS associated ANG variants

ANG mutation ∆∆G (kcal/mol) Reference Q12L -0.25 (Crabtree et al., 2007)

K17I -0.48 (Crabtree et al., 2007)

K17E 0.15 (Crabtree et al., 2007)

R31K -0.48 (Crabtree et al., 2007)

C39W -5.19 (Crabtree et al., 2007)

K40I -1.27 (Crabtree et al., 2007)

I46V -1.50 (Crabtree et al., 2007)

2.3 Results

Aggregation propensity is a significant hazard for disease duration (but not ALS onset) for patients

with ALS-associated SOD1 variants. ANG variant’s aggregation propensity was therefore

calculated as previously described, using the following equation 3 (Table 4).

Equation 3:

Chiti-Dobson solution

ln = 0.633 + 0.198 + 0.491 𝑉𝑉𝑚𝑚𝑚𝑚𝑚𝑚 �𝑉𝑉𝑊𝑊𝑊𝑊 � ∆𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻ℎ𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 �∆∆𝐺𝐺𝑐𝑐𝑜𝑜𝑜𝑜𝑜𝑜−𝛼𝛼 ∆∆𝐺𝐺𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐−𝛽𝛽� − ∆𝑐𝑐ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎

Change in aggregation propensity due to mutation is defined as ln . Change in 𝑉𝑉𝑚𝑚𝑚𝑚𝑚𝑚 �𝑉𝑉𝑊𝑊𝑊𝑊 � hydrophobicity and charge are represented by and , respectively.

Change in free energy upon change from alpha-helix∆𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 to theℎ random𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 coil is∆ defined𝑐𝑐ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 by ,

𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐−𝛼𝛼 and change in free energy upon change from beta-sheet to random coil represented∆∆𝐺𝐺 by

.Further evaluation with respect to ANG-associated ALS patients' clinical outcomes

𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐−𝛽𝛽 (Fabrizio∆∆𝐺𝐺 Chiti, Stefani, Taddei, Ramponi, & Dobson, 2003; Q. Wang et al., 2008).

Data were tested for normalcy using the Shapiro-Wilk test (Shapiro & Wilk, 1965), which indicated that ALS onset and aggregation propensity were normally distributed, but disease

38 duration, stability, and ribosomal activity were not. Non-parametric tests (Spearman’s, Kendall’s,

Log-rank, Tarone-ware, and Breslow’s tests), one semi-parametric statistical test (Cox proportional hazards analysis), and one parametric test (Pearson’s correlation) was employed, as was appropriate (i.e. when the model’s underlying assumptions were fulfilled). To determine whether stability, ribosomal activity, and aggregation propensity of ALS-associated ANG variants were related (i.e. dependent, as might occur if loss of stability resulted in a commiserate loss of activity), their relationships were tested using Spearman’s Rho and Kendall’s Tau, which demonstrated a lack of statistical correlation between these parameters (Figure 4).

These results were consistent with ANG stability and ribosomal activity being independent (or having a subset of components that are independent); with first principles arguments and previous reports indicating lack of correlation between stability and aggregation propensity; and indicated that each of these parameters needed to be evaluated separately with respect to clinical outcomes.

Table 4. Calculation of aggregation propensities for ANG variants associated with ALS.

Mutatio ∆Charg ∆Hydrophobici PαWT/Pαm ∆∆G_coi PβW ∆∆G_ Chiti- n e ty ut l-α T- β-coil Dobso Pβmu n t Solutio n Q12L 0.00 3.12 1.00 0.00 0.02 0.27 2.03 K17I 1.00 4.59 1.00 0.00 0.24 3.27 3.06 K17E -1.00 -0.14 1.00 0.00 -0.01 -0.14 0.38 R31K 0.00 1.18 1.23 0.54 0.01 0.14 0.88 C39W 1.00 1.88 0.50 -1.79 2.38 32.46 6.77 K40I 0.00 4.59 1.00 0.00 0.24 3.27 3.55 I46V -1.00 -0.52 1.50 1.05 -0.03 -0.41 0.29 K54E 0.00 -0.14 1.23 0.54 -0.01 -0.14 -0.01 T80S 0.00 -0.24 1.00 0.00 -0.24 -3.27 -0.80 F100I -1.00 -0.45 0.50 -1.79 0.03 0.41 -0.07

V103I 0.00 0.52 1.00 0.00 0.03 0.41 0.41 V113I -1.00 0.52 1.00 0.00 0.03 0.41 0.90 H114R 0.00 -3.31 1.00 0.00 0.02 0.27 -2.04 R121H 0.00 3.31 1.00 0.00 -0.02 -0.27 2.04 R121C -1.00 4.20 1.00 0.00 0.10 1.36 3.42

Figure 4: Lack of correlation between ANG stability and ANG ribonuclease activity and ANG

aggregation propensity. Spearman’s Coefficient and Kendall Tau’s coefficient were used for analyzing correlation. A significance level of 0.05 was used. a) Scatter plot demonstrating no

significant correlation between ANG stability and ANG ribonuclease activity b) Scatterplot

demonstrating no significant correlation between ANG ribonuclease activity and Aggregation

propensity c) Scatterplot showing no significant correlation between ANG stability and

aggregation propensity.

Separate analyses were therefore performed with respect to a given biochemical/physicochemical

characteristic (e.g. ribonuclease activity) and a given clinical outcome (e.g. ALS onset) in

accordance with previously published techniques (Abdolvahabi et al., 2017; Borchelt et al., 1994;

F. Chiti & Dobson, 2006; Cunningham et al., 2017; Ratovitski et al., 1999; Q. Wang et al., 2008).

These analyses included the following steps: 1) correlation of a biochemical/physiochemical

property and a clinical outcome; 2) establishing a threshold for the biophysical/physiochemical

property and separating data into two categories (e.g. categories with and without substantial

ribonuclease activity); 3) performing statistical comparisons between the clinical outcomes of

these categories; 4) if statistically appropriate, assigning an epidemiological hazard ratio to

statistically significant categories. Specifically, the monotonic relationship between a given

physicochemical parameter and clinical outcome was evaluated using Spearman’s coefficient (ρ)

(Spearman, 1904) and Kendall’s coefficient (τb) (Kendall, 1938). Survival data were then censored

to account for patients whose ALS onset times were reported, but whose survival times were not

reported; parsed according to their monotonic relationship; and Kaplan-Meier survival functions

(Ederer, Axtell, & Cutler, 1961) were generated. Statistical evaluations of the equality in survival

functions of these categories were performed using three independent tests, which taken together,

minimize potential weighting bias. These tests included Log-rank, which ranks all-time points equally (Mantel, 1966); Tarone-ware, which is weighted by a number of cases at risk at each timepoint (Tarone & Ware, 1977); and Breslow, which is weighted by the square root of the number

of cases at risk (Breslow & Day, 1980). Cox proportional hazards analysis was used to estimate

the hazard ratio. For Cox proportional hazards analysis, the data were analyzed using two models

1) Compared the mortality risk or ALS onset risk between two categories defined above using

them as categorical variables. Underlying assumptions for the Cox proportionality model were

evaluated graphically using log-log plots (Figure 12); 2) examined the association of a unit

increase in the physicochemical property to ALS onset risk or mortality risk using each

physicochemical parameter as a continuous variable. Underlying assumptions for the Cox model

were evaluated using Schoenfeld residuals (Cox, 1972; Zwiener, Blettner, & Hommel, 2011).

2.3.1 Loss of ANG stability correlates with faster ALS onset

To assess the relationship between ANG thermal stability and ALS onset, the stability of ANG

variants (∆∆G= difference in variant and WT ANG ∆Gs of unfolding) was correlated with the age

of patient’s at ALS onset. A significant correlation was observed; namely, greater ANG

destabilization was associated with earlier ALS onset (p value 0.01) (Figure 5a, Table 5). To further

investigate the relationship between ANG stability ALS onset, ANG variants were divided into

two categories, variants with ∆∆G less than or equal to -1, and variants with ∆∆G greater than -1,

and Kaplan-Meier analysis was performed (Figure 5b). Log-rank, Tarone-Ware, and Breslow analyses of the Kaplan-Meier curves each demonstrated a significant difference between these categories (p values in the range of 0.01-0.05) (Table 6). Specifically, patients harboring ANG variants with ∆∆G ≤ -1 (i.e. relatively destabilized) develop ALS 15 years earlier (47 ± 5 SE median age at onset) compared to those with relatively stable ANG variants (62 ± 4 SE median age at onset). The effect of stability on the risk of ALS onset could be evaluated using the Cox proportional hazards model using ∆∆G as a continuous variable. The overall model fit was

42 statistically significant (p value of 0.006); indicating a unit increase in stability, the risk onset decreases by 33% (hazard ratio 0.67, CI 0.50-0.89). Test for Cox proportional hazards assumption using Schoenfeld residuals (p value of 0.772), also represented by log-log plot (figure 13a), indicated there is no violation of the proportionality assumption.

Figure 5: Destabilization of ANG variants correlates with faster ALS onset. Spearman’s

Coefficient (p value 0.010) and Kendall Tau’s coefficient (p value 0.014) were used for analyzing correlation. Kaplan-Meier survival analysis was performed and the statistical significance of differences in survival between the categories was evaluated using Log-rank (p value 0.054),

Breslow (p value 0.015), and Tarone-ware (p value 0.027) tests. A significance level of 0.05 was used. a) Scatter plot demonstrating the correlation between thermal destabilization and ALS onset.

Sizes of points are proportional to the number of cases. b) Kaplan-Meier curves illustrating significant differences in ALS onset between patients with ANG variants with ∆∆G less than or equal to -1 kcal/mol and variants with ∆∆G greater than -1 kcal/mol.

To assess the relationship of ANG-ribonuclease activity and ALS onset, relative ribonuclease activities of ANG variants (to WT ANG) were correlated with the age of patients at ALS onset. No

significant correlation was observed between ALS onset and relative ribonuclease activity (p value range of 0.29-0.34) (Figure 6a, Table 5). To further investigate the relationship between loss of

ANG-ribonuclease activity and ALS onset, we divided the ANG variants into two categories, variants with less than or equal to ten percent WT ribonuclease activity and variants with greater than ten percent WT ribonuclease activity and performed Kaplan-Meier analysis (Figure 6b). The median ages of onset for patients with relatively low and high ANG activities were 51 ± 5 SE and

61 ± 4 SE years, respectively. Log Rank, Tarone-Ware, and Breslow analyses of the Kaplan-Meier curves each demonstrated no significant difference between these categories (p values in the range of 0.12-0.24) but did trend towards the significance threshold (Table 6). The effect of ribonuclease activity on the risk of ALS onset was evaluated using the Cox proportional hazards model using ribonuclease activity as a continuous variable, but the overall model fit was not significant.

Figure 6: Loss of ribonuclease activity of ANG variants does not correlate with ALS onset.

Spearman’s Coefficient (p value 0.340) and Kendall Tau’s coefficient (p value 0.294) were used for analyzing correlation. Kaplan-Meier survival analysis was performed and the statistical significance of differences in survival between the categories was evaluated using Log-rank (p value 0.235), Breslow (p value 0.119), and Tarone-ware (p value 0.173) tests. A significance level

of 0.05 was used. a) Scatter plot demonstrating no significant correlation between ribonuclease

activity and ALS onset. The sizes of points are proportional to the number of cases. b) Kaplan-

Meier curves illustrating no significant differences in ALS onset between patients with ANG

variants with %WT ribonuclease activity less than or equal to 10% and variants with %WT

ribonuclease activity greater than -10%.

2.3.2 Loss of ANG stability and ribonuclease activity correlate with longer ALS duration

To assess the relationship between ANG-thermal stability and ALS duration, the relative stability

of ANG variants was correlated with ALS patient’s disease durations. A significant negative

correlation was observed between loss of ANG stability and ALS duration (p value range of 0.01-

0.02) (Figure 7a, Table 5); namely, greater ANG destabilization was associated with longer ALS

duration. To further investigate the relationship between stability and ALS duration, we divided

the ANG variants into two categories, variants with ∆∆G less than or equal to -1 and variants with

∆∆G greater than -1 and performed Kaplan-Meier analysis (Figure 7b). The median survival for

patients with relatively low and high ANG activities were 48 ± 24 SE and 29 ± 8 8 SE months,

respectively. Log Rank, Tarone-Ware, and Breslow analyses of the Kaplan-Meier curves each

demonstrated p values slightly above the typical margin of significance (p values in the range of

0.07-0.09) (Table 6). The effect of ribonuclease activity on mortality risk was evaluated by Cox

proportional hazard analysis using ∆∆G as a continuous variable, but the overall model fit was not

significant.

Figure 7: Destabilization of ANG variants correlates with longer ALS duration. Spearman’s

Coefficient (p value 0.016) and Kendall Tau’s coefficient (p value 0.021) were used for analyzing correlation. Kaplan-Meier survival analysis was performed and the statistical significance of differences in survival between the categories was evaluated using Log-rank (p value 0.072),

Breslow (p value 0.086), and Tarone-ware (p value 0.077) tests. A significance level of 0.05 was used. a) Scatter plot representing the correlation between thermal destabilization and disease duration. The sizes of points are proportional to the number of cases. b) Kaplan-Meier curves illustrating differences in ALS onset between patients with ANG variants with ∆∆G less than or equal to -1 kcal/mol and variants with ∆∆G greater than -1 kcal/mol.

To assess the relationship between ANG ribonuclease activity and ALS duration, the relative ribonuclease activity of ANG variants was correlated to the disease duration of patients with ANG- associated ALS. A statistically significant negative correlation was observed between ALS duration and relative ribonuclease activity (p values in the range of 0.002-0.005) (Figure 8a, Table

5); namely, greater loss of ANG enzymatic activity correlated with longer ALS duration. To further

investigate the relationship between loss of ribonuclease activity and ALS duration, we divided

the ANG variants into two categories, variants with less than or equal to ten percent of WT ANG

ribonuclease activity and variants with greater than ten percent WT ribonuclease activity and

performed Kaplan-Meier analysis (Figure 8b). The median survival for patients with relatively low

ANG activities (48 ± 26 SE) was double that of patients with relatively high ANG activities (24 ±

7 SE months). Log-rank, Tarone-Ware, and Breslow analyses each demonstrated a statistically

significant difference between these categories (p values in the range of 0.002-0.01) (Table 5).

Likewise, Cox proportional hazards analysis examining mortality risk using two categories. The overall model fit was statistically significant (p value of 0.004) indicating ANG variants with

greater than ten percent WT ribonuclease activity was associated with increased risk of death

(hazard ratio = 4.1). Test for cox proportional assumptions demonstrated parallelism in log-log

plot indicating no violation of proportionality hazard (Figure 10d). The effect of ribonuclease

activity on mortality risk was evaluated using the Cox proportional hazards model using WT

ribonuclease activity as a continuous variable. The overall model was statistically significant (p

value of 0.029) indicating that for every unit increase in the relative percent of WT ribonuclease

activity, the mortality risk increases by 1% (hazard ratio 1.01, CI 1.001-1.019, i.e. 90% increase in

ribonuclease 90% increase in mortality). A test for the Cox proportional hazards assumption using

Schoenfeld residuals (p value of 0.837) indicated there is no violation of the proportionality

assumption. Therefore, using a number of conservative statistical tests, loss of ribonuclease

activity was significantly correlated to longer ALS duration.

Figure 8: Loss of ANG ribonuclease activity correlates with longer ALS duration. Spearman’s

Coefficient (p value 0.002) and Kendall Tau’s coefficient (p value 0.005) were used for analyzing

correlation. Kaplan-Meier survival analysis is performed and the statistical significance of

differences in survival between the categories was evaluated using Log-rank (p value 0.002),

Breslow (p value 0.01), and Tarone-ware (p value 0.005) tests. a) Scatter plot demonstrating a significant correlation between ribonuclease activity and ALS onset. Sizes of points are proportional to the number of cases. b) Kaplan-Meier curves illustrating significant differences in disease duration between patients with ANG variants with %WT ribonuclease activity less than or equal to 10% and variants with %WT ribonuclease activity greater than -10%.

Table 5. Non-parametric tests Spearman’s Rho and Kendall’s Tau to evaluate the

correlation of ALS onset and disease duration to the physicochemical properties of ANG.

Physicochemical property Spearman's Rho Kendall's Tau Correlation coefficient p value Correlation coefficient p value N Stability Ribonuclease 0.360 0.427 0.293 0.362 7 Activity Stability Aggregation -0.396 0.379 -0.390 0.224 7 Propensity Ribonuclease Aggregation -0.116 0.692 -0.033 0.870 14 Activity Propensity ALS Onset Stability 0.411 0.01* 0.313 0.014* 38 ALS Onset Ribonuclease 0.142 0.340 0.115 0.294 47 Activity Disease Stability -0.497 0.016* -0.385 0.02* 23 Duration Disease Ribonuclease -0.544 0.002* -0.399 0.005* 29 Duration Activity

* Indicates the significant statistical difference

2.3.3 Additional hypothesis testing

ANG-aggregation propensity did not significantly correlate with either onset (p value of 0.395)

(Figure 9a) or ALS duration (p values range of 0.10-0.13) (Figure 9b) or lifespan (p values range

of 0.40-0.48) (Figure 9c). Relative ANG-ribonuclease activity was correlated to the onset of PD

(n = 17). No significant correlation was observed between the onset of PD and %WT ribonuclease activity of ANG variants (p values range of 0.59-0.60) (Figure 10). Correlations with Parkinson’s disease duration couldn’t be performed as disease duration information was available for only two patients. Similarly, the stability of ANG variants cannot be correlated with either PD onset or PD duration as ∆∆G data was only available for two ANG-PD variants.

To illustrate the lifelong effects of ANG stability and activity, we have correlated these with the

total lifespan of ALS patients by adding disease duration to ALS onset time for patients having

both the data were reported (patients having only ALS onset data, but no survival data were

censored). Consistent with having averaged two competing forces, no significant correlations were

observed between the total lifespan of ALS patients and either stability (p values range of 0.07-

0.08) (Figure 11a) or ribonuclease activity (p values range of 0.56-0.73) (Figure 11b). Cox proportional hazards analysis examining the association of ALS lifespan using ∆∆G as a continuous variable, demonstrated statistical significance (p value of 0.009); indicating a unit

increase in stability, the risk mortality increases by 37% (hazard ratio 0.63, CI 0.44-0.89). Test for

Cox proportional hazards assumption (p value of 0.568) indicated there is no violation of the

proportionality assumption. Cox proportional hazards model examining the association of ALS

lifespan using ribonuclease activity as a continuous variable was performed. The data were not

reported as the overall model fit was not significant. Test for Cox proportional hazards assumption

(p value of 0.293) indicated there is no violation of the proportionality assumption. Likewise, using

the same thresholds to categorize ribonuclease activity and stability, Kaplan-Meier analyses

performed above: the median lifespans of patients with low versus high ANG stability and

ribonuclease activity were 850 ± 40.7 and 876± 148, and 876 ± 115 versus 834 ± 62.6 months,

respectively; and no significant difference was observed between the survival functions of stability

categories with respect to total lifespan (p values range of 0.18-0.28) or ribonuclease activity (p

values range of 0.62-1) (Figures 11c and 11d). Using the same thresholds to categorize

ribonuclease activity and stability, Cox proportional hazards analyses were performed: no

significant difference was observed between the hazard ratio of stability categories with respect to

total lifespan or ribonuclease activity categories with respect to lifespan and the data was not

reported. Tests for proportionality assumption demonstrated non-parallelism between the

categories indicating a violation of the Cox proportionality hazard model assumption (Figure 12).

Note, however, that patients with high ANG stability and activity survive an average of 13 and 9

years longer than the remaining ALS patients.

Correlation of ANG nuclear translocation activity to ALS onset and survival was not performed

as onset and survival data was only available for 7 patients and 5 patients, respectively with

mutations showing loss of nuclear translocation ability. Finally to account for multiple hypotheses

testing the false discovery rate was controlled using the Benjamini-Hochberg method (Table 7).

Three covariates (ANG stability; ANG aggregation propensity; ANG ribonuclease activity) were

correlated, independently, with two datasets (survival and onset). False discovery rate was

therefore corrected with respect to correlating three covariates to a particular dataset. Benjamini-

Hochberg method was used to adjust for the false discovery rate. A false discovery rate of 5% was used to adjust for type I error (Table 7). A total of 25 analyses were performed on our data.

All our p values found significant in our analysis were found to be significant using Benjamini-

Hochberg correction except, analysis of mortality risk using the Cox proportional hazards model using WT ribonuclease activity as a continuous variable. To increase stringency, we also required significance to be achieved through multiple statistical tests of each hypothesis (e.g. correlation,

Kaplan Meier, and Cox; as well as the use of survival hypothesis testing with different weighing functions using Log-rank, Tarone-ware, and Breslow).

Figure 9. ALS onset, disease duration, and lifespan do not correlate with aggregation propensity. a) Scatter plot demonstrating no significant correlation between aggregation propensity and ALS onset. b) Scatter plot demonstrating no significant correlation between aggregation propensity and disease duration. c) Scatter plot demonstrating no significant correlation between aggregation propensity and ALS lifespan. The definition of the aggregation propensity calculated here is important—its components (i.e. the mathematical terms used to calculate) are the fundamental physicochemical parameters of charge, hydrophobicity, and entropy. Aggregation propensity denotes the likelihood of undergoing a phase transition from an aqueous phase to an aggregated phase (i.e. it pertains to phenomena that occur following the nucleation step that “seeds” the

aggregate), and to a first approximation is independent of protein stability or activity. Our study

could not address the nucleation step that initiates the aggregation cascade—which could be related to protein stability—because the data were unavailable and cannot be estimated.

Table 6. Kaplan-Meier curves’ log-rank, Breslow and Tarone-ware tests are used to evaluate statistical equivalence in ALS onset, disease duration and lifespan of ANG variants with ∆∆G less than or equal to -1 and variants with ∆∆G greater than -1 or relative

(WT) ribonuclease activity less than or equal to 10% or ribonuclease activity greater than

10%.

Physicochemical property Log-rank Breslow Tarone-Ware Chi- p Chi- p value Chi- p value Square value Square Square Onset Stability 3.72 0.054 5.93 0.015* 4.89 0.027* Onset Ribonucleas 1.41 0.235 2.43 0.119 1.86 0.173 e Activity Disease Stability 3.23 0.072 2.94 0.086 3.14 0.077 Duration Disease Ribonucleas 9.65 0.002* 6.72 0.01* 8.02 0.005* Duration e Activity

* Indicates the significant statistical difference

Figure 10. PD onset does not correlate with ANG ribonuclease activity. Scatter plot demonstrating no significant correlation between ribonuclease activity and PD onset.

Figure 11. ALS lifespan does not correlate with stability, loss of ribonuclease activity of ANG

variants. Spearman’s Coefficient, Kendall Tau’s coefficient were used for analyzing correlation.

Kaplan-Meier survival analysis was performed and the statistical significance of differences in

survival between the categories was evaluated using Log-rank, Breslow, and Tarone-ware tests. A significance level of 0.05 is used. a) Scatter plot demonstrating no significant correlation between thermal destabilization and lifespan. b) Scatter plot demonstrating no significant correlation between ribonuclease activity and lifespan. c) Kaplan-Meier curves illustrating no significant

differences in lifespan between patients with ANG variants with ∆∆G less than or equal to -1

kcal/mol and variants with ∆∆G greater than -1 kcal/mol. d) Kaplan-Meier curves illustrating no

significant differences in lifespan between patients with ANG variants with %WT ribonuclease

activity less than or equal to 10% and variants with %WT ribonuclease activity greater than 10%.

Table 7. Adjustment of false discovery rate using the Benjamini-Hochberg method. The only p value lost significance upon Benjamini-Hochberg adjustment was highlighted in bold.

Physicochemical property Type of p Benjamini- Benjamini- Analysis value Hochberg Hochberg significance value Disease Ribonuclease Correlation 0.002 significant 0.036 Duration Activity Disease Ribonuclease Cox 0.004 significant 0.036 Duration Activity ALS Onset Stability Cox 0.006 significant 0.036 Lifespan Stability Correlation 0.007 significant 0.036 Lifespan Stability Cox 0.009 significant 0.036 ALS Onset Stability Correlation 0.01 significant 0.036 Disease Ribonuclease Kaplan-Meier 0.01 significant 0.036 Duration Activity ALS Onset Stability Kaplan-Meier 0.015 significant 0.044 Disease Stability Correlation 0.016 significant 0.044 Duration Disease Ribonuclease Cox 0.029 not significant 0.073 Duration Activity Disease Stability Kaplan-Meier 0.086 not significant 0.195 Duration Disease Aggregation Correlation 0.1 not significant 0.208 Duration Propensity ALS Onset Ribonuclease Kaplan-Meier 0.119 not significant 0.229 Activity Lifespan Stability Kaplan-Meier 0.18 not significant 0.321 ALS Onset Ribonuclease Correlation 0.34 not significant 0.510 Activity ALS Onset Ribonuclease Cox 0.343 not significant 0.510 Activity

Stability Aggregation Correlation 0.379 not significant 0.510 Propensity ALS Onset Aggregation Correlation 0.395 not significant 0.510 Propensity Lifespan Aggregation Correlation 0.4 not significant 0.510 Propensity Stability Ribonuclease Correlation 0.427 not significant 0.510 Activity Disease Stability Cox 0.428 not significant 0.510 Duration Lifespan Ribonuclease Correlation 0.56 not significant 0.617 Activity Lifespan Ribonuclease Cox 0.568 not significant 0.617 Activity Lifespan Ribonuclease Kaplan-Meier 0.62 not significant 0.646 Activity Ribonuclease Aggregation Correlation 0.692 not significant 0.692 Activity Propensity

a. b.

c. d.

Figure 12: Test for Cox proportionality assumption. Test for proportionality was performed using

log-log plots. a. Lack of parallelism was demonstrated using log-log plots between both categories

variants with between patients with ANG variants with ∆∆G less than or equal to -1 and variants with ∆∆G greater than -1. b. Lack of parallelism was demonstrated using log-log plots between

both categories variants with between patients with ANG variants with %WT ribonuclease activity

less than or equal to 10% and variants with %WT ribonuclease activity greater than -10%. c.

Parallelism was demonstrated using log-log plots between both categories variants with between

patients with ANG variants with ∆∆G less than or equal to -1 and variants with ∆∆G greater than

-1. d. of parallelism was demonstrated using log-log plots between both categories variants with

between patients with ANG variants with %WT ribonuclease activity less than or equal to 10% and variants with %WT ribonuclease activity greater than -10%.

2.3.4 Preclinical validation of a possible deleterious effect of ANG post-ALS onset.

We observed that in humans severe angiogenin deficiency is related to a more rapid onset of ALS

(10-15 years) and conversely less severe disease progression (c.a. order 2 years). Taken together these suggest a protective role for presymptomatic ANG treatment and a deleterious role for post- symptomatic ANG treatment. We previously demonstrated that in SOD1G93A-ALS mice, a preclinical mouse model of ALS, delivery of 1 μg of human recombinant ANG intraperitoneally

(3 times/ week) can extend lifespan, reduce aberrant molecular expression, and improve motor performance (Crivello et al., 2018; Kieran et al., 2008). Since our results from ALS patients

indicated that higher ANG activity or stability could be deleterious post-onset, a higher 10 μg dose

of ANG was administered intraperitoneally (i.p.) 3 times/week to tgSOD1G93A-ALS mice

following the onset of ALS symptoms. Whereas 1 μg improved multiple phenotypes in

tgSOD1G93A-ALS mice compared to vehicle-treated mice,(Crivello et al., 2018) Kaplan-Meier

analysis indicated that treatment with 10 μg of ANG did not (p value of 0.43, Figure 13, Table 8).

With the caveat that the mouse line used may not be the ideal approximation of the ANG ALS

population, our preclinical data provides initial evidence of enhanced ANG activity or stability

indeed not been beneficial or possibly being deleterious, post-ALS onset.

Figure 13: Intraperitoneal treatment of tgSOD1G93A-ALS mice with 10 μg recombinant huANG post disease onset did not prolong survival. a) Kaplan-Meier curves illustrated no significant differences in survival between tgSOD1G93A-ALS mice treated with vehicle and tgSOD1G93A-

ALS mice treated with 10 μg recombinant hANG (i.p., 3 times/week). Kaplan-Meier survival analysis was performed and the statistical significance of differences in survival between the categories was evaluated using Log-rank (p value 0.426), Breslow (p value 0.702), and Tarone-

ware (p value 0.55) tests. A significance level of 0.05 was used.

2.4 Discussion

Our results indicate that less stable and less active ANG variants correlate with more rapid ALS onset (15 and 10 years earlier, respectively), but paradoxically are also related to prolonged survival after onset (c.a. two years longer). These findings have implications for how ANG

replacement therapy is administered and how clinical trials for ANG replacement are designed.

Clearly, if a viable ANG replacement therapy becomes available, genotyping at a young age is

critical, because administering this therapy presymptomatically has the potential to increase

symptom-free life by an average of 15 years. Conversely, our data raise the possibility that

administering ANG replacement following ALS onset might be harmful and decrease the average

duration by 2 years. Indeed, if the data used in this study are representative of ANG variant’s

biochemical and biophysical properties, and the corresponding clinical parameters of ANG ALS patient’s, ANG replacement is indicated for presymptomatic patients, and ANG gene silencing is indicated for symptomatic ALS patients. What remains unclear is whether post-symptomatic ANG toxicity is specific to ANG-ALS variants (i.e. ANG-ALS variants gain a new toxic function and wild-type ANG may still be beneficial). At least one model does not support the idea of wild-type

ANG becoming toxic during progression. With SOD1G93A-ALS mice, daily delivery of 1 μg of human recombinant ANG intraperitoneally can extend lifespan, reduce aberrant molecular expression, and improve motor performance when treatment begins before onset or during progression (Crivello et al., 2018; Kieran et al., 2008).

How can ANG be of benefit before ALS symptoms but become harmful after ALS onset? The many beneficial roles ANG are well-understood and led to the prevailing hypothesis that ANG replacement is warranted in ALS. Given the consensus regarding the protective roles of ANG, these are only briefly discussed here and readers are referred to the following excellent review

(Sheng & Xu, 2016). In healthy cells, ANG is known to be important for neurite growth and

pathfinding (Subramanian & Feng, 2007). As a response to cellular stress, motor neurons secrete

ANG, which is taken up by the astrocytes where it acts as a neuroprotective agent in a paracrine manner (Subramanian & Feng, 2007). ANG variant K40I was also taken up by the astrocytes but

lacks neuroprotective function (Skorupa et al., 2012). Under stress ANG is known to trigger

multiple cell signaling pathways, altering cellular transcription and cleaving tRNA to form

secretory granules, and ANG knockdown using siRNA in motor neurons increased cell

susceptibility to excitotoxic injury indicating the neuroprotective function of ANG (Kieran et al.,

2008). These are all consistent with a loss-of-ANG function being particularly detrimental to

neurons that are under stress.

The observation that ANG activity and stability are related to faster ALS progression suggests a toxic role for ANG variants during the final stages of ALS. The ability of R121C to retain its ribonuclease activity and nuclear translocation ability (Padhi et al., 2013) provides initial evidence for an unknown role of ANG in ALS. A toxic role for ANG post-onset is further substantiated by clinical evidence from a patient carrying both G93D-SOD1 and ANG R121C mutations (131.2 %

WT ribonuclease activity). Symptoms of ALS were reported at the age of 72 years and the disease rapidly progressed for 2.5 years (Luigetti et al., 2011), compared to two reported patients carrying

G93D-SOD1 mutation only, previously described as slow progressing ALS mutation (Onset ranged from 45-71 years and disease progression ranged from 4 years to 22 years) (Restagno et al., 2008). The idea of beneficial entities becoming detrimental as ALS progresses is well- established. For example, the expression of Bcl2a1 is protective in G93A-SOD1-derived primary spinal cord cell cultures, until exposure to TNF-α, which mimics the condition of neuroinflammation that occurs after the onset of ALS. The authors suggested that Bcl2a1 serves a protective role before onset, and promotes cell death during disease progression (Crosio, Casciati,

Iaccarino, Rotilio, & Carri, 2006). Similarly, such context dependant, bimodal function ix was demonstrated by NF-kB in astrocytes of tgSOD1G93A-ALS mice. ANG is involved in the regulation of the NF-κB pathway by regulating the expression of four and a half LIM domains

62 protein 3.58 During the presymptomatic stage activation of NF-kB initiates microglial response showing neuroprotective role delaying onset of ALS. Whereas, post ALS onset NF-kB activation accelerates disease progression by switching the microglial phenotype.(Alami et al., 2018) It has been shown that inhibition of NF-kB rescues motor neuron survival in vitro and in vivo ALS models.(Frakes et al., 2014)

Table 8. Survival data of tgSOD1G93A-ALS mice dosed with 10 μg ANG.

Mice ID (10ug Survival

dose) (days)

7189 136

7211 142

7187 147

7184 154

7180 157

7185 162

7186 162

7178 164

7175 165

7176 167

7182 174

Pre-symptomatic ALS is characterized by mitochondrial deficiency, proteasome deficiency

(Kabashi, Agar, Strong, & Durham, 2012; Kabashi, Agar, Taylor, Minotti, & Durham, 2004), and

motor neuron retraction (Boillee, Vande Velde, & Cleveland, 2006). ALS progression, on the other

hand, is characterized by inflammation and an immune response that includes dominant astrocyte

and microglial activation [51] and activation of immunoproteasomes (Kabashi et al., 2012;

Kabashi et al., 2004). ANG alters the astrocyte secretome (Skorupa et al., 2013) and regulates

critical cellular pathways that could be either proapoptotic or antiapoptotic. ANG has been known

to cause angiogenesis by activating PKB/AKT pathway (H. M. Kim, Kang, Kim, Kang, & Chang,

2007), activation of the AKT pathway is shown to be proapoptotic by phosphorylating CDK2

(Maddika et al., 2008). Nitric oxide synthesis is increased in cells by ANG (Trouillon, Kang, Park,

Chang, & O'Hare, 2010), nitric oxide is known for its involvement in ALS increasing cellular

oxidative stress (Urushitani & Shimohama, 2001). ANG is also involved in the activation of the

NF-κB pathway by regulating the expression of four and a half LIM domains protein 3 (Xia et al.,

2015). It has been shown that inhibition of NF-kB in microglia rescues motor neuron survival in

vitro and in vivo ALS models (Frakes et al., 2014). Furthermore, Maruyama et al. suggested

inhibition of NF-κB as a potential treatment for ALS. ALS-associated optineurin mutants lack inhibitory effect on NF-κB activation demonstrated by wild type optineurin (Maruyama et al.,

2010) this provides evidence that activation of NF-κB could be deleterious after the onset of ALS.

The involvement of ANG in various pathways that are involved in pro and antiapoptotic signaling provides a reasonable doubt about its context-dependent role in cells. The context-dependent role has been previously associated with proteins such as prion protein, involved in both neuroprotective and neurotoxic functions (Steele et al., 2009). Finally, during stress ANG cleaves tRNA into fragments known as tiRNA which initiate the formation of stress granules. In cells

lacking NSun2, a protein that regulates tRNA methylation upon stress causes ANG induced tRNA

cleavage. In this model accumulation of tiRNA was able to induce apoptosis. Further, these cells

could be rescued by inhibiting ANG (Blanco et al., 2014). Aparicio-Erriu et al. demonstrated that the ANG internalized by astrocytes yield RNA fragments that are unlike tiRNA fragments generated within neurons and the role of these fragments and also the ANG substrates are not completely characterized (Aparicio-Erriu & Prehn, 2012). These examples further substantiate our data suggesting different roles of ANG during different cell types and phases of the disease. ANG performs paradoxical functions depending on cell types, for example, inhibiting proliferation of hematopoietic stem/progenitor cells and enhancing proliferation in lineage-committed myeloid- restricted progenitor cells (Goncalves et al., 2016).

Additional data are required to validate our physicochemical-epidemiological findings. The current data bottleneck is clinical data, which requires that more neurologists understand the importance of these data and publish them to case studies or repositories such as Amyotrophic

Lateral Sclerosis Database (ALSOD) (Wroe et al., 2008) and the ALS mutation database (Yoshida et al., 2010). Alternatively, if this does not happen, patient’s families should consider self-reporting these data to repositories. The major caveats of this study are: a few ANG mutants (e.g. K17I and

I46V) have relatively more clinical data, leading to these mutants having a relatively large effect upon the survival analyses; and there being relatively few biochemical studies. On the other hand, biochemical studies were performed within the same labs at the same time, which bodes well for their precision. These caveats notwithstanding, the whole of the biophysical and biochemical data for ANG and usable clinical data were employed here and provided consistent, statistically significant results using multiple conservative statistical approaches. These results are consistent with ALS-associated variants possessing full ANG stability and activity: prolonging lifespan by

65 delaying onset by as much as 15 years, and paradoxically decreasing survival time after onset by as much as two years.

3. Cyclic Thiosulfinates as a Novel Class of Disulfide Cleavable Cross-linkers for

Facile Hydrogel Synthesis.

Krishna C. Aluri1,2, Md Amin Hossain1,2, Ninad Kanetkar4, Brandon C. Miller1, Matthew G.

Dowgiallo1, Durgalakshmi Sivasankar1,2, Roman Manetsch1,3, Adam Ekenseair4and

Jeffrey N. Agar1,2,3*.

1Department of Chemistry and Chemical Biology, Northeastern University, 360 Huntington

Avenue, Boston, Massachusetts 02115, United States.

2Barnett Institute of Chemical and Biological Analysis, Northeastern University, 360 Huntington

Avenue, Boston, Massachusetts 02115, United States.

3Department of Pharmaceutical Sciences, Northeastern University, 360 Huntington Avenue,

Boston, Massachusetts 02115, United States.

4Department of Chemical Engineering, Northeastern University, 360 Huntington Avenue,

Boston, Massachusetts 02115, United States.

"Reproduced (or 'Reproduced in part') with permission from ACS Macro Letter, submitted for

publication. "Aluri, C. K., Hossain, A. Md., Kanetkar, N., Miller, C. M., Dowgiallo, G.D,

Durgalakshmi, S., Manetsch, R, Ekenseair, A, & Agar, J.N. (2018). 2. Cyclic Thiosulfinates

as a Novel Class of Disulfide Cleavable Cross-linkers for Facile Hydrogel Synthesis. Journal of

the American Chemical Society, 140(24), 7377-7380. Unpublished work copyright [2020]

American Chemical Society.

3.1 Statement of contributions

Experimental contributions to this chapter by Krishna C. Aluri are as follows: Krishna C. Aluri

designed the experimental plan and performed experiments. Cyclicthiosulfinate synthesis was

performed by Krishna C.Aluri, Brandon Miller and Matthew G. Dowgiallo. Amin Hossain performed cell culture work. The manuscript was written by Krishna C. Aluri and Jeffrey N. Agar with contributions Roman Manetsch and Adam Ekenseair. All figures were prepared by Krishna

C. Aluri.

3.2 Introduction

3.2.1 Hydrogels

Hydrogels are defined as the macromolecular polymeric networks that can hold large amounts of

water without losing their chemical and physical integrity (Wichterle & Lim, 1960). Cross-links

between these monomers help them to avoid dissolution in the water maintaining their integrity.

Wichterle and Lim demonstrated the synthesis of the first hydrogel in 1954 using 2-hydroxyethyl methacrylate and ethylene dimethacrylate (Wichterle & Lim, 1960). These hydrogels demonstrate both viscous as well as elastic behavior due to their ability to retain water and provide a wide range of use in various fields such as material chemistry, protein, drug delivery, tissue engineering, separation science, etc. (Hennink & van Nostrum, 2012). Hydrogels can be natural or synthetic polymers depending on the monomers used for synthesis (Majee, 2016). Natural hydrogels are synthesized by crosslinking natural monomers such as sugars (e.g.hyaluronic acid)(Luo, Kirker,

& Prestwich, 2000), proteins (e.g.chitosan)(Bhattarai et al., 2010), etc. Chemical hydrogels are synthesized using chemical monomers (e.g. Polyethylene glycol (PEG)(Abeysekera &

Sooriyarachchi, 2009)). Hydrogels are also classified based on the method of synthesis. Physical gels are formed by the phase transition in response to external stimuli such as changes in pH, ionic strength, photoactivation, and temperature. Chemical gels are formed by formation of covalent cross-links upon copolymerization of monomers using chemical cross-linkers. Biochemical gels are formed using enzymatic reactions activating gelation.(Majee, 2016)

Hydrogels are widely used for tissue engineering, drug, and biomolecule delivery (e.g. peptides, proteins, nucleic acids, and cells).(Drury & Mooney, 2003; Garcia, 2014; Hoare & Kohane, 2008;

Krebs, Jeon, & Alsberg, 2009; Lee & Mooney, 2001; Vermonden et al., 2012) Biodegradable

hydrogels are particularly of interest for these applications because of their high hydrophilicity,

ability to mimic soft tissue environment with their high biocompatibility, and ability to safely

disintegrate in vivo (Chai, Jiao, & Yu, 2017). Biodegradable hydrogels can be safely disintegrated

into monomers in the body and excreted out through renal filtration. Hydrogel degradation and the kinetics of biomolecule release can be manipulated by changing physical and chemical properties

such as altering their nanostructure (i.e. mesh size) of the hydrogel by manipulating the monomer

size, polymer volume and density, introduction of enzymatically or chemically cleavable sites

(Figure 14). These properties of hydrogels make them an attractive model for drug delivery

(Zustiak & Leach, 2010).

Figure 14. Schematic representing changes in properties of hydrogels that can be altered to change

the drug release properties from hydrogels.

3.2.2 Crosslinkers in Hydrogel synthesis

Poly (ethylene glycol) (PEG) is the most commonly used macromer for the synthesis of hydrogels.

PEG by itself is inert, non-reactive, and non-degradable, but the cross-linkers used to synthesize

70 hydrogels are more toxic (Sanborn, Messersmith, & Barron, 2002). For example, PEG molecules are routinely cross-linked via functional groups such as thiols, amines, acrylates, imides, methacrylates, and vinyl sulfone. Michael-type addition or free radical polymerization are widely explored for crosslinking.(C.-C. Lin & K. S. Anseth, 2009; Lowe, 2010; Nair et al., 2013)

Unfortunately, these chemical reactions come with drawbacks such as the need for photoinitiation for the generation of free radicals or the addition of catalysts such as triethylamine, limiting their use for in vivo applications.(Caliari & Burdick, 2016; C.-C. Lin & K. S. Anseth, 2009) Disulfide bond formation and thiol epoxy reaction are also commonly employed for hydrogel synthesis but these reactions are relatively slow (on the order of few hours) and require high concentrations of reactants, making these unreliable for in vivo applications (J. Su, 2018). In addition, the toxicity of most cross-linkers is high enough to prohibit their use in vivo. To minimize non-selective binding to free thiols in vivo and elicit toxicity, these crosslinkers need to be functionalized on to

PEG monomers (e.g. PEG-maleimide, PEG-acrylate, etc.).(Bakaic, Smeets, & Hoare, 2015;

Phelps et al., 2012) Moreover, most of these chemistries are not inherently biodegradable and require the inclusion of additional degradable groups such as poly (lactic acid), which increase the hydrophobicity and compromising the safety profile by forming pro-inflammatory metabolites

(Ishii et al., 2008).

3.2.3 Cyclic Thiosulfinates

We introduced cyclic thiosulfinates as novel cross-linkers that undergo disulfide exchange to form disulfide-linked cross-links between closely spaced free cysteine thiols whilst avoiding terminal

(a.k.a. “dead-end”) modifications of lone thiols (D. P. Donnelly et al., 2018). Cyclic thiosulfinates were shown to have nearly 100-fold increased reaction rates compared to cyclic disulfides, making

71 these molecules favorable for application as cross-linkers in polymer chemistry (Daniel P

Donnelly, Agar, & Lopez, 2019).

Critical features of cyclic thiosulfinate reactivity include (1) Cyclic thiosulfinates can selectively cross-link thiol pairs while the cross-links with lone thiols is reversible; (2) cross-linking occurs within the order of seconds at physiological pH; (3) does not react with other chemically reactive groups such as amines, disulfides, carboxylic acids, etc.; and (4) presence of reducing agents during cross-linking does not affect the cross-link formation.

In this manuscript, we use 4-arm PEG thiol as a monomer to evaluate the ability of cyclic thiosulfinates to form biocompatible and biodegradable hydrogels that can be used to deliver drugs, proteins, and cells. (D. P. Donnelly et al., 2018)

3.3 Results and Discussion

3.3.1 Synthesis of 1,2-dithiane-1-oxide

Synthesis of 1,2-dithiane-1-oxide was performed as reported (Figure 15) (Ananikov, Gayduk,

Beletskaya, Khrustalev, & Antipin, 2009; D. P. Donnelly et al., 2018). The initial step involved the synthesis of 1,2-dithiane from butane-1,4-dithiol (Figure 15). 41 grams of silica gel and distilled water (102 mL) were added to a round-bottomed flask and stirred vigorously to form a uniform suspension. 200 mL of dichloromethane and 2 grams of starting material (1,4- butanedithiol) were added while stirring. Bromine solution at 1.10 equivalents was added and stirred for 5 minutes until the reaction was complete and detected by thin-layer chromatography.

The reaction solution was filtered over celite in presence of 1 M sodium hydroxide. The organic phase was further washed three times with 50 mL of water and dried over sodium sulfate. The

72 solvent was evaporated under pressure and crystallized over hexanes to yield white crystalline 1,2- dithiane.

1,2-dithiane was dissolved in methanol and further oxidized using 1.10 equivalents of sodium periodate by dropwise addition. The reaction mixture was stirred for 16 hours to yield a white slurry which is further filtered over celite and concentrated under pressure. The aqueous layer was extracted with thrice with 40 mL of chloroform and the combined organic layer was dried over sulfate and dried under pressure. The synthesis product was finally purified by flash column chromatography using 2% methanol/ dichloromethane to yield a colorless solid of 1,2-dithiane-1- oxide confirmed by NMR (Figure 16). Chemical shifts matched with previously reported NMR spectra (Ananikov et al., 2009).

Figure 15. Synthesis of 1,2-dithiane-1-oxide.

Figure 16. NMR of 1,2-dithiane-1-oxide.

3.3.2 Cross-linking of 4-arm PEG thiol (PEG-4SH)

To demonstrate the ability of cyclic thiosulfinates to form hydrogels at physiological conditions,

PEG-4SH was cross-linked at a 1:2 molar ratio with 1,2-dithiane-1-oxide (i.e. equipotent) in PBS pH 7.4 at a feed ratio of 20% w/v for all the experiments. Hydrogel formation was monitored visually by the test tube inversion method,(Raghavan & Cipriano, 2006) and took place within a few seconds of the addition of the cross-linker (1,2-dithiane-1-oxide) (Figure 17). To evaluate the potential for the formation of the hydrogel in physiological preparations, we dissolved 1mM glutathione (GSH) with PEG-4SH before crosslinking. The presence of the reducing agent, glutathione, did not affect the hydrogel formation time (Figure 18). This result was possible because our cross-linkers were designed to bind lone thiols reversibly (via entropically favorable ring closure), and is consistent with our previous observation that protein sulfhydryl group cross- linking could occur in the presence of 10 mM GSH (D. P. Donnelly et al., 2018). Although the free thiols in the PEG-4SH are susceptible to autooxidation this did not appear to affect hydrogel formation, probably because the rate of the thiolate-disulfide interchange reaction is several orders of magnitude faster than the autoxidation reaction (Choh, Cross, & Wang, 2011). The effect of pH on gelation time was evaluated by dissolving the crosslinkers and PEG-4SH in buffers with varying pH (Figure 19). The gelation occurred fastest at low pH (4.8) and is significantly delayed at higher pH (11).

Figure 17. Mechanism of hydrogel formation using PEG-4SH and 1,2-dithiane-1-oxide.

Figure 18. Inverted test tubes demonstrating the formation of hydrogel using PEG-4SH and 1,2- dithiane-1-oxide. a. PEG-4SH-only indicating no gelation; b. 1,2-dithiane-1-oxide-only indicating no gelation; c. gelation of PEG-4SH and 1,2-dithiane-1-oxide in PBS; d. gelation of PEG-4SH and

1,2-dithiane-1-oxide in presence of 1 mM glutathione.

Figure 19. Effect of pH on gelation time.

3.3.3 Rheological characterization and Swelling of the hydrogel

Rheological characterization of hydrogels was performed to determine the elasticity [i.e. storage

modulus (G’)]. The hydrogel was synthesized in a 40 mm glass-bottom culture dish. Both the reactants were dissolved in phosphate buffer saline pH 6.5 to reduce the speed of reaction and allow homogenous mixing of PEG-4SH and 1,2-dithiane-1-oxide solutions. The hydrogel swollen for 24 hours in PBS pH 7.4 and storage modulus at equilibrium swelling was measured with AR

2000ex rheometer in parallel plate geometry with a 40-mm diameter upper plate, at room temperature, a frequency of 10 rad/s, and a constant 2% strain. For monitoring degradation of the hydrogel in the presence of glutathione, each hydrogel was incubated with 10 mL of 0.1,1, and 10 mM reduced glutathione. At each time point (0, 1, 3, and 6 hours) the storage modulus was measured using the same conditions. Our fully swollen hydrogel had G’ of 519 ± 110 Pa, which is comparable to reported G’ values (ca. 500 Pa ) (10% w/v PEG-4acrylate and 4 % PEG-4maleimide

hydrogels) (Phelps et al., 2012). The equilibrium swelling degree (qM) of the hydrogel is widely

used for determining the ability of the hydrogel to uptake water, calculating the distance between

cross-links, and the mesh size of the hydrogel. The swelling ratio of hydrogels is used to determine the mechanical properties, e.g., average molecular weight between crosslinks and mesh size of the

hydrogel, to optimize the hydrogel drug delivery (Ganji, Vasheghani, & VASHEGHANI, 2010;

Lin & Metters, 2006). To determine equilibrium swelling duration the gels were cast in the

NuncTM Lab-TekTM chamber slide system. The hydrogel was incubated in 2 ml PBS pH 7.4. At

each time point (the swollen hydrogels were weighed to determine the swollen weight (MS) followed by lyophilization for 24 hours to remove all the water and dry weight (MD) of hydrogel

was measured. The equilibrium swelling (qM) of hydrogel was determined using the following

equation 3 (Zustiak & Leach, 2010).

Equation 3:

( ) = 𝑀𝑀𝑆𝑆−𝑀𝑀𝐷𝐷 𝑞𝑞𝑀𝑀 𝑀𝑀𝐷𝐷

Our hydrogels did not change in swelling ratio (qm) overtime for up to 24 hours indicating the

initial volume of water allowed maximum swelling of our hydrogels (Figure 20). For our hydrogel

at equilibrium swelling, the calculated average molecular weight between crosslinks was 1721 ±

25 Da and mesh size was 5.2 ± 0.1 nm, which is comparable to 20% w/v 5 kDa PEG-norbornene

hydrogels (6.3 nm) (Rehmann et al., 2017).

3.3.4 Determination of mesh size

Mesh size was determined using the same constants for PEG polymer as described by Zustiak et al. The swelling ratio qM was used to determine swelling volume qV using equation 2. ρp is the density of dry PEG and ρs is the density of water.

Equation 4:

= 1 + ( 1) 𝑝𝑝 𝑉𝑉 𝜌𝜌 𝑀𝑀 𝑞𝑞 𝑠𝑠 𝑞𝑞 − 𝜌𝜌

The reciprocal of qV is defined as a polymer volume fraction (v2) and used for determining molecular weights between crosslinks (Mc) using Flory-Rehner calculation equation 5. is the

𝑛𝑛 average molecular weight of PEG-4SH, in our case 5000 Da. 𝑀𝑀

Equation 5:

1 2 ( (1 ) + + ) = 2 𝑣𝑣 2 2 1 2 𝑉𝑉1 𝐿𝐿𝐿𝐿 − 𝑣𝑣 𝑣𝑣 𝑋𝑋 𝑣𝑣 − 1 2 𝑀𝑀𝑐𝑐 𝑀𝑀𝑛𝑛 3 𝑣𝑣2 𝑣𝑣2 − The number of bonds in the crosslink (n) was calculated using equation 6. Mr is the molecular weight of the repeat unit in PEG.

Equation 6:

= 2 𝑀𝑀𝑐𝑐 𝑛𝑛 𝑟𝑟 𝑀𝑀

Root mean square end to end distance of the hydrogel was determined from equation 7. l is the average bond length.

Equation 7:

= 2 �𝑟𝑟0 𝑙𝑙�𝐶𝐶𝑛𝑛√𝑛𝑛

Finally, Canal and Peppas equation (equation 8) was used to determine mesh size (ξ)

Equation 8:

/ = −1 3 2 ξ 𝑣𝑣2 �𝑟𝑟0

10 Qm 9

7 0 10 20 30 Tim e (h)

Figure 20. Determination of equilibrium swelling of PEG-4SH-1,2-dithiane -1-oxide hydrogel.

Qm of the hydrogel did not change up to 24 hours indicating equilibrium swelling reached upon

hydrogel formation.

3.3.5 Degradation of the hydrogel

Whereas cyclic thiosulfinates can resist cleavage by glutathione, the cross-links formed by these compounds are reductively labile so long as the disulfide “banana” bond is accessible to nucleophilic attack. Disulfide cross-linked hydrogels are expected to be cleaved by reducing agents such as glutathione, dithiothreitol, and tris(2-carboxyethyl) phosphine. Glutathione is one of the widely expressed reducing agents in the human body. In vivo concentrations of glutathione range from 0.1 to 10 mM (Lafleur, Hoorweg, Joenje, Westmijze, & Retèl, 1994). We followed the degradation kinetics of hydrogels in the presence of 0.1, 1, and 10 mM glutathione using a swelling experiment (Figure 21). Hydrogel swelling was monitored for 24 hours. No change in swelling ratio was observed for control hydrogels incubated with PBS pH 7.4 within this 24 hour period, which indicated the swelling reached equilibrium and there was no degradation. Conversely, hydrogel mass increased ~150% over the period of 24 hours when incubated with low concentrations of glutathione (0.1 mM), consistent with an increased uptake of water due to loss of crosslinks (Choh et al., 2011). The hydrogels incubated with 1 mM and 10 mM glutathione exhibited an initial minor increase in mass. This was followed by a reduction in mass, starting at one and six hours, and culminating in disintegration at 6 and 14 hours, for the 1 and 10 mM glutathione treated samples, respectively. The storage modulus of the hydrogel in the presence of

1 mM glutathione was monitored using rheology. The data from the rheology experiment (Figure

22) was consistent with swelling data for the hydrogel that was incubated with 10 mM glutathione

(Figure 21).

200 PBS Glutathione 0.1 mM Glutathione 1 mM 150 Glutathione 10 mM

100

Relativemass hydrogel of 0 0 0.5 1 2 4 6 10 14 24 Time (h)

Figure 21. Characterization of hydrogel degradation in the presence of glutathione. Swelling of hydrogel relative to initial swelling upon incubation with PBS, 0.1 mM, 1 mM, and 10 mM glutathione (data connected using lines to guide the reader for pattern). Loss of weight of hydrogel was observed at higher concentrations of glutathione (1 mM and 10 mM) losing physical integrity within 24 hours.

10000

1000 0.1 mM Gluathione

1 mM Gluathione

10 mM Gluathione

Storage Modulus (G', Pa) (G', Modulus Storage 100 0 2 4 6 8 Time (hr)

Figure 22. Loss of storage modulus of hydrogel when incubated with 0.1,1 and 10 mM glutathione

was monitored using rheology. The loss of storage modulus was observed over 6 hours. No

significant loss of storage modulus was observed in hydrogels incubated with0.1,1 and 10 mM glutathione. Hydrogels incubated with 10 mM glutathione their physical integrity and completely dissolved after 3 hours.

3.3.6 Cytotoxicity and biocompatibility

Biocompatibility is a vital characteristic of any novel cross-linkers. A safe hydrogel intended for in vivo applications should support healthy cell growth, morphology, and survival. To determine the possible cytotoxicity of our hydrogel, PEG-4SH was cross-linked with 1,2-dithiane 1 oxide, and HepG2 cells were grown in the presence of hydrogel at the bottom of the well for 24 hours

(Figure 25). Both PEG-4SH and 1,2-dithiane-1-oxide were dissolved in cell culture media and filtered through 0.2 μM syringe filter and hydrogel was prepared in the 96-well plate at a final

concentration of 1 mm (based on final PEG-4SH concertation). We also evaluated the effect of

varying PEG-4SH to the cross-linker ratio (molar concentration ratio) on cytotoxicity. The

viability of the cells was monitored using colorimetric methyl thiazolyl diphenyl tetrazolium

bromide (MTT) assay (Figure 23a), where the MTT reagent is converted into formazan by NADPH

dependent oxidoreductase enzymes. No cytotoxicity was observed in both the conditions tested.

In fact, at 1:2 molar ratio of PEG-4SH to cross-linker, the number of cells increased compared to

controls. This interesting finding requires further study. We suspected free thiols interfering with

the assay and evaluated this using 1mM PEG-4SH as control (Figure 23b). When compared to

cells only control the PEG-4SH showed increased signal intensity indicating possible interference

of free thiols. Despite MTT assay being widely used in the determination of cell viability in

hydrogel synthesis using PEG-SH(Cai et al., 2016; Zustiak & Leach, 2010), we observed that free

thiols interfere with assay consistent with other reports showing similar interference (Neufeld,

Tapia, Lutzke, & Reynolds, 2018).

300

200

100 % Cell viability

ker

V ehicle

Chlorprom azine

1:1 PEG-4SH:Crosslinker1:2 PEG-4SH:Crosslin

150

100

50 % Cell viability

CELLS ONLY PEG -4SH O N LY C hlorprom azine

Figure 23. Cells are cultured as a monolayer in presence of 1 mM hydrogel in a culture plate. a.

The cell viability was based on the metabolic activity of cells determined using MTT assay. The data were normalized to media only control (vehicle). No significant changes in cell viability

observed for hydrogel synthesized using 1:1 molar PEG-4SH to cross-linker. An increase in

metabolic activity was observed in hydrogel synthesized with 1:2 molar ratio PEG-4SH to cross- linker. All samples were assessed in triplicate for the cytotoxicity assay and 500 µM of chlorpromazine was used as a positive control. b. MTT assay comparing 1mM PEG-4SH to cells

only control. Increased activity was observed in the PEG-4SH group indicating possible

interference with MTT assay.

To avoid the interference of free thiols and accurately measure the cell viability we used ATPlite™

assay. The viability of the cells was monitored using Adenosine Triphosphate (ATP)- monitoring

luminescence assay. The amount of ATP in the cells is directly related to the metabolic activity of

cells. The ATP released upon cell lysis reacts with added luciferin and produces luminescence

which can be measured. No significant cytotoxicity was observed in the presence of our hydrogel

with 1:1 PEG-4SH:cross-linker (p value 0.59), While 1:2 PEG-4SH:cross-linker showed 88% cell

survival (comparable to previously published hydrogels ~90 %), but one way-ANOVA analysis

showed significant statistical difference (p value 0.02) (Figure 24).

150

100 *

50 % Cell viability Cell %

Vehicle

Chlorpromazine

1:1 PEG-4SH:Cross-linker1:2 PEG-4SH:Cross-linker

Figure 24. Cell viability in the presence of hydrogel cross-linked using 1,2-dithiane-1-oxide. The cell viability was based on the metabolic activity of cells determined using ATPlite™ assay. The data were normalized to media only control (vehicle). No significant cytotoxicity was observed in the presence of our hydrogel with 1:1 PEG-4SH:cross-linker (p value 0.59), While 1:2 PEG-

4SH:cross-linker (indicated by *)showed 88% with statistical significance (p value 0.02). All samples were assessed in triplicate for the cytotoxicity assay and 500 µM of chlorpromazine was used as a positive control.

Figure 25. Cells are cultured as a monolayer in presence of 1 mM hydrogel in a culture plate.

Bright-field microscopic image demonstrating the proliferation of cells into the hydrogel.

For an application such as injectable hydrogels undergoing gelation in contact with cells, the

reaction should occur rapidly with minimal uptake of cross-linker. In addition, the cross-linker

should be tolerable to the cells, which is unprecedented for commonly used cross-linkers. For

example, maleimides and other small molecule cross-linkers show toxicity when used in vivo for

either drug and cell delivery or scaffolds (Jensen et al., 2002; Kharkar, Rehmann, Skeens,

Maverakis, & Kloxin, 2016). To determine whether live cells could survive in the presence of our

hydrogel precursors, we encapsulated live cells within the hydrogels using a 1:2 molar ratio of

PEG-4SH and 1,2-dithiane 1 oxide. To evaluate the effect of hydrogel encapsulation on HepG2

cell survival, 67 μL of 0.3 million/mL cells were added to PEG-4SH dissolved in cell culture

medium and crosslinking was performed using 1,2-dithiane-1-oxide to achieve a final molar ratio

of 1:2 PEG-4SH to 1,2-dithiane-1-oxide. The hydrogel encapsulating the cells was incubated for

1 hour in the incubator at 37oC before capturing fluorescent images using manufacturers protocol for Live/Dead cell assay. Calcein AM and ethidium homodimer-1 were used to detect live and

dead cells after encapsulation respectively. 500 μM chlorpromazine was used as a positive control

for cytotoxicity and DMSO for negative control for both MTT and Live/Dead cell assays Calcein

AM dye is taken up by live cells and produces green fluorescence in response to esterase activity.

Whereas, ethidium bromide produces red fluoresce due to loss of plasma membrane integrity. No

significant cell death was observed in Hep G2 cells encapsulated in the hydrogel (Figure 26).

Figure 26. Live/Dead cell assay of Hep G2 cells encapsulated in hydrogel synthesized by mixing

PEG-4SH to 1,2-dithiane 1-oxide at 1:2 molar ratio. a) cells treated with 500 µM chlorpromazine are all dead or dying. b) control cells were treated with media only or c) cells encapsulated in cross- linked PEG-4SH-1,2-dithiane-1-oxide hydrogel exhibit similar viability. Fluorescence images were acquired on Zeiss Axio Observer.Z1 Microscope (Zeiss, Germany) equipped with a

Hamamatsu digital camera C10600 Orca-R2, 5–40X objectives and standard FITC/DAPI/TRITC filters.

3.3.7 Protein encapsulation and release

Bovine serum albumin (BSA) encapsulation and in vitro release from the hydrogel.

To demonstrate the application of our disulfide cross-linked hydrogel for protein release, BSA was used as a model protein. Bovine serum albumin was alkylated with 10-fold Iodoacetamide before encapsulation, to mimic proteins with no accessible free cysteines. To mimic disulfide bound proteins with no solvent-accessible thiols, alkylated BSA is mixed with PEG-4SH before the addition of 1,2-dithiane-1-oxide. 1 mg of total protein was encapsulated in 100 μL of the hydrogel in a 2 mL Eppendorf microcentrifuge tube. The hydrogel was incubated with 1mL of PBS pH 7.4 at 37 oC while shaking at 300 rpm. 50 μL of PBS was sampled at 0, 0.5, 1, 2, 4, 6, 8, 24 and 96

hours. After each sampling, the media was replaced with fresh PBS. The protein concentration was

estimated using the Bradford assay following the manufacturer's protocol. BSA release data from

0 to 8 hours (i.e. ~60% of total protein released) was used to derive diffusion coefficient as

described by Choh et al. (Choh et al., 2011). The data were fit using the Goal Seek tool in Microsoft

Excel to estimate the diffusion coefficient of BSA release (Cornell, 2006).

A complete release of encapsulated BSA was observed by 96 hours (Figure 27a). First 60% of total BSA was released in the first 8 hours of incubation, this data was fitted using the Ritger-

Peppas model (Ritger & Peppas, 1987). The Diffusion coefficient of in vitro BSA release was

estimated using the Fickian diffusion model by fitting Fickian exponent of 0.45 for cylinders based

on our aspect ratio of 1.6, as described by Choh et al (Figure 27b) (Choh et al., 2011). The

estimated diffusion coefficient is 1.51*10-7 cm2/sec which was ~1.5 times slower than the theoretical diffusion coefficient of BSA in aqueous solutions (1.00*10-7 cm2/sec) (Rehmann,

Garibian, & Kloxin, 2013)., The ability to recover almost all of the encapsulated BSA indicates

there is limited chain transfer of native disulfides using our cyclic thiosulfinate cross-linkers.

In order to evaluate the ability of our hydrogel to cross-link proteins containing solvent-accessible free cysteines (e.g. reactive Cys 34 of BSA), we used BSA without the alkylation step. The first

50% of BSA was released over 4 hours. BSA release reached a plateau after 4 hours and 55% of total BSA was recovered over 24 hours. Our results indicated ~45% of BSA was cross-linked to the hydrogel (Figure 27).

0.8

0.6 ∞ 0.4 M /M

0.2 M /M ∞ (m odelled)

M /M ∞ (measured)

0.0 0 2 4 6 8 Tim e (h)

Figure 27. In vitro release of BSA from the hydrogel. a. 1 mg of BSA ~ 100 % of the encapsulated protein was recovered over 96 hours. b. Determination of the diffusion coefficient from the cumulative fractional release of BSA (M/M∞). Goal seek in excel was used to fit the Fickian diffusion model for cylinders to the first 60% of BSA released.

1 .0

0 .8

0 .6

0 .4

Total BSA (mg) Total 0 .2

0 .0 0 10 20 30 Tim e (h)

Figure 28. In vitro of release of BSA without alkylation. 0.55 mg of BSA ~ 55 % of the

encapsulated protein was recovered over 24 hours in PBS.

Expression, purification, and characterization of Wid-type human Angiogenin

ANG was originally reported to be isolated, purified and characterized by Fett et al. (Fett et al.,

1985) from the human adenocarcinoma cell line (HT-29). The procedure for was further optimized

by many labs to increase the efficiency expression and purification from various cell lines(Futami

et al., 1997; Moenner, Gusse, Hatzi, & Badet, 1994) and biofluids such as plasma (Shapiro & Wilk,

1965) and milk(Maes et al., 1988). E.Coli was first used for expression and purification of human

ANG by Yoon et al. (Yoon, Kim, Kwon, Han, & Kim, 1999) this procedure was further optimized

by Holloway et al (Holloway, Hares, Shapiro, Subramanian, & Acharya, 2001). We further

optimized these methods for the expression of human WT-ANG. WT-ANG is expressed in the

insoluble fraction of E.Coli and required reducing conditions for purification. The protein was later

refolded as described by Holloway et al. (Holloway et al., 2001). Plasmids expressing WT-ANG

were purchased from Genescript (Piscataway, NJ). BL21(DE3) Competent E. coli cells were

transformed with the WT-ANG DNA plasmid. The antibiotic-resistant colony and cells were

grown in the YCP medium (Yeast ext. 20g, Casamino acid 10g, KH2PO 4 3g, N~2I-IPO 4 2g, O's

4 2.5g, MgSO47H20 0.24g, CaC12 0.01g, Glucose 5g/liter) to reach OD of 2 at 30oC. E.coli was

then treated with 1 mM Isopropyl β- d-1-thiogalactopyranoside (IPTG) to induce protein transcription. IPTG treatment time (1, 2, and 14 hours post-treatment) was optimized to enable the highest protein expression monitored by SDS-PAGE (sodium dodecyl sulfate–polyacrylamide gel electrophoresis). ANG identifed based on molecular weight of 14 kDa. 1hour post IPTG treatment yielded highest expression of WT-ANG.

The E.coli cells were mechanically lysed using sonication with lysis buffer (50 mM Tris-HCl,

2mM EDTA, pH 8.0). The cells were centrifuged for 20 min and the cells were resuspended in lysis buffer containing 1%v/v Triton X-100. SDS-PAGE was run to detect the presence of WT-

ANG and supernatant and pellet fraction (Figure 29). The highest expression of WT-ANG was observed in the insoluble pellet fraction (Figure 29). The cell pellet fraction was resolubilized in reducing buffer containing 7 M guanidine hydrochloride, 0.15 M reduced glutathione, 0.1 M Tris–

HCl, 2 mM EDTA, pH 8.0, stirred under nitrogen for 2 h. To refold the protein with minimum aggregation we used L-arginine–HCl and 0.6 mM oxidized glutathione as described previously.

WT-ANG retention at various steps of expression, solubilization and refolding was confirmed by

SDS-PAGE (Figure 31).

Figure 29. Optimization of IPTG incubation time. The highest expression of ANG was observed

after 1 hr of incubation with IPTG.

Finally, resolubilized ANG was purified using hydrophobic interaction chromatography using a

phenyl sepharose fast flow column. The proteins were separated under salt gradient-based in the

hydrophobicity (Figure 32). The gradient started with a high salt gradient of 2 M ammonium

sulfate pH7 as mobile phase B and gradually decreased to 0 % ammonium sulfate. SDS-PAGE

was performed on the collected fragments to confirm the elution of WT-ANG. Fractions 36-46

(Figure 33) demonstrate the elution of WT-ANG. These fractions were desalted using 10 kDa molecular weight cut off the filter and buffer exchange in 10 mM Ammonium acetate buffer. The sequence of angiogenin was confirmed in-solution trypsin digestion. The tryptic peptides were separated by reverse-phase chromatography using a 40-minute gradient of 0 - 40% organic phase

(0.1% formic acid in acetonitrile) followed by MS/MS (Figure 34). WT-ANG sequence as

confirmed with 89.4% sequence coverage (Figure 35).

Figure 30. Detection of ANG expression in cell pellets and supernatant media. The highest

expression of WT-ANG was observed in the insoluble pellet fraction.

Figure 31. WT-ANG retention at various steps of expression, solubilization and refolding was confirmed by SDS-PAGE.

Angiogeni

Figure 32. Purification of WT-ANG using Hydrophobic interaction chromatography.

100

Figure 33. SDS-PAGE of HIC purification. Fractions 36-46 demonstrate elution of WT-ANG.

Figure 34. Representative total ion chromatogram of WT-ANG trypsin digest - purified from

E.Coli.

101

Figure 35. Sequence coverage of LC-MS/MS of tryptic digest of WT-ANG purified from E.coli.

102

Angiogenin encapsulation and in vitro release from hydrogel

ANG is responsible for angiogenesis and known to play a crucial role in cell survival and apoptosis

(Sheng & Xu, 2016). Mutations in ANG are associated with ALS. (Greenway et al., 2004)

Angiogenin replacement has been suggested as a therapeutic strategy for treating Amyotrophic

lateral sclerosis associated with mutations in ANG (Kishikawa et al., 2008). We also demonstrated

that in ALS with ANG mutations the stability and ribonuclease activity of ANG variants correlates

with ALS epidemiology. ANG has an in vivo half-life of 2 hours (Belli et al., 2011), therefore

encapsulation using an injectable hydrogel provides in-creased protection from proteolysis making

its therapeutic delivery feasible by reducing the number of doses. ANG has two disulfide bonds

that are required for its stability and activity. Thus, ANG serves as an ideal model protein for

evaluating protein release from our hydrogel encapsulation. The schematic of protein

encapsulation was shown in Figure 36.

Recombinant human angiogenin 1 ng of total protein was mixed with PEG-4SH before the addition

of 1,2-dithiane-1-oxide. The hydrogel was incubated with 2 mL of PBS pH 7.4 at 37 oC while

shaking at 300 rpm to ensure homogeneous mixing of released protein. 200 µL of PBS was

sampled at 0, 0.5, 1, 2, 3, 4, 6, 8 and 12 hours. After each sampling, the media was replaced with

the same volume of fresh PBS. The protein concentration was estimated using quantikine ELISA

kit for human angiogenin following the manufacturer's protocol.

A complete release of encapsulated ANG was observed within 4 hours. Our in vitro ANG release

data was fit using the Fickian diffusion model (Figure 37) for cylinders as described by Choh et al

(Choh et al., 2011). The diffusion coefficient was estimated to be 1.51E-07 which was times 1.03

slower than the theoretical diffusion coefficient calculated using Stokes-Einstein’s equation

(1.46E-07) indicated our hydrogel did not significantly delay ANG release. Because of the smaller

103 size of the ANG (hydrodynamic radius of 1.7 nm calculated using Stokes-Einstein equation) compared to our mesh size (5.7 nm), ANG was able to freely diffuse through the hydrogel with minimal hindrance. To reduce the rate of ANG release the mesh size of the hydrogel needs to be optimized, but our experiment with BSA demonstrates the application of our hydrogel for diffusion mediated release of proteins with no accessible free cysteines.

Figure 36. Schematic demonstrating encapsulation of protein in the hydrogel. a. formation of the hydrogel in the presence of protein with no solvent-accessible free cysteines b. formation of cross-links between hydrogel and protein in presence of solvent accessible free cysteines.

104

Figure 37. In vitro release of ANG. ANG was observed for up to 12 hours. About 100 % of total

ANG encapsulated in the hydrogel was released by 4 hours. Mass of ANG recovered at a given time (Mt) was plotted against the total mass of ANG released (Minf).

105

4. Cross-linking Superoxide dismutase 1 using cyclic thiosulfinates

Donnelly DP1,2, Dowgiallo MG1, Salisbury JP1,2, Aluri KC1,2, Iyengar S1, Chaudhari M1,2,

Mathew M1, Miele I1, Auclair JR1,2, Lopez SA1, Manetsch R1,3, Agar JN1,2,3.

1Department of Chemistry and Chemical Biology, Northeastern University, 360 Huntington

Avenue, Boston, Massachusetts 02115, United States.

2Barnett Institute of Chemical and Biological Analysis, Northeastern University, 360 Huntington

Avenue, Boston, Massachusetts 02115, United States.

3Department of Pharmaceutical Sciences, Northeastern University, 360 Huntington Avenue,

Boston, Massachusetts 02115, United States

Reproduced (or 'Reproduced in part') with permission Donnelly, D. P., Dowgiallo, M. G.,

Salisbury, J. P., Aluri, K. C., Iyengar, S., Chaudhari, M., ... Manetsch, R & Agar, J.N. (2018).

Cyclic thiosulfinates and cyclic disulfides selectively cross-link thiols while avoiding modification of lone thiols. Journal of the American Chemical Society, 140(24), 7377-7380. Copyright [2018]

American Chemical Society.

106

4.1 Statement of Contribution

Experimental contributions to this chapter by Krishna C. Aluri are as follows: Krishna C. Aluri performed the synthesis of beta-lipoic acid and used it for in vitro cross-linking experiments. All lipoic acid mass spectrometry experiments were performed by Krishna C. Aluri. NMR data was acquired by Matthew G. Dowgiallo. The manuscript was written and figures were prepared by

Daniel P. Donelly with contributions from all coauthors.

107

4.2 Introduction

4.2.1 Structure of superoxide dismutase 1 and role in ALS

Cu/Zn Superoxide dismutase 1 was the first known and well-studied protein involved in ALS. It

is metalloenzyme encoded by chromosome 21q (Rosen et al., 1993). SOD1 is a homodimer with

a highly conserved structure made of 153 amino acids (Figure 38). The dimer interface is held in

place by hydrophobic interactions making SOD1 one of the most stable proteins (Forman &

Fridovich, 1973). Each SOD1 subunit is made of eight antiparallel beta-strands arranged in a Greek

key motif. Two major loops include metal binding containing amino acids 49-84 and bind to metals

one zinc ion and one copper ion. The metal-binding loop contains an intramolecular disulfide bond between Cys 57 ad Cys 146, that stabilizes the structure. The electrostatic loop containing amino acids 122-143 acts as an active site channel (Rakhit & Chakrabartty, 2006). Binding of SOD1 to copper and zinc ions is also important for structural stability and also helps SOD1 for catalyzing dismutation reaction (Kaur, McKeown, & Rashid, 2016; Perry, Shin, Getzoff, & Tainer, 2010).

SOD1 undergoes post-translation modifications such a metal ion binding, N terminal acetylation,

and disulfide formation, critical for its structure and function. Canonical role of SOD1 is the

.- dismutation of reactive and toxic superoxide anions (O2 ) into hydrogen peroxide (H2O2) and

oxygen (O2) (Fridovich, 1986). Figure 1 shows the 3D structure of SOD1 with important structural

information. More than 170 mutations (Sangwan et al., 2017) in SOD1 are implicated in ALS,

with A4V being the most common mutation in North America. (Ajroud-Driss & Siddique, 2015).

The exact role of SOD1 in ALS is not completely understood, but a gain of toxic function is the

most accepted hypothesis (Borchelt et al., 1994). Proposed hypothesis for SOD1 toxic gain of

function include alterations instability due to loss of metal-binding (Crow, Sampson, Zhuang,

108

Thompson, & Beckman, 1997; Stathopulos et al., 2003), misfolding, changes in post-translation modifications (Bosco et al., 2010), loss of dimer stability (Molnar et al., 2009), increased hydrophobicity (Tiwari, Xu, & Hayward, 2005), changes in copper-mediated reactions (Beckman,

Carson, Smith, & Koppenol, 1993), and increased aggregation propensity (Q. Wang et al., 2008).

Figure 38. Structure of human SOD1. Created using PYMOL. Three disulfide bonds were

represented by spheres. Circles in magenta and grey represent metal ions Zn and Cu.

109

4.2.2 Stabilizing SOD1 monomers using chemical cross-linking

Loss of dimer stability and gain of toxic function of ALS-SOD1 variants was a major hypothesis associated for the mechanism of toxicity of familial ALS (Andersen, 2006; Durham, Kabashi,

Taylor, & Agar, 2007; Gurney et al., 1994; Reaume et al., 1996). Further many studies suggest these ALS-SOD1 variants upon loss of dimer stability dissociate into monomers which expose the hydrophobic dimer interface leading to misfolding and increased propensity to form higher-order aggregates (Doucette et al., 2004; Hörnberg, Logan, Marklund, & Oliveberg, 2007; Rakhit et al.,

2004; Rakhit et al., 2007). Wang et al. further showed the aggregation propensity of these mutants correlate with disease onset (Qi Wang, Joshua L Johnson, Nathalie YR Agar, & Jeffrey N Agar,

2008). Many SOD1 variants such as G93A, A4V, H46r, G85R, G93R, E100G, and I113T are observed to be forming monomers in vivo and showed aggregate formation consistent with the destabilization and aggregation hypothesis (Hough et al., 2004; Vassall, Stathopulos, Rumfeldt,

Lepock, & Meiering, 2006; J. Wang et al., 2007).

Stabilizing the misfolded proteins using small molecules is a well-known strategy in drug development. A most recent example of a drug that stabilized misfolded proteins is Tafamidis.

Tafamidis is recently approved by the FDA to treat transthyretin familial polyneuropathy (Scott,

2014). Transthyretin is a tetrameric protein in healthy individuals, mutations in the protein

destabilize tetramer forming monomers which misfold and aggregate and form amyloids causing

deadly disease. Tefamnidis meglumine is an NSAID benzoxazole derivative having high affinity

and selectivity to transthyretin protein. Upon binding to transthyretin tafamidis kinetically

stabilizes the protein reducing aggregation of the protein (Coelho et al., 2012; Coelho et al., 2016;

S. M. Johnson et al., 2005). Further examples of such stabilizing misfolded proteins using small

110 molecules include polyglutamine diseases such as Huntington disease, spinocerebellar ataxia where CAG trinucleotide repeats using trehalose (Tanaka, Machida, & Nukina, 2005).

Stabilization of N370S β-glucosidase using chemical chaperones in Gaucher disease (Sawkar et al., 2002).

Using SOD1 G93A and G85R variants causing fALS, Auclair at al. demonstrated that maleimide functional groups and thiol-disulfide exchange reactions can be exploited to cross-link adjacently spaced Cys111 in SOD1 monomers. This chemical crosslinking increased the stability of G93A

SOD1 by 20 °C and G85R SOD1 by 45 °C (monitored by thermo-fluorescence assay). Cross- linking also restored the activity of otherwise inactive mutants (Auclair, Boggio, Petsko, Ringe, &

Agar, 2010).

4.3 Results and Discussion

4.3.1 Cyclic thiosulfinates for SOD1 stabilization

Though thiol-ene cross-linking chemistries provide a valuable tool for demonstrating in vitro activity and proof of concept for stabilization of SOD1 (Auclair, Boggio, Petsko, Ringe, & Agar,

2010), these are not compatible for in vivo application. These tools are not selective and form terminal “dead-end” modifications (leave terminal reactive groups when reacted with single free cysteines, that can react with other active cysteines) which are toxic and elicit adverse immune response (Baillie, 2016).

Cyclic disulfides are a major class of cysteine reactive molecules extensively studied by

Whitesides’ group (J. A. Burns & Whitesides, 1990; Singh & Whitesides, 1990; Szajewski &

Whitesides, 1980). These molecules are highly selective as they are entropically driven by the

111

propensity to remain oxidized and cyclic; Moreover, the ability to form dead-end modifications is

driven by ring strain which is 3 orders of magnitude higher (J. A. Burns & Whitesides, 1990;

Szajewski & Whitesides, 1980). Cyclic disulfides also have a higher therapeutic window (5 g/day)

and can be transported across cell membranes (Abegg et al., 2017; Zong et al., 2017) especially,

α-lipoic acid is a cyclic disulfide routinely used as a nutritional supplement.

We hypothesized cyclic disulfides could be efficiently used for cross-linking cysteines in SOD1 monomers while avoiding dead-end modifications (D. P. Donnelly et al., 2018). We also reasoned that using mono-S-oxo cyclic disulfide i.e. cyclic thiosulfinates can increase the reaction rate while avoiding the slowest step of cross-linking reaction (Figure 39).

112

Figure 39. Proposed mechanism of cross-linking by cyclic thiosulfinates. Dead-end modification

is minimized by entropically favorable ring closure Cross-linking proceeds through condensation of CysA (a), sulfenic acid derived from either rate-limiting S-oxidation of thiolate (b) or a cyclic thiosulfinate (c). Amended and reproduced with permission Donnelly, D. P., Dowgiallo, M. G.,

Salisbury, J. P., Aluri, K. C., Iyengar, S., Chaudhari, M., ... Manetsch, R & Agar, J.N. (2018).

113

4.3.2 Synthesis of β-lipoic acid

The synthesis of β-lipoic acid was adapted from literature (Figure 40)(SAITO & FUKUI, 1967).

1 mL of aqueous hydrogen peroxide was added to starting material solution of DL-thioctic acid in acetone and allowed to stir for 24 hours. 40 mL of acidified chloroform was added to the reaction mixture. The solvent was removed under reduced pressure, diluted with dichloromethane. The solvent was removed under reduced pressure and purified by flash column chromatography on silica gel with 3% methanol/chloroform, 0.1 % acetic acid. The final product is concentrated under pressure to yield a colorless oil. β-lipoic acid formation was confirmed by NMR (Figure 41) and mass spectrometry (Figure 42). Chemical shifts matched with previously reported NMR spectra

(Müller, Knaack, & Olbrich, 1997). High-resolution electron spray ionization mass spectrometry

(HRMS; ESI-orbitrap) confirmed β-lipoic acid. HRMS (ESI-orbitrap) calculated for C8H14O3S2

[M + H]+ was 223.04571 Da; found 223.0451.

Figure 40. Synthesis of β-lipoic acid. Reproduced with permission Donnelly, D. P., Dowgiallo,

M. G., Salisbury, J. P., Aluri, K. C., Iyengar, S., Chaudhari, M., Mathew, M., Miele, I., Auclair, J.

R., Lopez, S, A., Manetsch, R & Agar, J.N. (2018).

114

115

Figure 41. 1H NMR of β-lipoic acid. a. Full 1H NMR of β-lipoic acid. b. Expanded 1H NMR of

β-lipoic acid. c. 13C NMR of β-lipoic acid. d. Expanded 13C NMR of β-lipoic acid.

116

Figure 42. High-resolution mass spectrum of β-lipoic acid.

117

4.3.3 Crosslinking of SOD1 monomers

To broaden the applicability of cyclic disulfide and cyclic thiosulfinate mediated cross-linking, α-

lipoic acid was purchased and β-lipoic acid was synthesized. Both α-lipoic acid and β-lipoic acid incubated with purified SOD1. The ability of α-lipoic acid and β-lipoic acid (mixture of isomers) to cross-link solvent-accessible cysteines 111 of SOD1 was evaluated by high-resolution mass spectrometry. α-lipoic acid cross-linked ~ 56% of SOD1 over 4650 minutes, while ~81% of

SOD1 was cross-linked by β-lipoic acid. Compared to α-lipoic acid, β-lipoic acid cross-linked 25

% more SOD1 in vitro (Figure 43). Cyclic disulfide reactivity, including reversible binding to lone thiols, is predictable and highly tunable. Cyclic thiosulfinate cross-linkers have potential as a less toxic alternative to Cys specific diene cross-linkers. The potential of β-lipoic acid, as well as other cyclic thiosulfinates, is being evaluated in vivo for cross-linking SOD1 variants.

Our data demonstrated the ability of cyclic thiosulfinates to avoid dead-end modifications while cross-linking closely spaced cysteines. We also demonstrated that the cyclic disulfides and cyclic thiosulfinates can cross-link SOD1. These molecules can provide a safer alternative for stabilizing

SOD1 variants in vivo.

118

0 min

30 min

60 min

140 min

488 min

1920 min

4650 min

Figure 43. Mass spectrometry assay of α-lipoic acid and β-lipoic acid cross-linking SOD1. Both rate and extent of covalent cross-linking of SOD1 is increased by oxidation of α-lipoic acid to β- lipoic acid. Representative deconvoluted mass spectra. The 31 894 Da molecular mass of the cross- linked dimer (D) supports the mechanism proposed in Figure 14 (e.g., two SOD1 monomers [2 ×

15 844 Da (M)] + β-lipoic acid [222 Da] − oxygen [16 Da]). Reproduced with permission

Donnelly, D. P., Dowgiallo, M. G., Salisbury, J. P., Aluri, K. C., Iyengar, S., Chaudhari, M., ...

Manetsch, R & Agar, J.N. (2018).

119

5. Conclusions of the Dissertation and Future Directions

In conclusion, we demonstrated 1) the etiology of ANG-ALS patients correlate with

physicochemical properties of ANG-ALS variants. 2) introduce cyclic thiosulfinates as a novel

class of cross-linkers that can efficiently cross-link SOD1dimers while avoiding dead-end thiol modifications and 3) cyclic thiosulfinates can be used novel class of cross-linkers for facile synthesis of safe and biodegradable hydrogels for drug and cell delivery.

Loss of stability of ANG correlates with disease onset and duration i.e. increased stability before ALS onset delays the onset by 15 years and reduced stability post-onset increases survival by 2 years. Similarly, loss of ANG activity i.e. ribonuclease activity after ALS onset increases the survival by 2 years. This hypothesis was validated using the G93A-SOD1 transgenic mice model.

Administration of increased dose of ANG mitigated the benefit of prolonged lifespan demonstrated by small doses of ANG. Future directions to this study are to evaluate the efficacy of ANG in

G93ASOD1 mice when delivered pre-symptomatically. Our data suggests a need for genotyping families with ANG-fALS and administration of ANG before symptom onset could prolong disease-free survival for an average of 15 years. We also propose ANG knockdown as a novel therapeutic strategy for the treatment of ANG-ALS after the onset of symptoms.

We demonstrated the application of cyclic thiosulfinates for the facile synthesis of hydrogels. At physiological pH, the cross-linking occurs within seconds after the addition of the six-membered cyclic thiosulfinate cross-linkers (1,2 dithiane-1-oxide). The presence of glutathione did not stop the hydrogel formation. These cyclic thiosulfinate cross-linked hydrogels are susceptible to glutathione (reduction) mediated degradation. Our in vitro experiments measuring cell viability demonstrated the cells did not show significant toxicity when grown in the presence of hydrogel.

120

Our encapsulation experiments demonstrated that our precursors demonstrate minimal toxicity while encapsulating the cells in the hydrogel during formation. We demonstrated these hydrogels are safe and can be employed for cell encapsulation and delivery. Finally, our hydrogels demonstrated controlled release via Fickian diffusion in proteins with no free cysteines. We demonstrated our hydrogels can successfully cross-link solvent-accessible cysteines in protein.

However, in this study, we did not optimize our hydrogels for the release of smaller proteins such as ANG. The future direction for this study is to study the safety of these hydrogels in vivo (in situ formation of the hydrogel in subcutaneous space to evaluate long term safety in rodents). In addition to six-membered cyclic thiosulfinates used in the study, other ring sizes such as 5 and 7 membered cyclic thiosulfinates should also be evaluated for applications in hydrogel synthesis.

Finally, using beta-lipoic acid we demonstrated that the cyclic thiosulfinates can be used to efficiently crosslink Cys111 of SOD1 monomers and provide a viable therapeutic strategy for the treatment of SOD1-ALS. We showed cyclic thiosulfinates crosslink at faster rates compared to cyclic disulfides. The ring strain of the cyclic thiosulfinates makes them more selective cross- linkers avoiding dead-end modifications making them safer alternatives to currently available more toxic cross-linkers such as maleimides. The future directions for this project are to evaluate the potency of cyclic thiosulfinates in stabilizing SOD1 in G93ASOD1 mice and evaluate the toxicity of these molecules.

121

6. References

Abdolvahabi, A., Shi, Y., Rasouli, S., Croom, C. M., Aliyan, A., Marti, A. A., & Shaw, B. F.

(2017). Kaplan-Meier Meets Chemical Kinetics: Intrinsic Rate of SOD1 Amyloidogenesis

Decreased by Subset of ALS Mutations and Cannot Fully Explain Age of Disease Onset. ACS

Chem Neurosci, 8(6), 1378-1389. doi:10.1021/acschemneuro.7b00029

Abegg, D., Gasparini, G., Hoch, D. G., Shuster, A., Bartolami, E., Matile, S., & Adibekian, A.

(2017). Strained cyclic disulfides enable cellular uptake by reacting with the transferrin receptor.

Journal of the American Chemical Society, 139(1), 231-238.

Abeysekera, W., & Sooriyarachchi, R. (2009). Use of Schoenfeld’s global test to test the

proportional hazards assumption in the Cox proportional hazards model: an application to a

clinical study.

Acharya, K. R., Shapiro, R., Allen, S. C., Riordan, J. F., & Vallee, B. L. (1994). Crystal structure

of human angiogenin reveals the structural basis for its functional divergence from ribonuclease.

Proceedings of the National Academy of Sciences, 91(8), 2915-2919.

Ajroud-Driss, S., & Siddique, T. (2015). Sporadic and hereditary amyotrophic lateral sclerosis

(ALS). Biochimica Et Biophysica Acta (BBA)-Molecular Basis of Disease, 1852(4), 679-684.

Alami, N. O., Schurr, C., Heuvel, F. O., Tang, L., Li, Q., Tasdogan, A., . . . Raposo, C. (2018).

NF‐κB activation in astrocytes drives a stage‐specific beneficial neuroimmunological response in

ALS. The EMBO journal, 37(16).

Alemasov, N. A., Ivanisenko, N. V., Taneja, B., Taneja, V., Ramachandran, S., Ivanisenko, V.

A. J. J. o. M. G., & Modelling. (2019). Improved regression model to predict an impact of SOD1

122

mutations on ALS patients survival time based on analysis of hydrogen bond stability. 86, 247-

255.

Ananikov, V. P., Gayduk, K. A., Beletskaya, I. P., Khrustalev, V. N., & Antipin, M. Y. (2009).

Catalyst leaching as an efficient tool for constructing new catalytic reactions: Application to the

synthesis of cyclic vinyl sulfides and vinyl selenides. European Journal of Inorganic Chemistry,

2009(9), 1149-1161.

Andersen, P. M. (2006). Amyotrophic lateral sclerosis associated with mutations in the CuZn

superoxide dismutase gene. Current neurology and neuroscience reports, 6(1), 37-46.

Aparicio-Erriu, I. M., & Prehn, J. H. (2012). Molecular Mechanisms in Amyotrophic Lateral

Sclerosis: The Role of Angiogenin, a Secreted RNase. Front Neurosci, 6, 167.

doi:10.3389/fnins.2012.00167

Auclair, J. R., Boggio, K. J., Petsko, G. A., Ringe, D., & Agar, J. N. (2010). Strategies for

stabilizing superoxide dismutase (SOD1), the protein destabilized in the most common form of

familial amyotrophic lateral sclerosis. Proceedings of the National Academy of Sciences,

107(50), 21394-21399.

Austin, J. A., Wright, G. S., Watanabe, S., Grossmann, J. G., Antonyuk, S. V., Yamanaka, K., &

Hasnain, S. S. (2014). Disease causing mutants of TDP-43 nucleic acid binding domains are resistant to aggregation and have increased stability and half-life. Proc Natl Acad Sci U S A,

111(11), 4309-4314. doi:10.1073/pnas.1317317111

Baillie, T. A. (2016). Targeted covalent inhibitors for drug design. Angewandte Chemie

International Edition, 55(43), 13408-13421.

Bakaic, E., Smeets, N. M., & Hoare, T. (2015). Injectable hydrogels based on poly (ethylene glycol) and derivatives as functional biomaterials. RSC Advances, 5(45), 35469-35486.

123

Bannwarth, S., Ait-El-Mkadem, S., Chaussenot, A., Genin, E. C., Lacas-Gervais, S., Fragaki, K.,

. . . Moore, D. G. (2014). A mitochondrial origin for frontotemporal dementia and amyotrophic

lateral sclerosis through CHCHD10 involvement. Brain, 137(8), 2329-2345.

Beckman, J. S., Carson, M., Smith, C. D., & Koppenol, W. H. (1993). ALS, SOD and

peroxynitrite. Nature, 364(6438), 584.

Belli, M., Ramazzotti, M., & Chiti, F. (2011). Prediction of amyloid aggregation in vivo. EMBO

Rep, 12(7), 657-663. doi:10.1038/embor.2011.116

embor2011116 [pii]

Bertram, L., & Tanzi, R. E. J. T. J. o. c. i. (2005). The genetic epidemiology of

neurodegenerative disease. 115(6), 1449-1457.

Bhattarai, N., Gunn, J., & Zhang, M. (2010). Chitosan-based hydrogels for controlled, localized

drug delivery. Advanced drug delivery reviews, 62(1), 83-99.

Blanco, S., Dietmann, S., Flores, J. V., Hussain, S., Kutter, C., Humphreys, P., . . . Frye, M.

(2014). Aberrant methylation of tRNAs links cellular stress to neuro-developmental disorders.

EMBO J, 33(18), 2020-2039. doi:10.15252/embj.201489282 embj.201489282 [pii]

Boillee, S., Vande Velde, C., & Cleveland, D. W. (2006). ALS: a disease of motor neurons and their nonneuronal neighbors. Neuron, 52(1), 39-59. doi:S0896-6273(06)00725-2 [pii]

10.1016/j.neuron.2006.09.018

Borchelt, D. R., Lee, M. K., Slunt, H. S., Guarnieri, M., Xu, Z. S., Wong, P. C., . . . Cleveland,

D. W. (1994). Superoxide dismutase 1 with mutations linked to familial amyotrophic lateral sclerosis possesses significant activity. Proc Natl Acad Sci U S A, 91(17), 8292-8296.

124

Bosco, D. A., Morfini, G., Karabacak, N. M., Song, Y., Gros-Louis, F., Pasinelli, P., . . .

McKenna-Yasek, D. (2010). Wild-type and mutant SOD1 share an aberrant conformation and a common pathogenic pathway in ALS. Nature neuroscience, 13(11), 1396.

Bradburn, M. J., Clark, T. G., Love, S., & Altman, D. (2003). Survival analysis part II: multivariate data analysis–an introduction to concepts and methods. British journal of cancer,

89(3), 431.

Bradshaw, W. J., Rehman, S., Pham, T. T., Thiyagarajan, N., Lee, R. L., Subramanian, V., &

Acharya, K. R. (2017). Structural insights into human angiogenin variants implicated in

Parkinson's disease and Amyotrophic Lateral Sclerosis. Sci Rep, 7, 41996. doi:10.1038/srep41996 srep41996 [pii]

Breslow, N. E., & Day, N. E. (1980). Statistical methods in cancer research. Volume I - The analysis of case-control studies. IARC Sci Publ(32), 5-338.

Brooks, B. R., Miller, R. G., Swash, M., & Munsat, T. L. (2000). El Escorial revisited: revised criteria for the diagnosis of amyotrophic lateral sclerosis. Amyotrophic lateral sclerosis and other motor neuron disorders, 1(5), 293-299.

Burdick, J. A., & Prestwich, G. D. J. A. m. (2011). Hyaluronic acid hydrogels for biomedical applications. 23(12), H41-H56.

Burns, J. A., & Whitesides, G. M. (1990). Predicting the stability of cyclic disulfides by molecular modeling: Effective concentrations in thiol-disulfide interchange and the design of strongly reducing dithiols. Journal of the American Chemical Society, 112(17), 6296-6303.

125

Burns, J. N., Turnage, K. C., Walker, C. A., & Lieberman, R. L. (2011). The stability of

myocilin olfactomedin domain variants provides new insight into glaucoma as a protein

misfolding disorder. Biochemistry, 50(26), 5824-5833. doi:10.1021/bi200231x

Cai, X. Y., Li, J. Z., Li, N. N., Chen, J. C., Kang, E.-T., & Xu, L. Q. (2016). PEG-based

hydrogels prepared by catalyst-free thiol–yne addition and their post-antibacterial modification.

Biomaterials science, 4(11), 1663-1672.

Caliari, S. R., & Burdick, J. A. (2016). A practical guide to hydrogels for cell culture. Nature methods, 13(5), 405.

Chai, Q., Jiao, Y., & Yu, X. (2017). Hydrogels for biomedical applications: their characteristics and the mechanisms behind them. Gels, 3(1), 6.

Chao, Y., Li, C.-P., Chau, G.-Y., Chen, C.-P., King, K.-L., Lui, W.-Y., . . . Lee, S.-D. (2003).

Prognostic significance of vascular endothelial growth factor, basic fibroblast growth factor, and angiogenin in patients with resectable hepatocellular carcinoma after surgery. Annals of surgical

oncology, 10(4), 355-362.

Charcot, J.-M. (1869). Deux cas d'atrophie musculaire progressive avec lesions de la substance

grice et des faisceaux anterolateraux de la moelle epiniere. Arch Physiol Norm Pathol, 2,

354,629,744.

Chen, M.-L., Wu, R.-M., Tai, C.-H., & Lin, C.-H. (2014). Mutational analysis of angiogenin

gene in Parkinson's disease. PLoS One, 9(11), e112661.

Chen, S., Sayana, P., Zhang, X., & Le, W. (2013). Genetics of amyotrophic lateral sclerosis: an

update. Molecular neurodegeneration, 8(1), 28.

Chiti, F., & Dobson, C. M. (2006). Protein misfolding, functional amyloid, and human disease.

Annu Rev Biochem, 75, 333-366. doi:10.1146/annurev.biochem.75.101304.123901

126

Chiti, F., Stefani, M., Taddei, N., Ramponi, G., & Dobson, C. M. (2003). Rationalization of the effects of mutations on peptide andprotein aggregation rates. Nature, 424(6950), 805.

Choh, S. Y., Cross, D., & Wang, C. (2011). Facile synthesis and characterization of disulfide- cross-linked hyaluronic acid hydrogels for protein delivery and cell encapsulation.

Biomacromolecules, 12(4), 1126-1136. doi:10.1021/bm101451k

Clark, T., Bradburn, M., Love, S., & Altman, D. (2003). Survival analysis part I: basic concepts and first analyses. British journal of cancer, 89(2), 232.

Coelho, T., Maia, L. F., da Silva, A. M., Cruz, M. W., Planté-Bordeneuve, V., Lozeron, P., . . .

Schmidt, H. H.-J. (2012). Tafamidis for transthyretin familial amyloid polyneuropathy: a randomized, controlled trial. Neurology, 79(8), 785-792.

Coelho, T., Merlini, G., Bulawa, C. E., Fleming, J. A., Judge, D. P., Kelly, J. W., . . . Mundayat,

R. (2016). Mechanism of action and clinical application of tafamidis in hereditary transthyretin amyloidosis. Neurology and therapy, 5(1), 1-25.

Conforti, F. L., Sprovieri, T., Mazzei, R., Ungaro, C., La Bella, V., Tessitore, A., . . . Quattrone,

A. (2008). A novel Angiogenin gene mutation in a sporadic patient with amyotrophic lateral sclerosis from southern Italy. Neuromuscul Disord, 18(1), 68-70. doi:S0960-8966(07)00676-1

[pii]

10.1016/j.nmd.2007.07.003

Conway, K. A., Lee, S.-J., Rochet, J.-C., Ding, T. T., Williamson, R. E., & Lansbury, P. T.

(2000). Acceleration of oligomerization, not fibrillization, is a shared property of both α- synuclein mutations linked to early-onset Parkinson's disease: implications for pathogenesis and therapy. Proceedings of the National Academy of Sciences, 97(2), 571-576.

127

Cornell, P. (2006). Beginning Excel What-If Data Analysis Tools: Getting Started with Goal

Seek, Data Tables, Scenarios, and Solver (Vol. 72): Apress.

Cox, D. R. (1972). Regression Models and Life-Tables. Journal of the Royal Statistical Society.

Series B (Methodological), 34(2), 187-220.

Crabtree, B., Thiyagarajan, N., Prior, S. H., Wilson, P., Iyer, S., Ferns, T., . . . Acharya, K. R.

(2007). Characterization of human angiogenin variants implicated in amyotrophic lateral sclerosis. Biochemistry, 46(42), 11810-11818. doi:10.1021/bi701333h

Crivello, M., O'Riordan, S. L., Woods, I., Cannon, S., Halang, L., Coughlan, K. S., . . . Prehn, J.

H. M. (2018). Pleiotropic activity of systemically delivered angiogenin in the SOD1(G93A) mouse model. Neuropharmacology, 133, 503-511. doi:10.1016/j.neuropharm.2018.02.022

Crosio, C., Casciati, A., Iaccarino, C., Rotilio, G., & Carri, M. T. (2006). Bcl2a1 serves as a switch in death of motor neurons in amyotrophic lateral sclerosis. Cell Death Differ, 13(12),

2150-2153. doi:4401943 [pii]

10.1038/sj.cdd.4401943

Crow, J. P., Sampson, J. B., Zhuang, Y., Thompson, J. A., & Beckman, J. S. (1997). Decreased zinc affinity of amyotrophic lateral sclerosis‐associated superoxide dismutase mutants leads to enhanced catalysis of tyrosine nitration by peroxynitrite. Journal of neurochemistry, 69(5), 1936-

1944.

Cunningham, A. D., Colavin, A., Huang, K. C., & Mochly-Rosen, D. (2017). Coupling between

Protein Stability and Catalytic Activity Determines Pathogenicity of G6PD Variants. Cell Rep,

18(11), 2592-2599. doi:S2211-1247(17)30246-2 [pii]

10.1016/j.celrep.2017.02.048

128

de Groot, N. S., Aviles, F. X., Vendrell, J., & Ventura, S. (2006). Mutagenesis of the central

hydrophobic cluster in Aβ42 Alzheimer's peptide: Side‐chain properties correlate with

aggregation propensities. The FEBS journal, 273(3), 658-668.

Deng, H. X., Chen, W., Hong, S. T., Boycott, K. M., Gorrie, G. H., Siddique, N., . . . Siddique,

T. (2011). Mutations in UBQLN2 cause dominant X-linked juvenile and adult-onset ALS and

ALS/dementia. Nature, 477(7363), 211-215. doi:10.1038/nature10353

nature10353 [pii]

Donnelly, D. P., Agar, J. N., & Lopez, S. A. (2019). Nucleophilic substitution reactions of cyclic

thiosulfinates are accelerated by hyperconjugative interactions. Chemical Science, 10(21), 5568-

5575.

Donnelly, D. P., Dowgiallo, M. G., Salisbury, J. P., Aluri, K. C., Iyengar, S., Chaudhari, M., . . .

Agar, J. N. (2018). Cyclic Thiosulfinates and Cyclic Disulfides Selectively Cross-Link Thiols

While Avoiding Modification of Lone Thiols. J Am Chem Soc, 140(24), 7377-7380.

doi:10.1021/jacs.8b01136

Doucette, P. A., Whitson, L. J., Cao, X., Schirf, V., Demeler, B., Valentine, J. S., . . . Hart, P. J.

(2004). Dissociation of Human Copper-Zinc Superoxide Dismutase Dimers Using Chaotrope and Reductant INSIGHTS INTO THE MOLECULAR BASIS FOR DIMER STABILITY.

Journal of Biological Chemistry, 279(52), 54558-54566.

Drury, J. L., & Mooney, D. J. (2003). Hydrogels for tissue engineering: scaffold design variables and applications. Biomaterials, 24(24), 4337-4351.

Durham, H., Kabashi, E., Taylor, D., & Agar, J. (2007). The proteasome in neurodegeneration.

Motor Neuron Disease. New York: Springer, 247-264.

129

Duyao, M., Ambrose, C., Myers, R., Novelletto, A., Persichetti, F., Frontali, M., . . . et al. (1993).

Trinucleotide repeat length instability and age of onset in Huntington's disease. Nat Genet, 4(4),

387-392. doi:10.1038/ng0893-387

Ederer, F., Axtell, L. M., & Cutler, S. J. (1961). The relative survival rate: a statistical methodology. Natl Cancer Inst Monogr, 6, 101-121.

Elden, A. C., Kim, H.-J., Hart, M. P., Chen-Plotkin, A. S., Johnson, B. S., Fang, X., . . . Lu, M.

M. (2010). Ataxin-2 intermediate-length polyglutamine expansions are associated with increased risk for ALS. Nature, 466(7310), 1069.

Estrada, L. H., & Champion, J. J. B. s. (2015). Protein nanoparticles for therapeutic protein delivery. 3(6), 787-799.

Etoh, T., Shibuta, K., Barnard, G. F., Kitano, S., & Mori, M. (2000). Angiogenin expression in human colorectal cancer: the role of focal macrophage infiltration. Clinical Cancer Research,

6(9), 3545-3551.

Fedde, K. N., Michell, M. P., Henthorn, P. S., & Whyte, M. P. (1996). Aberrant properties of alkaline phosphatase in patient fibroblasts correlate with clinical expressivity in severe forms of hypophosphatasia. The Journal of Clinical Endocrinology & Metabolism, 81(7), 2587-2594.

Fernandez-Santiago, R., Hoenig, S., Lichtner, P., Sperfeld, A. D., Sharma, M., Berg, D., . . .

Ludolph, A. (2009). Identification of novel Angiogenin (ANG) gene missense variants in

German patients with amyotrophic lateral sclerosis. J Neurol, 256(8), 1337-1342. doi:10.1007/s00415-009-5124-4

Fett, J. W., Strydom, D. J., Lobb, R. R., Alderman, E. M., Bethune, J. L., Riordan, J. F., &

Vallee, B. L. (1985). Isolation and characterization of angiogenin, an angiogenic protein from human carcinoma cells. Biochemistry, 24(20), 5480-5486.

130

Forman, H.-J., & Fridovich, I. (1973). On the stability of bovine superoxide dismutase. J. Biol.

Chem, 248, 2645-2649.

Frakes, A. E., Ferraiuolo, L., Haidet-Phillips, A. M., Schmelzer, L., Braun, L., Miranda, C. J., . .

. Kaspar, B. K. (2014). Microglia induce motor neuron death via the classical NF-kappaB pathway in amyotrophic lateral sclerosis. Neuron, 81(5), 1009-1023. doi:S0896-6273(14)00018-

X [pii]

10.1016/j.neuron.2014.01.013

Freischmidt, A., Wieland, T., Richter, B., Ruf, W., Schaeffer, V., Müller, K., . . . Weydt, P.

(2015). Haploinsufficiency of TBK1 causes familial ALS and fronto-temporal dementia. Nature neuroscience, 18(5), 631.

Fridovich, I. (1986). Superoxide dismutases. Adv Enzymol Relat Areas Mol Biol, 58(6), 61-97.

Futami, J., TSUSHIMA, Y., MURATO, Y., TADA, H., SASAKI, J., SENO, M., & YAMADA,

H. (1997). Tissue-specific expression of pancreatic-type RNases and RNase inhibitor in humans.

DNA and cell biology, 16(4), 413-419.

Ganji, F., Vasheghani, F. S., & VASHEGHANI, F. E. (2010). Theoretical description of hydrogel swelling: a review.

Gao, X., & Xu, Z. (2008). Mechanisms of action of angiogenin. Acta Biochim Biophys Sin

(Shanghai), 40(7), 619-624.

Garcia, A. J. (2014). PEG-maleimide hydrogels for protein and cell delivery in regenerative medicine. Ann Biomed Eng, 42(2), 312-322. doi:10.1007/s10439-013-0870-y

Gellera, C., Colombrita, C., Ticozzi, N., Castellotti, B., Bragato, C., Ratti, A., . . . Silani, V.

(2008). Identification of new ANG gene mutations in a large cohort of Italian patients with amyotrophic lateral sclerosis. Neurogenetics, 9(1), 33-40.

131

Gendron, T. F., Belzil, V. V., Zhang, Y.-J., & Petrucelli, L. (2014). Mechanisms of toxicity in

C9FTLD/ALS. Acta neuropathologica, 127(3), 359-376.

Gieselmann, V. (2005). What can cell biology tell us about heterogeneity in lysosomal storage

diseases? Acta Paediatr Suppl, 94(447), 80-86; discussion 79.

Goel, M. K., Khanna, P., & Kishore, J. (2010). Understanding survival analysis: Kaplan-Meier

estimate. International journal of Ayurveda research, 1(4), 274.

Goncalves, K. A., Silberstein, L., Li, S., Severe, N., Hu, M. G., Yang, H., . . . Hu, G. F. (2016).

Angiogenin Promotes Hematopoietic Regeneration by Dichotomously Regulating Quiescence of

Stem and Progenitor Cells. Cell, 166(4), 894-906. doi:10.1016/j.cell.2016.06.042

Goode, A., Butler, K., Long, J., Cavey, J., Scott, D., Shaw, B., . . . Oldham, N. J. (2016).

Defective recognition of LC3B by mutant SQSTM1/p62 implicates impairment of autophagy as

a pathogenic mechanism in ALS-FTLD. Autophagy, 12(7), 1094-1104.

Gorrie, G. H., Fecto, F., Radzicki, D., Weiss, C., Shi, Y., Dong, H., . . . Li, S. (2014). Dendritic

spinopathy in transgenic mice expressing ALS/dementia-linked mutant UBQLN2. Proceedings of the National Academy of Sciences, 111(40), 14524-14529.

Greenway, M. J., Alexander, M. D., Ennis, S., Traynor, B. J., Corr, B., Frost, E., . . . Hardiman,

O. (2004). A novel candidate region for ALS on chromosome 14q11.2. Neurology, 63(10), 1936-

1938. doi:63/10/1936 [pii]

Greenway, M. J., Andersen, P. M., Russ, C., Ennis, S., Cashman, S., Donaghy, C., . . . Hardiman,

O. (2006). ANG mutations segregate with familial and 'sporadic' amyotrophic lateral sclerosis.

Nat Genet, 38(4), 411-413. doi:ng1742 [pii]

10.1038/ng1742

132

Gurney, M. E., Pu, H., Chiu, A. Y., Dal Canto, M. C., Polchow, C. Y., Alexander, D. D., . . .

Deng, H.-X. (1994). Motor neuron degeneration in mice that express a human Cu, Zn superoxide

dismutase mutation. Science, 264(5166), 1772-1775.

Hart, M. P., & Gitler, A. D. (2012). ALS-associated ataxin 2 polyQ expansions enhance stress- induced caspase 3 activation and increase TDP-43 pathological modifications. Journal of

Neuroscience, 32(27), 9133-9142.

Hayward, C., Colville, S., Swingler, R. J., & Brock, D. J. (1999). Molecular genetic analysis of the APEX nuclease gene in amyotrophic lateral sclerosis. Neurology, 52(9), 1899-1899.

Hennink, W. E., & van Nostrum, C. F. (2012). Novel crosslinking methods to design hydrogels.

Advanced drug delivery reviews, 64, 223-236.

Hoare, T. R., & Kohane, D. S. (2008). Hydrogels in drug delivery: Progress and challenges.

Polymer, 49(8), 1993-2007.

Holloway, D. E., Hares, M. C., Shapiro, R., Subramanian, V., & Acharya, K. R. (2001). High- level expression of three members of the murine angiogenin family in Escherichia coli and purification of the recombinant proteins. Protein expression and purification, 22(2), 307-317.

Hörnberg, A., Logan, D. T., Marklund, S. L., & Oliveberg, M. (2007). The coupling between disulphide status, metallation and dimer interface strength in Cu/Zn superoxide dismutase.

Journal of molecular biology, 365(2), 333-342.

Hough, M. A., Grossmann, J. G., Antonyuk, S. V., Strange, R. W., Doucette, P. A., Rodriguez, J.

A., . . . Valentine, J. S. (2004). Dimer destabilization in superoxide dismutase may result in disease-causing properties: structures of motor neuron disease mutants. Proceedings of the

National Academy of Sciences, 101(16), 5976-5981.

133

Hoyle, C. E., Lowe, A. B., & Bowman, C. N. J. C. S. R. (2010). Thiol-click chemistry: a multifaceted toolbox for small molecule and polymer synthesis. 39(4), 1355-1387.

Ingre, C., Roos, P. M., Piehl, F., Kamel, F., & Fang, F. (2015). Risk factors for amyotrophic lateral sclerosis. Clin Epidemiol, 7, 181-193. doi:10.2147/CLEP.S37505

Ishii, D., Ying, T. H., Mahara, A., Murakami, S., Yamaoka, T., Lee, W.-k., & Iwata, T. (2008).

In vivo tissue response and degradation behavior of PLLA and stereocomplexed PLA nanofibers.

Biomacromolecules, 10(2), 237-242.

Jensen, L. H., Renodon-Corniere, A., Wessel, I., Langer, S. W., Søkilde, B., Carstensen, E. V., . .

. Jensen, P. B. (2002). Maleimide is a potent inhibitor of topoisomerase II in vitro and in vivo: a new mode of catalytic inhibition. Molecular pharmacology, 61(5), 1235-1243.

Johnson, J. O., Mandrioli, J., Benatar, M., Abramzon, Y., Van Deerlin, V. M., Trojanowski, J.

Q., . . . Wuu, J. (2010). Exome sequencing reveals VCP mutations as a cause of familial ALS.

Neuron, 68(5), 857-864.

Johnson, J. O., Pioro, E. P., Boehringer, A., Chia, R., Feit, H., Renton, A. E., . . . Winborn, B. J.

(2014). Mutations in the Matrin 3 gene cause familial amyotrophic lateral sclerosis. Nature neuroscience, 17(5), 664.

Johnson, S. M., Wiseman, R. L., Sekijima, Y., Green, N. S., Adamski-Werner, S. L., & Kelly, J.

W. (2005). Native state kinetic stabilization as a strategy to ameliorate protein misfolding diseases: a focus on the transthyretin amyloidoses. Accounts of chemical research, 38(12), 911-

921.

Kabashi, E., Agar, J. N., Strong, M. J., & Durham, H. D. (2012). Impaired proteasome function in sporadic amyotrophic lateral sclerosis. Amyotrophic Lateral Sclerosis, 13(4), 367-371.

134

Kabashi, E., Agar, J. N., Taylor, D. M., Minotti, S., & Durham, H. D. (2004). Focal dysfunction of the proteasome: a pathogenic factor in a mouse model of amyotrophic lateral sclerosis.

Journal of neurochemistry, 89(6), 1325-1335.

Kaplan, E. L., & Meier, P. (1958). Nonparametric estimation from incomplete observations.

Journal of the American statistical association, 53(282), 457-481.

Kaur, S. J., McKeown, S. R., & Rashid, S. (2016). Mutant SOD1 mediated pathogenesis of amyotrophic lateral sclerosis. Gene, 577(2), 109-118.

Kendall, M. G. (1938). A New Measure of Rank Correlation. Biometrika, 30(1/2), 81-93. doi:10.2307/2332226

Kharkar, P. M., Rehmann, M. S., Skeens, K. M., Maverakis, E., & Kloxin, A. M. (2016). Thiol– ene click hydrogels for therapeutic delivery. ACS biomaterials science & engineering, 2(2), 165-

179.

Kieran, D., Sebastia, J., Greenway, M. J., King, M. A., Connaughton, D., Concannon, C. G., . . .

Prehn, J. H. (2008). Control of motoneuron survival by angiogenin. J Neurosci, 28(52), 14056-

14061. doi:10.1523/JNEUROSCI.3399-08.2008

28/52/14056 [pii]

Kiernan, M. C., Vucic, S., Cheah, B. C., Turner, M. R., Eisen, A., Hardiman, O., . . . Zoing, M.

C. (2011). Amyotrophic lateral sclerosis. The lancet, 377(9769), 942-955.

Kim, H. J., Kim, N. C., Wang, Y.-D., Scarborough, E. A., Moore, J., Diaz, Z., . . . Molliex, A.

(2013). Mutations in prion-like domains in hnRNPA2B1 and hnRNPA1 cause multisystem proteinopathy and ALS. Nature, 495(7442), 467.

Kim, H. M., Kang, D. K., Kim, H. Y., Kang, S. S., & Chang, S. I. (2007). Angiogenin-induced protein kinase B/Akt activation is necessary for angiogenesis but is independent of nuclear

135 translocation of angiogenin in HUVE cells. Biochem Biophys Res Commun, 352(2), 509-513. doi:S0006-291X(06)02519-8 [pii]

10.1016/j.bbrc.2006.11.047

Kirby, J., Highley, J. R., Cox, L., Goodall, E. F., Hewitt, C., Hartley, J. A., . . . Shaw, P. J.

(2013). Lack of unique neuropathology in amyotrophic lateral sclerosis associated with p.K54E angiogenin (ANG) mutation. Neuropathol Appl Neurobiol, 39(5), 562-571. doi:10.1111/nan.12007

Kishikawa, H., Wu, D., & Hu, G. F. (2008). Targeting angiogenin in therapy of amyotropic lateral sclerosis. Expert Opin Ther Targets, 12(10), 1229-1242. doi:10.1517/14728222.12.10.1229

Kobsa, S., & Saltzman, W. M. J. P. r. (2008). Bioengineering approaches to controlled protein delivery. 63(5), 513-519.

Krebs, M. D., Jeon, O., & Alsberg, E. (2009). Localized and sustained delivery of silencing RNA from macroscopic biopolymer hydrogels. Journal of the American chemical Society, 131(26),

9204-9206.

Kuh, D., & Shlomo, Y. B. (2004). A life course approach to chronic disease epidemiology:

Oxford University Press.

Lafleur, M. V. M., Hoorweg, J. J., Joenje, H., Westmijze, E. J., & Retèl, J. (1994). The ambivalent role of glutathione in the protection of DNA against singlet oxygen. Free radical research, 21(1), 9-17.

Lattante, S., Ciura, S., Rouleau, G. A., & Kabashi, E. (2015). Defining the genetic connection linking amyotrophic lateral sclerosis (ALS) with frontotemporal dementia (FTD). Trends in

Genetics, 31(5), 263-273.

136

Lee, K. Y., & Mooney, D. J. (2001). Hydrogels for tissue engineering. Chemical reviews, 101(7),

1869-1880.

Lin, C.-C., & Anseth, K. S. (2009). PEG hydrogels for the controlled release of biomolecules in

regenerative medicine. Pharmaceutical research, 26(3), 631-643.

Lin, C.-C., & Anseth, K. S. J. P. r. (2009). PEG hydrogels for the controlled release of

biomolecules in regenerative medicine. 26(3), 631-643.

Lin, C.-C., & Metters, A. T. (2006). Hydrogels in controlled release formulations: network design and mathematical modeling. Advanced drug delivery reviews, 58(12-13), 1379-1408.

Lomen-Hoerth, C., Anderson, T., & Miller, B. (2002). The overlap of amyotrophic lateral

sclerosis and frontotemporal dementia. Neurology, 59(7), 1077-1079.

Lowe, A. B. (2010). Thiol-ene “click” reactions and recent applications in polymer and materials

synthesis. Polymer Chemistry, 1(1), 17-36.

Luheshi, L. M., Tartaglia, G. G., Brorsson, A. C., Pawar, A. P., Watson, I. E., Chiti, F., . . .

Crowther, D. C. (2007). Systematic in vivo analysis of the intrinsic determinants of amyloid Beta

pathogenicity. PLoS Biol, 5(11), e290. doi:07-PLBI-RA-2184 [pii]

10.1371/journal.pbio.0050290

Luigetti, M., Lattante, S., Zollino, M., Conte, A., Marangi, G., Del Grande, A., & Sabatelli, M.

(2011). SOD1 G93D sporadic amyotrophic lateral sclerosis (SALS) patient with rapid

progression and concomitant novel ANG variant. Neurobiol Aging, 32(10), 1924 e1915-1928.

doi:10.1016/j.neurobiolaging.2011.04.004

S0197-4580(11)00119-9 [pii]

Luo, Y., Kirker, K. R., & Prestwich, G. D. (2000). Cross-linked hyaluronic acid hydrogel films:

new biomaterials for drug delivery. Journal of controlled release, 69(1), 169-184.

137

Maddika, S., Ande, S. R., Wiechec, E., Hansen, L. L., Wesselborg, S., & Los, M. (2008). Akt-

mediated phosphorylation of CDK2 regulates its dual role in cell cycle progression and

apoptosis. J Cell Sci, 121(Pt 7), 979-988. doi:10.1242/jcs.009530

121/7/979 [pii]

Maes, P., Damart, D., Rommens, C., Montreuil, J., Spik, G., & Tartar, A. (1988). The complete

amino acid sequence of bovine milk angiogenin. FEBS letters, 241(1-2), 41-45.

Majee, S. B. (2016). Emerging concepts in analysis and applications of hydrogels: BoD–Books on Demand.

Mantel, N. (1966). Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep, 50(3), 163-170.

Maruyama, H., Morino, H., Ito, H., Izumi, Y., Kato, H., Watanabe, Y., . . . Suzuki, H. (2010).

Mutations of optineurin in amyotrophic lateral sclerosis. Nature, 465(7295), 223.

Mayeux, R. J. A. r. o. n. (2003). Epidemiology of neurodegeneration. 26(1), 81-104.

McLaughlin, R. L., Phukan, J., McCormack, W., Lynch, D. S., Greenway, M., Cronin, S., . . .

Hardiman, O. (2010). Angiogenin levels and ANG genotypes: dysregulation in amyotrophic lateral sclerosis. PLoS One, 5(11), e15402. doi:10.1371/journal.pone.0015402

Millecamps, S., Salachas, F., Cazeneuve, C., Gordon, P., Bricka, B., Camuzat, A., . . .

Meininger, V. (2010). SOD1, ANG, VAPB, TARDBP, and FUS mutations in familial amyotrophic lateral sclerosis: genotype-phenotype correlations. J Med Genet, 47(8), 554-560. doi:10.1136/jmg.2010.077180 jmg.2010.077180 [pii]

138

Moenner, M., Gusse, M., Hatzi, E., & Badet, J. (1994). The widespread expression of angiogenin

in different human cells suggests a biological function not only related to angiogenesis.

European journal of biochemistry, 226(2), 483-490.

Molnar, K. S., Karabacak, N. M., Johnson, J. L., Wang, Q., Tiwari, A., Hayward, L. J., . . . Agar,

J. N. (2009). A common property of amyotrophic lateral sclerosis-associated variants

destabilization of the copper/zinc superoxide dismutase electrostatic loop. Journal of Biological

Chemistry, 284(45), 30965-30973.

Morgan, S., & Orrell, R. W. (2016). Pathogenesis of amyotrophic lateral sclerosis. British

medical bulletin, 119(1), 87-97.

Müller, A., Knaack, M., & Olbrich, A. (1997). NMR Spectroscopic Characterization of Isomeric

S‐Oxides Derived from α‐Lipoic Acid. Magnetic resonance in chemistry, 35(2), 111-114.

Nair, D. P., Podgorski, M., Chatani, S., Gong, T., Xi, W., Fenoli, C. R., & Bowman, C. N.

(2013). The thiol-Michael addition click reaction: a powerful and widely used tool in materials

chemistry. Chemistry of Materials, 26(1), 724-744.

Neufeld, B. H., Tapia, J. B., Lutzke, A., & Reynolds, M. M. (2018). Small molecule interferences in resazurin and MTT-based metabolic assays in the absence of cells. Analytical

chemistry, 90(11), 6867-6876.

Oakes, J. A., Davies, M. C., & Collins, M. O. (2017). TBK1: a new player in ALS linking

autophagy and neuroinflammation. Molecular brain, 10(1), 5.

Orlacchio, A., Babalini, C., Borreca, A., Patrono, C., Massa, R., Basaran, S., . . . Bernardi, G.

(2010). SPATACSIN mutations cause autosomal recessive juvenile amyotrophic lateral

sclerosis. Brain, 133(2), 591-598.

139

Padhi, A. K., Jayaram, B., & Gomes, J. (2013). Prediction of functional loss of human

angiogenin mutants associated with ALS by molecular dynamics simulations. Sci Rep, 3, 1225.

doi:10.1038/srep01225

Palmer, K., Scheraga, H., Riordan, J., & Vallee, B. (1986). A preliminary three-dimensional structure of angiogenin. Proceedings of the National Academy of Sciences, 83(7), 1965-1969.

Pan, L., Deng, X., Ding, D., Leng, H., Zhu, X., & Wang, Z. (2015). Association between the

Angiogenin (ANG) K17I variant and amyotrophic lateral sclerosis risk in Caucasian: a meta-

analysis. Neurological Sciences, 36(12), 2163-2168.

Parkinson, N., Ince, P., Smith, M., Highley, R., Skibinski, G., Andersen, P., . . . Collinge, J.

(2006). ALS phenotypes with mutations in CHMP2B (charged multivesicular body protein 2B).

Neurology, 67(6), 1074-1077.

Paubel, A., Violette, J., Amy, M., Praline, J., Meininger, V., Camu, W., . . . Vourc'h, P. (2008).

Mutations of the ANG gene in French patients with sporadic amyotrophic lateral sclerosis. Arch

Neurol, 65(10), 1333-1336. doi:10.1001/archneur.65.10.1333

65/10/1333 [pii]

Perry, J., Shin, D., Getzoff, E., & Tainer, J. (2010). The structural biochemistry of the superoxide

dismutases. Biochimica et Biophysica Acta (BBA)-Proteins and Proteomics, 1804(2), 245-262.

Phelps, E. A., Enemchukwu, N. O., Fiore, V. F., Sy, J. C., Murthy, N., Sulchek, T. A., . . .

García, A. J. (2012). Maleimide cross‐linked bioactive peg hydrogel exhibits improved reaction

kinetics and cross‐linking for cell encapsulation and in situ delivery. Advanced materials, 24(1),

64-70.

Pisal, D. S., Kosloski, M. P., & Balu-Iyer, S. V. J. J. o. p. s. (2010). Delivery of therapeutic

proteins. 99(6), 2557-2575.

140

Pratt, A. J., Shin, D. S., Merz, G. E., Rambo, R. P., Lancaster, W. A., Dyer, K. N., . . . Getzoff,

E. D. (2014). Aggregation propensities of superoxide dismutase G93 hotspot mutants mirror

ALS clinical phenotypes. Proc Natl Acad Sci U S A, 111(43), E4568-4576.

doi:10.1073/pnas.1308531111

Puls, I., Jonnakuty, C., LaMonte, B. H., Holzbaur, E. L., Tokito, M., Mann, E., . . . Oh, S. J.

(2003). Mutant dynactin in motor neuron disease. Nature genetics, 33(4), 455.

Raghavan, S. R., & Cipriano, B. H. (2006). Gel formation: phase diagrams using tabletop

rheology and calorimetry. In Molecular gels (pp. 241-252): Springer.

Rakhit, R., & Chakrabartty, A. (2006). Structure, folding, and misfolding of Cu, Zn superoxide dismutase in amyotrophic lateral sclerosis. Biochimica Et Biophysica Acta (BBA)-Molecular

Basis of Disease, 1762(11-12), 1025-1037.

Rakhit, R., Crow, J. P., Lepock, J. R., Kondejewski, L. H., Cashman, N. R., & Chakrabartty, A.

(2004). Monomeric Cu, Zn-superoxide dismutase is a common misfolding intermediate in the oxidation models of sporadic and familial amyotrophic lateral sclerosis. Journal of Biological

Chemistry, 279(15), 15499-15504.

Rakhit, R., Robertson, J., Velde, C. V., Horne, P., Ruth, D. M., Griffin, J., . . . Chakrabartty, A.

(2007). An immunological epitope selective for pathological monomer-misfolded SOD1 in ALS.

Nature medicine, 13(6), 754-759.

Ratovitski, T., Corson, L. B., Strain, J., Wong, P., Cleveland, D. W., Culotta, V. C., & Borchelt,

D. R. (1999). Variation in the biochemical/biophysical properties of mutant superoxide dismutase 1 enzymes and the rate of disease progression in familial amyotrophic lateral sclerosis kindreds. Hum Mol Genet, 8(8), 1451-1460.

141

Reaume, A. G., Elliott, J. L., Hoffman, E. K., Kowall, N. W., Ferrante, R. J., Siwek, D. R., . . .

Brown, R. H. (1996). Motor neurons in Cu/Zn superoxide dismutase-deficient mice develop

normally but exhibit enhanced cell death after axonal injury. Nature genetics, 13(1), 43-47.

Rehmann, M. S., Garibian, A. C., & Kloxin, A. M. (2013). Hydrolytically degradable thiol–ene

hydrogels for protein release. Paper presented at the Macromolecular symposia.

Rehmann, M. S., Skeens, K. M., Kharkar, P. M., Ford, E. M., Maverakis, E., Lee, K. H., &

Kloxin, A. M. (2017). Tuning and predicting mesh size and protein release from step growth

hydrogels. Biomacromolecules, 18(10), 3131-3142.

Restagno, G., Lombardo, F., Sbaiz, L., Mari, C., Gellera, C., Alimonti, D., . . . Chio, A. (2008).

The rare G93D mutation causes a slowly progressing lower motor neuron disease. Amyotroph

Lateral Scler, 9(1), 35-39. doi:10.1080/17482960701788198

788218754 [pii]

Rich, J. T., Neely, J. G., Paniello, R. C., Voelker, C. C., Nussenbaum, B., & Wang, E. W.

(2010). A practical guide to understanding Kaplan-Meier curves. Otolaryngology—Head and

Neck Surgery, 143(3), 331-336.

Ritger, P. L., & Peppas, N. A. (1987). A simple equation for description of solute release I.

Fickian and non-fickian release from non-swellable devices in the form of slabs, spheres, cylinders or discs. Journal of controlled release, 5(1), 23-36.

Robinson, R. (2017). FDA Approves Edaravone for ALS: Phase 3 Trial Finds It Slows

Progression. Neurology Today, 17(11), 1-30.

Rosen, D. R., Siddique, T., Patterson, D., Figlewicz, D. A., Sapp, P., Hentati, A., . . . Deng, H.-

X. (1993). Mutations in Cu/Zn superoxide dismutase gene are associated with familial amyotrophic lateral sclerosis. Nature, 362(6415), 59.

142

Rowland, L. P., & Shneider, N. A. (2001). Amyotrophic lateral sclerosis. New England Journal

of Medicine, 344(22), 1688-1700.

SAITO, I., & FUKUI, S. (1967). Studies on the oxidation products of lipoic acid. The Journal of

vitaminology, 13(2), 115-121.

Sanborn, T. J., Messersmith, P. B., & Barron, A. E. (2002). In situ crosslinking of a biomimetic

peptide-PEG hydrogel via thermally triggered activation of factor XIII. Biomaterials, 23(13),

2703-2710.

Sangwan, S., Zhao, A., Adams, K. L., Jayson, C. K., Sawaya, M. R., Guenther, E. L., . . .

Soriaga, A. B. (2017). Atomic structure of a toxic, oligomeric segment of SOD1 linked to

amyotrophic lateral sclerosis (ALS). Proceedings of the National Academy of Sciences, 114(33),

8770-8775.

Sawkar, A. R., Cheng, W.-C., Beutler, E., Wong, C.-H., Balch, W. E., & Kelly, J. W. (2002).

Chemical chaperones increase the cellular activity of N370S β-glucosidase: a therapeutic strategy for Gaucher disease. Proceedings of the National Academy of Sciences, 99(24), 15428-

15433.

Schmitt, N. D., & Agar, J. N. J. J. o. M. S. (2017). Parsing disease‐relevant protein modifications

from epiphenomena: perspective on the structural basis of SOD1‐mediated ALS. 52(7), 480-491.

Scott, L. J. (2014). Tafamidis: a review of its use in familial amyloid polyneuropathy. Drugs,

74(12), 1371-1378.

Seilhean, D., Cazeneuve, C., Thuries, V., Russaouen, O., Millecamps, S., Salachas, F., . . .

Duyckaerts, C. (2009). Accumulation of TDP-43 and alpha-actin in an amyotrophic lateral

sclerosis patient with the K17I ANG mutation. Acta Neuropathol, 118(4), 561-573.

doi:10.1007/s00401-009-0545-9

143

Shapiro, S. S., & Wilk, M. B. (1965). An analysis of variance test for normality (complete samples). Biometrika, 52(3/4), 591-611.

Sheng, J., & Xu, Z. (2016). Three decades of research on angiogenin: a review and perspective.

Acta Biochim Biophys Sin (Shanghai), 48(5), 399-410. doi:10.1093/abbs/gmv131

Shimoyama, S., Gansauge, F., Gansauge, S., Negri, G., Oohara, T., & Beger, H. G. (1996).

Increased angiogenin expression in pancreatic cancer is related to cancer aggressiveness. Cancer

Research, 56(12), 2703-2706.

Singh, R., & Whitesides, G. M. (1990). Degenerate intermolecular thiolate-disulfide interchange involving cyclic five-membered disulfides is faster by. apprx. 103 than that involving six-or seven-membered disulfides. Journal of the American Chemical Society, 112(17), 6304-6309.

Skorupa, A., King, M. A., Aparicio, I. M., Dussmann, H., Coughlan, K., Breen, B., . . . Prehn, J.

H. (2012). Motoneurons secrete angiogenin to induce RNA cleavage in astroglia. J Neurosci,

32(15), 5024-5038. doi:10.1523/JNEUROSCI.6366-11.2012

32/15/5024 [pii]

Skorupa, A., Urbach, S., Vigy, O., King, M. A., Chaumont-Dubel, S., Prehn, J. H., & Marin, P.

(2013). Angiogenin induces modifications in the astrocyte secretome: relevance to amyotrophic lateral sclerosis. J Proteomics, 91, 274-285. doi:10.1016/j.jprot.2013.07.028

S1874-3919(13)00433-8 [pii]

Smith, B. N., Ticozzi, N., Fallini, C., Gkazi, A. S., Topp, S., Kenna, K. P., . . . Miller, J. W.

(2014). Exome-wide rare variant analysis identifies TUBA4A mutations associated with familial

ALS. Neuron, 84(2), 324-331.

Spearman, C. (1904). The Proof and Measurement of Association between Two Things. The

American Journal of Psychology, 15(1), 72-101. doi:10.2307/1412159

144

Sreedharan, J., Blair, I. P., Tripathi, V. B., Hu, X., Vance, C., Rogelj, B., . . . Buratti, E. (2008).

TDP-43 mutations in familial and sporadic amyotrophic lateral sclerosis. Science, 319(5870),

1668-1672.

Stathopulos, P., Rumfeldt, J., Scholz, G., Irani, R., Frey, H., Hallewell, R., . . . Meiering, E.

(2003). Cu/Zn superoxide dismutase mutants associated with amyotrophic lateral sclerosis show enhanced formation of aggregates in vitro. Proceedings of the National Academy of Sciences,

100(12), 7021-7026.

Steele, A. D., Zhou, Z., Jackson, W. S., Zhu, C., Auluck, P., Moskowitz, M. A., . . . Lindquist, S.

(2009). Context dependent neuroprotective properties of prion protein (PrP). Prion, 3(4), 240-

249. doi:10135 [pii]

Su, J. (2018). Thiol-Mediated Chemoselective Strategies for In Situ Formation of Hydrogels.

Gels, 4(3), 72.

Su, X. W., Broach, J. R., Connor, J. R., Gerhard, G. S., & Simmons, Z. (2014). Genetic heterogeneity of amyotrophic lateral sclerosis: implications for clinical practice and research.

Muscle & nerve, 49(6), 786-803.

Subramanian, V., Crabtree, B., & Acharya, K. R. (2008). Human angiogenin is a neuroprotective factor and amyotrophic lateral sclerosis associated angiogenin variants affect neurite extension/pathfinding and survival of motor neurons. Hum Mol Genet, 17(1), 130-149. doi:ddm290 [pii]

10.1093/hmg/ddm290

Subramanian, V., & Feng, Y. (2007). A new role for angiogenin in neurite growth and pathfinding: implications for amyotrophic lateral sclerosis. Hum Mol Genet, 16(12), 1445-1453. doi:ddm095 [pii]

145

10.1093/hmg/ddm095

Szajewski, R. P., & Whitesides, G. M. (1980). Rate constants and equilibrium constants for thiol- disulfide interchange reactions involving oxidized glutathione. Journal of the American

Chemical Society, 102(6), 2011-2026.

Tanaka, M., Machida, Y., & Nukina, N. (2005). A novel therapeutic strategy for polyglutamine diseases by stabilizing aggregation-prone proteins with small molecules. Journal of molecular medicine, 83(5), 343-352.

Tarone, R. E., & Ware, J. (1977). On distribution-free tests for equality of survival distributions.

64(1), 156-160. doi:10.1093/biomet/64.1.156

Taylor, J. P., Brown Jr, R. H., & Cleveland, D. W. (2016). Decoding ALS: from genes to mechanism. Nature, 539(7628), 197.

Team, R. C. (2013). R: A language and environment for statistical computing.

Thiyagarajan, N., Ferguson, R., Subramanian, V., & Acharya, K. R. (2012). Structural and molecular insights into the mechanism of action of human angiogenin-ALS variants in neurons.

Nat Commun, 3, 1121. doi:10.1038/ncomms2126 ncomms2126 [pii]

Tiwari, A., Xu, Z., & Hayward, L. J. (2005). Aberrantly increased hydrophobicity shared by mutants of Cu, Zn-superoxide dismutase in familial amyotrophic lateral sclerosis. Journal of

Biological Chemistry, 280(33), 29771-29779.

Trouillon, R., Kang, D. K., Park, H., Chang, S. I., & O'Hare, D. (2010). Angiogenin induces nitric oxide synthesis in endothelial cells through PI-3 and Akt kinases. Biochemistry, 49(15),

3282-3288. doi:10.1021/bi902122w

146

Tsuji, T., Sun, Y., Kishimoto, K., Olson, K. A., Liu, S., Hirukawa, S., & Hu, G. F. (2005).

Angiogenin is translocated to the nucleus of HeLa cells and is involved in ribosomal RNA

transcription and cell proliferation. Cancer Res, 65(4), 1352-1360. doi:10.1158/0008-5472.CAN-

04-2058

Urushitani, M., & Shimohama, S. (2001). The role of nitric oxide in amyotrophic lateral

sclerosis. Amyotroph Lateral Scler Other Motor Neuron Disord, 2(2), 71-81.

van Blitterswijk, M., Vlam, L., van Es, M. A., van der Pol, W. L., Hennekam, E. A., Dooijes, D.,

. . . van den Berg, L. H. (2012). Genetic overlap between apparently sporadic motor neuron

diseases. PLoS One, 7(11), e48983. doi:10.1371/journal.pone.0048983

PONE-D-12-26807 [pii]

van Es, M. A., Diekstra, F. P., Veldink, J. H., Baas, F., Bourque, P. R., Schelhaas, H. J., . . . van

den Berg, L. H. (2009). A case of ALS-FTD in a large FALS pedigree with a K17I ANG

mutation. Neurology, 72(3), 287-288. doi:10.1212/01.wnl.0000339487.84908.00

72/3/287 [pii]

Van Es, M. A., Schelhaas, H. J., Van Vught, P. W., Ticozzi, N., Andersen, P. M., Groen, E. J., . .

. Diekstra, F. P. (2011). Angiogenin variants in Parkinson disease and amyotrophic lateral

sclerosis. Annals of neurology, 70(6), 964-973.

Vance, C., Rogelj, B., Hortobágyi, T., De Vos, K. J., Nishimura, A. L., Sreedharan, J., . . .

Wright, P. (2009). Mutations in FUS, an RNA processing protein, cause familial amyotrophic

lateral sclerosis type 6. Science, 323(5918), 1208-1211.

Vassall, K. A., Stathopulos, P. B., Rumfeldt, J. A., Lepock, J. R., & Meiering, E. M. (2006).

Equilibrium thermodynamic analysis of amyotrophic lateral sclerosis-associated mutant apo Cu,

Zn superoxide dismutases. Biochemistry, 45(23), 7366-7379.

147

Vermonden, T., Censi, R., & Hennink, W. E. (2012). Hydrogels for protein delivery. Chemical reviews, 112(5), 2853-2888.

Wang, J., Caruano-Yzermans, A., Rodriguez, A., Scheurmann, J. P., Slunt, H. H., Cao, X., . . .

Borchelt, D. R. (2007). Disease-associated mutations at copper ligand histidine residues of superoxide dismutase 1 diminish the binding of copper and compromise dimer stability. Journal of Biological Chemistry, 282(1), 345-352.

Wang, Q., Johnson, J. L., Agar, N. Y., & Agar, J. N. (2008). Protein aggregation and protein instability govern familial amyotrophic lateral sclerosis patient survival. PLoS biology, 6(7).

Wang, Q., Johnson, J. L., Agar, N. Y., & Agar, J. N. (2008). Protein aggregation and protein instability govern familial amyotrophic lateral sclerosis patient survival. PLoS Biol, 6(7), e170. doi:10.1371/journal.pbio.0060170

08-PLBI-RA-0487 [pii]

Wang, S., Taraballi, F., Tan, L. P., Ng, K. W. J. C., & research, t. (2012). Human keratin hydrogels support fibroblast attachment and proliferation in vitro. 347(3), 795-802.

Watanabe, S., Kaneko, K., & Yamanaka, K. (2013). Accelerated disease onset with stabilized familial amyotrophic lateral sclerosis (ALS)-linked mutant TDP-43 proteins. J Biol Chem,

288(5), 3641-3654. doi:10.1074/jbc.M112.433615

Watson, D., Castaño, E., Kokjohn, T. A., Kuo, Y.-M., Lyubchenko, Y., Pinsky, D., . . . Stine, W.

B. (2005). Physicochemical characteristics of soluble oligomeric A β and their pathologic role in

Alzheimer's disease. Neurological research, 27(8), 869-881.

Wichterle, O., & Lim, D. (1960). Hydrophilic gels for biological use. Nature, 185(4706), 117-

118.

148

Wijesekera, L. C., & Leigh, P. N. (2009). Amyotrophic lateral sclerosis. Orphanet J Rare Dis, 4,

3. doi:10.1186/1750-1172-4-3

1750-1172-4-3 [pii]

Wong, Y. C., & Holzbaur, E. L. (2014). Optineurin is an autophagy receptor for damaged

mitochondria in parkin-mediated mitophagy that is disrupted by an ALS-linked mutation.

Proceedings of the National Academy of Sciences, 111(42), E4439-E4448.

Wroe, R., Wai-Ling Butler, A., Andersen, P. M., Powell, J. F., & Al-Chalabi, A. (2008).

ALSOD: the Amyotrophic Lateral Sclerosis Online Database. Amyotroph Lateral Scler, 9(4),

249-250. doi:10.1080/17482960802146106

793701370 [pii]

Wu, C.-H., Fallini, C., Ticozzi, N., Keagle, P. J., Sapp, P. C., Piotrowska, K., . . . Baron, D. M.

(2012). Mutations in the profilin 1 gene cause familial amyotrophic lateral sclerosis. Nature,

488(7412), 499.

Wu, D., Yu, W., Kishikawa, H., Folkerth, R. D., Iafrate, A. J., Shen, Y., . . . Hu, G. F. (2007).

Angiogenin loss-of-function mutations in amyotrophic lateral sclerosis. Ann Neurol, 62(6), 609-

617. doi:10.1002/ana.21221

Xia, W., Fu, W., Cai, X., Wang, M., Chen, H., Xing, W., . . . Xu, D. (2015). Angiogenin promotes U87MG cell proliferation by activating NF-kappaB signaling pathway and

downregulating its binding partner FHL3. PLoS One, 10(2), e0116983.

doi:10.1371/journal.pone.0116983

PONE-D-14-34157 [pii]

149

Yang, Y., Hentati, A., Deng, H.-X., Dabbagh, O., Sasaki, T., Hirano, M., . . . Azim, A. C. (2001).

The gene encoding alsin, a protein with three guanine-nucleotide exchange factor domains, is mutated in a form of recessive amyotrophic lateral sclerosis. Nature genetics, 29(2), 160.

Yoon, J. M., Kim, S. H., Kwon, O. B., Han, S. H., & Kim, B. K. (1999). High level expression of soluble angiogenin in Escherichia coli. IUBMB Life, 47(2), 267-273.

Yoshida, M., Takahashi, Y., Koike, A., Fukuda, Y., Goto, J., & Tsuji, S. (2010). A mutation database for amyotrophic lateral sclerosis. Hum Mutat, 31(9), 1003-1010. doi:10.1002/humu.21306

Yoshioka, N., Wang, L., Kishimoto, K., Tsuji, T., & Hu, G.-f. (2006). A therapeutic target for prostate cancer based on angiogenin-stimulated angiogenesis and cancer cell proliferation.

Proceedings of the National Academy of Sciences, 103(39), 14519-14524.

Zoccolella, S., Beghi, E., Palagano, G., Fraddosio, A., Guerra, V., Samarelli, V., . . . Logroscino,

G. (2007). Riluzole and amyotrophic lateral sclerosis survival: a population-based study in southern Italy. Eur J Neurol, 14(3), 262-268. doi:ENE1575 [pii]

10.1111/j.1468-1331.2006.01575.x

Zong, L., Bartolami, E., Abegg, D., Adibekian, A., Sakai, N., & Matile, S. (2017).

Epidithiodiketopiperazines: Strain-promoted thiol-mediated cellular uptake at the highest

tension. ACS central science, 3(5), 449-453.

Zou, Z. Y., Wang, X. N., Liu, M. S., Sun, Q., Li, X. G., Cui, L. Y., & Kong, J. (2012).

Identification of a novel missense mutation in angiogenin in a Chinese amyotrophic lateral sclerosis cohort. Amyotroph Lateral Scler, 13(3), 270-275. doi:10.3109/17482968.2011.643900

Zuazo-Gaztelu, I., & Casanovas, O. (2018). Unraveling the Role of Angiogenesis in Cancer

Ecosystems. Front Oncol, 8, 248. doi:10.3389/fonc.2018.00248

150

Zustiak, S. P., & Leach, J. B. (2010). Hydrolytically degradable poly (ethylene glycol) hydrogel scaffolds with tunable degradation and mechanical properties. Biomacromolecules, 11(5), 1348-

1357.

Zwiener, I., Blettner, M., & Hommel, G. (2011). Survival analysis: part 15 of a series on evaluation of scientific publications. Deutsches Ärzteblatt International, 108(10), 163.

151