Volume 52 Number 6 December 2018 ISSN 0023-6772

Laboratory Animals THE INTERNATIONAL JOURNAL OF LABORATORY ANIMAL SCIENCE, MEDICINE, TECHNOLOGY AND WELFARE

Official Journal of AFSTAL, ECLAM, ESLAV, FELASA, GV-SOLAS, ILAF, LASA, NVP, SECAL, SGV, SPCAL

Published on behalf of Laboratory Animals Ltd. journals.sagepub.com/home/lan by SAGE Publications Ltd. We build the model, you do the research.

Ever considered the consequences of using a model verification. Erroneous modifications can lead to the carrying the wrong genes? The advent of recent generation of mutant and transgenic lines carrying genome editing technologies like CRISPR/Cas9 are non-functional alleles or unwanted mutations. These revolutionising the way new models are created, but faulty models can seriously impact your research their impact on the mouse genome requires careful programme, wasting resources on misleading results.

Learn how we can help at www.criver.com/modelcreation Volume 52 Number 6 December 2018 Contents

Review Article The African spiny mouse (Acomys spp.) as an emerging model for development and regeneration 565 G Pinheiro, DF Prata, IM Arau´jo and G Tiscornia

Original Articles Diurnally active rodents for laboratory research 577 R Refinetti and GJ Kenagy

Anaesthetic effects of alfaxalone administered intraperitoneally alone or combined with dexmedetomidine and fentanyl in the rat 588 M Arenillas and IAG de Segura

Refinement and reduction through application of a quantitative score system for estimation of TB-induced disease burden using computed tomography 599 SA Sharpe, D Smyth, A McIntyre, F Gleeson and MJ Dennis

Color matters: They would choose if they could (see)! 611 T Nitezki, N Schulz and S Kra¨mer

Four simple ways to increase power without increasing the sample size 621 SE Lazic

Pigs are useful for the molecular study of bone inflammation and regeneration in humans 630 FL Lu¨thje, K Skovgaard, HE Jensen and LK Jensen

News Registration is opened for the 14th FELASA Congress! 643

What does the GV-SOLAS actually do? 645 B Kra¨nzlin and N Linklater

Meeting report GV-SOLAS annual conference 2018 646 C Johner

En Espan˜a se presenta el primer informe anual del Acuerdo de Transparencia en Experimentacio´n Animal 649 S Vila

Newly created title of the Diplomate Specialized Veterinarian in Laboratory Animal Science (Dipl. SVLAS) awarded by the Swiss Association of Veterinarians in Industry and Research (SAVIR) 653 M Heimann

Calendar of events/Index to advertisers 656

*compared to previous generation Tunnel GP

ARCADIA_210X280_DEF-PROFILI.indd 1 13/10/2018 12:30:35 Catalogue 2019 Request your free copy NOW AVAILABLE! at: finescience.de

FINE SURGICAL INSTRUMENTS FOR RESEARCHTM ...... 553 Laboratory Animals Subscription information Annual subscription (2018) including postage: AALAS Member, (print and electronic) [US$120]. Combined Institutional Rate (print and electronic) [£304/US$562]. Electronic only and print Editorial Board only subscriptions are available for institutions at a discounted Editor-in-Chief B Riederer rate. Note VAT is applicable at the appropriate local rate. Deputy Editors G Jarvis, P Jirkof Visit sagepublishing.com for more subscription details. To activate your subscription (institutions only) visit Section Section Editors online.sagepub.com. Abstracts, tables of contents and contents alerts are available on this site free of charge Anaesthesia, Analgesia, P Flecknell, M Leach, P Foley, for all. Student discounts and single issue rates are available from Pain & Stress P Hedenqvist SAGE Publications Ltd, 1 Oliver’s Yard, 55 City Road, London EC1Y 1SP, UK, tel. þ44 (0)20 7324 8500, email subscriptions@sa- Anatomy and Neuroscience B Riederer, S Wells (neuro) gepub.co.uk and in North America, SAGE Publications Inc, Aquatic Organisms K Finger-Baier (fish), PO Box 5096, Thousand Oaks, CA 91320, USA. JP Mocho, M Crim Behaviour D Preissmann, M Gyger, Advertisement Managers L Lewejohann Biostatisics & Experimental R-D Gosselin, H Wu¨ rbel PRC Associates Ltd, 1st Floor Offices, 115 Roebuck Road, Design Chessington, Surrey KT9 1JZ, UK; Tel: +44 (0) 20 8337 3749; Fax: Education P Vergara, C Tho¨ne-Reineke +44 (0) 20 8337 7346; Email: [email protected] Imaging Techniques L van der Weerd, J Tremoleda Laboratory Animals Ltd Large Animal Models M Jensen-Waern, D Anderson, Laboratory Animals Ltd is a company limited by guarantee and has T Morris no share capital. The Memorandum of Association obliges the Management of Animal Facilities J-B Prins, M Dennis company to apply all its resources to the advancement of public Molecular & Genetic Engineering T Ruelicke, P Cinelli Pathology & Microbiology P Clements, D Salvatori, education in laboratory animal science, technology and welfare. It is a registered charity (Registered Charity Number 261047) and A Bleich none of its directors may receive any fee or remuneration. Physiology & Clinical Chemistry M Sommers, T Hough Primates G Rainer, P Honess Registered Office: 3Rs & Ethics G Griffin, A Olsson Laboratory Animals Ltd, 44 Springfield Road, Horsham, West Reproductive Biology H Hedrich, B Pintado, Sussex, RH12 2PD, UK C Gilbert Small Animal Models M Berard, J-B Prins Council of Management (temporary), S Wells Surgical Procedures D Bouard, R Tolba Chairman J-B Prins Systematic Review M Ritskes-Hoitinga, Secretary E Weir BS Kousholt Treasurer J Gregory Toxicology F Rutten L Antunes P Nowlan Veterinary Medicine E Rivera, J Sanchez-Morgado, K Applebee J Orellana L Whitfield, V Baumans B Riederer N Kostomitsopoulos M Berard A Ritchie Nutrition and Diets G Tobin N Ezov M Ritkes-Hoitinga Special Issue Microbiota guest editor: C Gilbert A Shortland Axel Kornerup-Hansen and J Guillen S Wells Craig Franklin J Helppi M Wilkinson C Johner Comment and correspondence relating to editorial matters may be sent to the Chairman of the Editorial Board by email: laeditorial@- ß 2018 Laboratory Animals Ltd. Apart from any fair dealing for the sagepub.co.uk; or post: LAL, PO Box 373, Eye, Suffolk, IP22 9BS, purposes of research or private study, or criticism or review, as permitted UK. under the UK Copyright, Designs and Patents Act, 1988, no part of this See also http://www.lal.org.uk publication may be reproduced, stored, or transmitted, in any form or by Laboratory Animals, (ISSN 0023-6772) is published and distributed any means, without the prior permission of the publishers, or in the case bimonthly (February, April, June, August, October, December) in of reprographic reproduction in accordance with the terms of licences both print and electronic form by SAGE Publications Ltd, issued by the Copyright Licensing Agency in the UK, or in accordance 1 Oliver’s Yard, 55 City Road, London EC1Y 1SP, UK. with the terms of licences issued by the appropriate Reproduction Rights Organization outside the UK. Enquiries concerning reproduction outside All manuscripts submitted for publication should the terms stated here should be sent to SAGE at the address. be prepared in accordance with the Guidelines Whilst every effort is made to ensure that no inaccurate or misleading for Authors which can be found online at data, opinion or statement appears in the journal, Laboratory Animals Ltd journals.sagepub.com/home/lan. Please submit your paper wish to make it clear that the data and opinions appearing in the articles online at mc.manuscriptcentral.com/la, and advertisements herein are the responsibility of the contributor or contributions for news items can also be made via advertiser concerned. Accordingly, Laboratory Animals Ltd and their offi- manuscript central. cers and agents accept no liability whatsoever for the consequences of any such inaccurate or misleading data, opinion or statement. Printed in Great Britain ...... 554 AFSTAL Secretary President Nicole Linklater Teresa Rodrigo Association Franc¸aise des Sciences et Faculty of Biology Techniques de I’Animal de Laboratoire Philipps University Vice President President Karl-von-Frisch Str. 8 Isabel Blanco Sebastian Paturance 35043 Marburg Vice President Germany Secretary Elodie Bouchoux (www.gv-solas.de) Clara Mun˜oz

Secretariat: 28, rue Saint Dominique, 75007, Treasurer Paris, France David Mun˜oz (www.afstal.com) ILAF Israeli Laboratory Animal Forum Secretariat: c/Maestro Ripoll, 8, President 28006 Madrid, Amir Rosner Spain (www.secal.es) ECLAM Secretary European College of Laboratory Animal David Castel Medicine Neufeld Cardiac Research President Institute SGV Rony Kalman Sheba Medical Center Schweizerische Gesellschaft fu¨r Tel Hashomer 52621 Secretariat: PO Box 9101NL-6500 Versuchstierkunde Israel HB Nijmegen, The Netherlands Socie´te´ Suisse pour la Science des Animaux de (www.ilaf.org.il) (http://eslav-eclam.org/eclam) Laboratoire (Swiss Laboratory Animal Science Association) President ESLAV LASA Dr. Birgit Ledermann European Society of Laboratory Animal Laboratory Animal Science Association Secretary Veterinarians President Dr. med. vet. Andrina Zbinden President David Anderson Faculty of Science and Medicine Jose´ Manuel Sa´nchez-Morgado University of Fribourg Secretary General Ch. du Muse´e8 Honorary Secretary Miles Maxwell CH-1700 Fribourg, Switzerland Massimiliano Bardotti (www.naturalsciences.ch/organisations/sgv) PO Box 524, Hull, Secretariat: Sanofi-Aventis R&D, HU9 9HE, UK 1, Avenue Pierre Brossolette (www.lasa.co.uk) 91385 Chilly-Mazarin Cedex France ([email protected]; www.eslav-eclam.org) SPCAL Sociedade Portuguesa de Cieˆ ncias NVP em Animais de Laborato´ rio (Portuguese Society for Laboratory Animal Nederlandse Vereniging voor Science) FELASA Proefdierkunde Federation of European (Dutch Association for Laboratory Animal President Laboratory Animal Science Associations Science) Isabel Vito´ria Figueiredo President Vice-President Hanna-Marja Voipio President Ricardo Afonso President-elect Martje Fentener van Vlissingen Ana Santos Secretary Secretary Catarina Pinto Reis Past President Jan Langemans Secretariat: Laborato´rio de Heinz Brandstetter BPRC Farmacologia Lange Kleiweg 139 Hon. Secretary Faculdade de Farma´cia 2288 GJ Rijswik Jean-Philippe Mocho Largo de D. Dinis The Netherlands 3000 Coimbra (www.proefdierkunde.nl) Secretariat: PO Box 372, Portugal Eye, IP22 9BR, UK (www.spcal.pt) (www.felasa.eu) SECAL Sociedad Espanˇ ola para las Ciencias del GV-SOLAS Animal de Laboratorio Gesellschaft fu¨r Versuchstierkunde (Spanish Society for Laboratory Animal (Society for Laboratory Animal Science) Science) President Bettina Kraenzlin

Innovative solutions for laboratory animal containment

BioFlex® Isolators > High Quality > Expert Knowledge > Experience > Innovation > Latest Technology > Reliability > Design Flexibility > Specialise in Bespoke Designs > Flexible Film, Semi-Rigid or Fully-Rigid Designs > Large Animal Isolator Designs

Bell Isolation Systems lead the way in the design and manufacture of high quality flexible film and rigid isolators using the latest materials and technology, coupled with over forty years of innovation developing standard, customised and bespoke products.

Contact for further information: Bell Isolation Systems Ltd Telephone: +44 (0) 1506 442916 Unit 12a, Oakbank Park Way Fax: +44 (0) 1506 440780 Livingston EH53 0TH Email: [email protected] Scotland, UK Website: www.bell-isolation-systems.com Helping to advance science and medicine Over half of the most commonly prescribed therapies for humans and pets are developed through research involving dogs. This is why our dedication to quality, consistency and the highest standards of animal welfare is so important. The Marshall Beagle® has contributed to medical advancements for over 55 years, let us help make a positive impact on your results.

www.marshallbio.com

Delivering higher standards Environment Enrichment

DesResTM A multitude of happy home owners now living in Desirable Residences from LBS Biotech... Available in a range of specifications to suit the first time buyer or those with a large family, these affordable cage additions offer an excellent way to improve living conditions in laboratory housing.

• Low maintenance • Upwardly mobile • Low cost • Stimulating living conditions • Builder’s certificate • Environmentally friendly materials

Contact the experts: Tel: +44 (0)1293 827940 Email: [email protected] www.lbs-biotech.com

Visit Prague, felasa Save the Date! clasa

IMPORTANT DATES

On-line registration open October 22, 2018 Accommodation booking open October 22, 2018 Abstract submission deadline January 31, 2019 Workshop registration open February 28, 2019 Early registration deadline March 31, 2019 Congress dates June 10 – 13, 2019

CONGRESS MOBILE APPLICATION Download our congress app and stay updated!

Congress Secretariat: GUARANT International Na Pankráci 17, 140 21 Prague 4, Czech Republic Phone: +420 284 001 444, fax: +420 284 001 448 E-mail: [email protected]

www.felasa2019.eu conventional systems • air fl ow systems • enrichment • enclosure systems • transport systems • cleaning systems • conventional systems

Combining high quality work with attention to detail – and a passion for design

ZOONLAB is for over 60 years one of the leading manufacturers and suppliers of complete solutions for scientifi c animal husbandries. Not only the vast product range but also decades of experience, fi rst-class quality and innovative and individual product solutions are reasons why customers from all over the world rely on ZOONLAB.

Learn more about us – we look forward to your enquiry: [email protected] or +49 / 23 05 / 97 30 40

ZOONLAB GmbH Hermannstraße 6 | 44579 Castrop-Rauxel | Germany www.zoonlab.de

Review Article Laboratory Animals 2018, Vol. 52(6) 565–576 ! The Author(s) 2018 The African spiny mouse (Acomys spp.) Article reuse guidelines: sagepub.com/journals- as an emerging model for development permissions DOI: 10.1177/0023677218769921 and regeneration journals.sagepub.com/home/lan

Gonc¸alo Pinheiro1,2, Diogo Filipe Prata1,2 , Ineˆs Maria Arau´jo1,2,3 and Gustavo Tiscornia1,3,4

Abstract The African spiny mouse (Acomys spp.) is an emerging animal model with remarkable biological character- istics that make it a subject of interest for a broad range of research fields. Typically a desert species adapted to a low-calorie diet, spiny mice develop diabetes-related symptoms when switched to high-energy diets. Spiny mice undergo relatively long gestation periods and have small litters of highly developed pups, making them an adequate model for late organogenesis and perinatal biology. Recently, they have been shown to have remarkable healing and regeneration capabilities, which make them unique among mammals. In this work, we describe our experience in housing a colony of African spiny mice and cover all basic aspects of feeding, maintenance and breeding for research purposes.

Keywords handling techniques, ethics and welfare, husbandry, nutrition, laboratory animal welfare, animal model, organisms and models

Date received: 12 July 2017; accepted: 27 February 2018

Introduction as a model of mammalian regeneration due to its remark- The genus Acomys, commonly known as the spiny mouse, able healing and regenerative properties.8–10 In this paper, is a small rodent distributed throughout Africa, the we present an overview of Acomys husbandry based on Middle East and south-west Asia.1 It was first described several years of experience in housing a colony. Animal in 1803 by E. Geoffrey2 and derives its common name husbandry and procedures were carried out as regulated from the presence of thick hair spines on its back. The by both Portuguese law (DL 113/2013) and European law genus presents a number of interesting biological charac- (directive 2010/63/EU). Projects at our institute involving teristics that have been the subject of research since the Acomys spp. are approved by the Animal Welfare Body 1950s, when the first laboratory colonies were established. of the University of Algarve and by the competent Early interest focused on Acomys as a model of type 2 authority (Direcc¸a˜ o Geral de Alimentac¸a˜ oe diabetes, based on the observation that animals fed high- fat diets developed obesity and hyperglycaemia.3 A 1 second area of interest is based on the fact that this Center for Biomedical Research (CBMR), University of Algarve, Faro, Portugal genus has a relatively long gestation period (approxi- 2Department of Biomedical Sciences and Medicine, University of mately 39 days) after which pups are born in an advanced Algarve, Faro, Portugal state of development. Therefore, the spiny mouse is used 3Algarve Biomedical Center, University of Algarve, Faro, Portugal as a model for organogenesis and late pregnancy biol- 4Clı´nica Eugin, Research and Innovation Department, Spain ogy.4–6 More recently, Acomys has been demonstrated Corresponding author: to undergo menstruation (a rare characteristic in mam- Gustavo Tiscornia, University of Algarve, Campus de Gambelas, mals) and may become an important rodent model for the Building 2, Lab 3.66, Faro, 8005-139, Portugal. 7 study of the menstrual cycle. The genus has also emerged Email: [email protected] 566 Laboratory Animals 52(6)

Veterina´ria). The Center for Biomedical Research propose classifying Acomys within the subfamily (CBMR-UALG) bioterium is a licensed animal facility Gerbilinae. However, more recent studies based on for small rodents. the sequences of multiple nuclear genes support the inclusion of Acomys in the subfamily Deomyinae.23,24 Nomenclature, taxonomy and The spiny mouse is found throughout Africa, the geographical distribution Middle East and south-west Asia. Within the genus, some species receive their common names after their Acomys derives its scientific name from the Greek main geographical location and are therefore referred ‘acme’ (pointy, referring to the shape of the snout) to as Egyptian spiny mouse (Acomys cahirinus), Cretan and ‘mus’ (mouse).11 Its common name, ‘spiny spiny mouse (A. minous), Cyprus spiny mouse mouse’, describes the thick spine-like hairs along its (A. nesiotes), Arabian spiny mouse (A. dimidiatus)or dorsum, which are presumably an anti-predator adap- Turkish Spiny Mouse (A. cilicicus). Other species tation. Currently, the International Union for the names commonly refer to coat colouring (golden spiny Conservation of Nature (IUCN) recognizes 18 distinct mouse (A. russatus) and grey spiny mouse species of Acomys. However, the taxonomy within the (A. cineraceus)), while several others go exclusively by genus Acomys remains somewhat controversial, with their scientific names, such as A. kempi, A. percivali, the number of species that compose it varying consid- A. wilsoni, etc. A thorough description of its habitats erably as the genus has been studied over the years.12 and geographical range can be found in The IUCN As its common name suggests, Acomys had traditionally Red List of Threatened Species 2017-3.25 Briefly, been classified as a murid based on various morpho- within Africa, the genus has a wide geographical range logical traits.13–16 However, analysis of biochemical (Figure 1), with various species being found in Maghreb and immunological data called this classification into and Egypt, Ethiopia, Sudan, Kenya, Uganda, Tanzania question,17–19 and further DNA-based analysis, includ- and as far south as South Africa. In the Middle East, a ing DNA–DNA hybridization analysis20 and direct number of species are found in Turkey, Jordan, Israel sequencing of the anti-chemotrypsin or cytochrome b and the Arabian Peninsula. The A. cahirinus/A. dimidia- genes12,21,22, found strong evidence of a close phylogen- tus group has branched out to the Mediterranean islands etic relationship with gerbils, leading these authors to of Cyprus (A. nesiotes) and Crete (A. minous), but the

A. minous A. nesiotes A. cilicicus A. cahirinus A. airensis A. dimidiatus A. johannis A. cahirinus A. seura A. spinossisimus A. subspinosus

A. russatus A. mullah A. cineraceus A. kempi A. percivali A. ignitus A. wilsoni

A. lousiae

Figure 1. Geographical distribution of Acomys species (this image was created using information available in the International Union for the Conservation of Nature Red List of Threatened Species, Version 2017-3;25 www.iucnredlist.org). Pinheiro et al. 567 genus has not been found in the wild in southern Animals have pointy snouts, relatively large protrud- Europe. In south-west Asia, Acomys has been reported ing black eyes, large ears and a scaly tail with sparse in Pakistan and Iran.1,26–29 It typically inhabits dry, short fine hairs. The dorsal coat colour is light golden semi-arid, rocky desert areas or short grasslands. In brown peppered with reddish, black and white hairs Kenya, it has been described as inhabiting rocky can- (A. cahirinus), with some species presenting colour- yons, cliff bases and rock formations known as kopjes. ations ranging from grey (A. percivali and A. cinera- The species can be commensal with man, inhabiting the ceus) to an almost rusty orange (A. russatus). The crevices of human dwellings11 or associated with fruit ventral coat is significantly lighter, usually a creamy orchards, palm groves or gardens. Acomys do not dig white. Spiny mice have a complex social organization burrows or build nests, tending to take refuge in what- including individual and kin recognition,30,31 commu- ever shelter the terrain offers. nal breeding32 and paternal care.33 Pups are born in a relatively advanced stage of devel- Basic biology opment. This precocity contrasts with the more common altricial strategies of other rodents (mice and Acomys is somewhat larger that the average laboratory rats) whose offspring are born at significantly earlier Mus musculus.1 In our colony, at 1 year of age, their stages of development. Acomys pups are born covered weight ranges between 39.05 and 63.41 g (N ¼ 52). with a soft, grey fur coat, unfurled ears, open eyes and Average weight is 49.3 6.1 g. Males weigh 49.9 are mobile within 1 or 2 days (Figure 2(a)); indeed, by 6.7 g (N ¼ 24) and females 48.8 5.4 g (N ¼ 28), with one week of age they can be completely active and no statistical differences between sexes (p ¼ 0.87). start feeding on solid food in addition to lactation Body length (snout to anus) ranges from 114 to (Figure 2b). At 4 weeks of age, the pups start becoming 96 mm. Average body length is 106.8 4mm(N ¼ 52). juveniles and start changing their coat commencing on Males measure 107 4.2 mm (N ¼ 24), while females their backs (Figure 2(c)), where dermal papillae already measure 106.6 3.7 mm (N ¼ 28), with no statistical formed during embryogenesis undergo a second wave differences between sexes (p ¼ 0.67). Tail length ranges of proliferation (the first occurs during embryogenesis) between 102 and 88 mm. Average tail length is to produce the thicker awl hairs that eventually 94.4 3.6 mm (N ¼ 46). Male tails measure develop into the genus’s characteristic spines;34 colour 93.8 3.3 mm (N ¼ 22), while female tails measure changes gradually to the adult hue (Figure 2(d)). Males 94.9 3.8 mm (N ¼ 24), with no statistical differences and females become sexually mature between 3 between the sexes (p ¼ 0.3). In general, no sexual and 4 months, but may need to reach between 4 and dimorphism is evident. 6 months to breed successfully, presumably depending

Figure 2. Postnatal development of A. Cahirinus. (a) 2-day-old pup, (b) 1-week-old pup, (c) 4-week-old juvenile and (d) 3-month-old adult. 568 Laboratory Animals 52(6) on social conditions. A recent report has demonstrated experience, spiny mice thrive well between 20 and that Acomys females have an 11-day oestral cycle35 26C, but may show discomfort (lethargy and stretch- after which, remarkably, they undergo menstruation.7 ing out) at temperatures >28C and increased huddling After mating, females do not form an evident external (presumably to retain heat) at temperatures <18C. vaginal plug, a feature which renders the identification Pups in particular can suffer hypothermia if exposed of exact mating time (E0) difficult. Spiny mice have a to low temperatures for more than 10–15 minutes. long gestation period compared to other rodents Humidity is kept within 30 and 60% and lighting con- (39 1 days), after which they give birth to small litters ditions on a 12-hr light/dark cycle. Light intensity is (one to four pups, usually two). Spiny mice are rela- kept below 250 lux. tively long-lived compared to other murids; our oldest Our animals are housed in transparent plastic animals are now over 4 years of age, with some publi- cages of different sizes. Cages of 50 30 22 cm cations reporting longevities of up to 6 years in labora- (or 60 45 28 cm) are fitted with a metallic grid tory colonies.36–38 (1 cm2) lid. The cage should not offer any non-metal Acomys have notoriously weak skin with a tensile surface that the animals can bite into, as they will read- strength significantly lower than that found in other ily chew holes large enough to exit the enclosure. The murids,10 presumably an anti-predatory adaptation. number of animals per cage varies between 2–3 or 3–5 When handled vigorously, their skin tears easily and adults in the smaller and larger cages, respectively. large patches of skin can come cleanly off the animal, Alternatively, they can be kept in groups of up to 20 exposing the underlying muscle fascia. Remarkably, animals in large wire cages (18 24 16 inches).11 bleeding is minimal, and animals walk off apparently Spiny mice are highly social and communal breeders, unconcerned and quickly engage in normal behaviors and seem to do well in what could be considered like feeding or grooming. They go on to close their slightly crowded conditions for other rodents. Given wounds through what looks like epimorphic regener- their highly social nature, we do not house them indi- ation, an exceptional trait in a mammal.8–10 Equally vidually except in case of disease, injury caused by interesting is that they do not seem to develop infec- attack by cage-mates or due to experimental reasons, tions in spite of what can be extensive wounds. Lizards and if so, strive to minimize the time spent by the ani- are known to show autotomy and can shed their tails, mals in isolation. A thin (2-cm deep) layer of substrate which later regenerate.39 Acomys show ‘false autot- is adequate, as spiny mice are notoriously devoid of omy’, a tendency to lose the tail sheath due to mechan- digging and burrowing behaviors. We use non-scented ical stress. A spiny mouse that loses his tail sheath will commercial cat litter, but other substrates like corn quickly proceed to eat his own tail and is left with a bedding and pine shavings can be used. The desert short stump that does not regenerate. Therefore, adaptations of spiny mice and their water homeostasis Acomys should be handled with care. Transfer of ani- result in relatively odour-free animal rooms (compared mals to a new cage is best achieved by coaxing them to Mus). As a rule, cages are cleaned out and animals gently into a plastic container tall enough to impede moved to fresh substrate every 2 weeks. While increas- escape; if handled by hand, the handler should wear ing the substrate layer to a depth of 3 cm allows sub- two pairs of gloves: soft leather gloves covered with strate changes every 3 weeks without any deleterious disposable latex gloves, as Acomys (particularly effect of any aspect of the animal’s health or repro- adults) can occasionally bite if alarmed. Younger indi- ductive behavior, we prefer a 2-week changing sched- viduals can be picked up by the tail for quick transfers ule. We provide one, or more, small wooden boxes or sexing. Adults should in general not be picked up by and/or polyvinyl chloride pipe sections (8 cm diameter) the tail, particularly if the animal has been in isolation, as hiding places or burrows. This minimizes conse- as this results in significantly more agitated animals. quences of occasional aggressive behavior that can The animal should be gently grasped with the animal’s sometimes be observed in cages and seems to encour- back against the handler’s open palm with the fingers age breeding, as females usually have their litters gently closing in on the animal’s whole body. Holding inside the boxes. Dry wooden branches for gnawing their head and front paws immobilized also seems to and climbing, as well as a small running wheel, are calm the animals while being handled. provided for environmental enrichment and seem par- ticularly appreciated by the younger individuals. As animals are usually housed in groups, an individual Husbandry and colony maintenance identification system is required. Ear punches cannot Housing be used due to the fact that Acomys can regenerate extensive full-thickness wounds in the ear pinna (see Our animals are kept in a standard animal facility. below). Ear clips are equally unsuitable, as spiny Temperature is maintained at 24–26C. In our mice ears are relatively delicate and can tear easily. Pinheiro et al. 569 Breeding Given their regenerative potential, we have not tested toe clipping as an identification strategy. We currently For reproduction, we have used breeding groups con- use microchips (IDVet, Portugal) that can be inserted sisting of one male with one to three females without subcutaneously between the shoulder blades, allowing observing major differences in reproductive output; unambiguous identification of each particular however, published reports suggest that maximum individual. productivity is obtained with groups consisting of sev- eral related females paired to an unrelated male and Diet and nutrition their offspring, with litter size correlating to female weight and number of immature females in the Spiny mice are highly omnivorous. In the wild they feed group.33 Litters are small, ranging from one to four on green vegetation, assorted seeds, dry or fleshy fruits, pups, but most often two pups. Overall sex ratio has small arthropods, mushrooms and snails.40 They do not been reported as relatively stable at 1:1.47 If needed, hoard food. It is not known if Acomys consumes meat pups can be weaned by 3 weeks of age; alternatively, in their natural habitat,41,42 but cannibalism of pups we occasionally leave offspring in the parental cage for and dead cage-mates is a relatively common occurrence up to 2 months without having observed effects on in our colony. Pups can disappear without a trace, and productivity of the breeding group; indeed, mature a cannibalized adult can be reduced to a pile of bones females can become receptive shortly after parturition and skin relatively quickly. Acomys is adapted to dry and a new litter will be produced between 45 and 60 environments43 and derives a significant part of its days after the previous litter. However, this can be quite water intake from food.44 This makes Acomys more variable. Acomys breed year-round, but we have sensitive to salt and fat-rich diets, which results in a observed variable productivity decreases during reduction of fertility and an increase in the probability winter, despite the fact that bioterium environmental of appearance of type 2 diabetes.3,45,46 Food can be conditions remain unchanged. dispensed into a flat ceramic bowl or directly on to Sex can be readily determined at 3–4 weeks of age by the substrate. While they will eat almost everything inspecting the genitalia, with the distance between the that is put in front of them, they are not without anus and the sexual organs distinctly larger in males their preferences. After trying different feeding regimes, than in females (Figure 3). Sexing younger animals is we finally settled for a mixture of seeds (commercial more difficult. Establishing new breeding groups is best poultry seed mixes, rich in wheat, barley, corn, sun- done when the animals are still relatively young and flower seeds, etc). Animals are fed approximately sociable (<4 months old), regardless of whether they 15% of their weight two times a week. As a rule of have started to mature sexually (>1 month old). In thumb, no more than 10% of the previous meal this case, animals can be put together in a new cage should remain uneaten at the next feeding. Once a without any special precautions. Older individuals week, the animals diet is supplemented with a mixture become more territorial and may not accept new of vegetables and fruit. A number of breeding groups cage-mates readily. Females can be particularly aggres- from our colony were fed normal laboratory mouse sive to new males introduced into their territory, par- pellets ad libitum: after three months, breeding ticularly if there is more than one female;48 new males dropped off precipitously but recommenced quickly can be chased, attacked, severely injured or even killed. when animals were switched back to their normal We have not tested whether this behavior is dependent seed-based, vegetable/fruit-supplemented diet. Of on the time of year/season of male introduction. When note, the mouse chow used was standard maintenance attempting to establish a breeding group with older type (Ultragene Maintenance Diet, 12% protein con- animals, we put the male in the cage first and allow tent); it is possible that use of a high-protein level him 2 or 3 days to establish his territory. We then put mouse chow would allow reproduction to continue, a cardboard divider with a window covered with metal- although we have not yet tested this. Protein content lic mesh, dividing the cage into two equal areas. can be reinforced with live mealworms, crickets or A female is introduced into the second compartment, earthworms once a week. Pups less than 1 month giving animals the opportunity to become habituated to old or animals recovering from sickness or injury each other’s presence and scent. After 2 days (if the benefit from softer food. Animals receive a vitamin animals have not already chewed through it) the divider supplement (Wevit, Portugal) once a month and min- can be taken out. Usually some chasing and scuffling eral water is provided ad libitum. Unrestricted feeding occurs (particularly of females chasing males) and can can lead to overweight animals and possibly the devel- result in scratches, bitten tails and ears, or light skin opment of diabetes. In particular, carbohydrate-rich injuries. After a few hours, the animals calm down. The and especially fat-rich diets should be avoided animals should be monitored to ensure mutual accept- (see below). ance and separated if aggression results in severe injury. 570 Laboratory Animals 52(6)

Figure 3. Sexing of A. cahirinus.

If joining is successful, the divider may be reinstalled not form a visible vaginal plug upon pregnancy and a second female introduced. making determination of E0 problematic. Females become conspicuously heavier around midpregnancy, General and reproductive behavior which lasts approximately 39 days.

Once breeding groups are established, the group tends Health to live harmoniously. Acomys are generally nocturnal; one notable exception is A. russatus, which is nocturnal Acomys are generally robust and relatively easy to but adopts a diurnal lifestyle when it shares its habitat maintain in good health when given good living condi- with A. Cahirinus.49 During the day, animals sleep in tions and a healthy diet. Any new animals arriving at huddled groups, becoming active for short periods, par- the colony remain quarantined for 2 weeks and are ticularly if they are stimulated by human presence or monitored for signs of disease, external parasites, and fed. They are naturally curious and become readily nasal or ocular secretions. Healthy spiny mice are clear accustomed to their handlers. Of note, cage lids eyed, socially active and inquisitive. Discomfort or pain should be well secured, as Acomys are inquisitive, usually causes the animals to become lethargic and lively, and something of ‘escape artists’. If they do assume a hunched position. Other signs of sickness escape, recapturing them can be a challenge, as they are semi-closed eyes and puffy eyebags. After the quar- are quick runners capable of sizable jumps; recapture antine period is over, animals are integrated into the is best achieved by using a box-type mouse trap that colony. When cases of sickness do occur, we typically allows access to food and snaps shut when the animal isolate the animal in a separate room and monitor its nibbles at the morsel. Acomys do not burrow, dig or development. Occasionally, animals may fight and show foot-hopping behavior. Hierarchical relationships wounds may be found; these are usually small, of are established by chasing and tail nipping, and it is not little concern and heal quickly without requiring treat- uncommon for tails to be scratched, bitten and injured. ment. It will take a very severe skin wound to threaten Lower-status animals can be recognized by a greater the life of a spiny mouse; full-thickness wounds of up to number of scabs and bite-marks on their tails, and 30% of the surface of the back heal without any par- tend to occur in breeding groups under situations of ticular intervention; indeed, after 60 days, little external social stress. If a tail becomes severely injured, or if sign of the wound remains (Figure 4). Histological ana- the animal has lost its tail sheath due to poor handling, lysis has shown substantial regeneration of such the usual outcome is that the animal chews off the wounds, including the formation of new hair follicles entire tail, leaving a short stump. Females tend to be in the bed wound. Information about disease and para- dominant over males. A superovulation protocol for sites of Acomys, either in the wild or in captivity, is spiny mice has recently been reported and could con- limited. Wild-captured Acomys have been reported to stitute the first step towards the development of trans- carry ticks and fleas,51 cestodes, nematodes52 and coc- genic technologies.50 When the female becomes cidian protozoa.53 In addition, a number of blood- receptive, the male will chase the female until she is borne parasites have been detected, including Babesia, successfully mounted. The coitus itself involves rapid Bartonella, Haemobartonella, Hepatozoon and thrusting and lasts only for 1–2 seconds, after which Trypanosoma.54 Diagnostic tests specific for Acomys both animals lick their genitalia clean. Somewhat are not commercially available but standard tests avail- inconveniently for researchers, pregnant females do able for rodents (PCR assays, foecal floats, pelage and Pinheiro et al. 571

Figure 4. Remarkable healing and regeneration of skin wounds in adult A. cahirinus. (a) An adult animal caught in a dysfunctional running wheel tore himself free; the wound was trimmed clean but was otherwise not treated. (b) 2 days post-injury, (c) 12 days post-injury, (c) 21 days post-injury and (e) 50 days post-injury. tape tests) can be used to determine the pathogen status 1970s, the diabetic phenotype of Acomys under fat- or of the colony.11 In addition, we expose sentinels to sucrose-enriched diets was characterized. While a fat- Acomys soiled bedding and test them twice a year. rich diet induced obesity and diabetes, a sucrose-rich While we have never detected parasites or infectious diet induced hepatomegaly, hyperactive lipogenesis agents, we have relatively rarely had animals become and elevated cholesterol blood levels. Pancreatic hyper- sick and quickly die; on one occasion 11 animals died plasia and insulinaemia also occurred, but ketosis and within a 10-day period, suggesting an infectious out- pancreatic atrophy were delayed in comparison to a fat- break from an agent that we were unable to identify. rich diet. Overall, mortality was significantly higher for The ‘outbreak’ resolved spontaneously, did not the fat-rich diet.56,58 From the 1980s onward, diabetes propagate throughout the colony and has not reoc- research in Acomys decreased in favour of other more cured since. genetically tractable mouse models. From a practical point of view, it is not clear that Acomys individuals Uses as an experimental model within outbred colonies would all become diabetic on high-energy diets, but we proceed under the assumption The first spiny mouse laboratory colonies were estab- that they do have a potentially diabetic-prone lished in the 1950s. While Acomys has never been a metabolism. mainstream model, it has been used to study a range of subjects, including physiology (in particular, adapta- Pre-natal and perinatal development tions to desert habitats), ecology, evolution, behavior research and metabolism. Three major themes stand out: dia- betes, pre-natal development and regeneration. Spiny mice are an interesting model for a range of developmental studies focusing on organogenesis, late Diabetes research pregnancy and perinatal biology. Pups are born in an advanced state of development with open eyes, Obesity and diabetes were reported in an A. cahirinus unfurled ears and completely covered in fur. This pre- colony kept at the University of Geneva in the cocity results in most of their organ systems being rela- 1960s.3,55 Comparative dietary studies of spiny mice tively well developed at birth and contrasts markedly in two different colonies led to the conclusion that the with the state of development found in altricial rodent diabetic phenotype was a metabolic response of a desert pups. Spiny mouse pups are born capable of basic loco- species to nutritional plenty rather than selection of a motion and become completely mobile within 1 week. genetic variant.45,46 Up to 15% of 1-year-old animals They are also capable of a degree of thermoregulation, developed nutritionally induced diabetes, with beta-cell although pups under 3 weeks are sensitive to hypother- hyperplasia,3 increased pancreatic insulin, low insulin mia if isolated from contact with adults. Liver, lungs, responses, hyperinsulinaemia, hyperglycaemia and glu- certain brain regions and the kidney all show high levels cosuria.56 After some time, islets of Langerhans broke of maturity at birth.59–64 The spiny mouse model has down, the pancreas atrophied and ketosis occurred, been used to examine the effect of maternal exposure to eventually leading to death.57 During the 1960s and glucocorticoids. Exposure to elevated concentrations of 572 Laboratory Animals 52(6) dexamethasone levels during midgestation has deleteri- regeneration field, two important themes are the role ous effects on placental function and is affected by the of the extracellular matrix (ECM) and the immune sex of the foetus and time elapsed since exposure,65,66 system in the regulation of the fibrotic vs. regenerative an observation that parallels what is observed in human response. During Acomys regeneration, the ECM has pregnancies. Another aspect of human pregnancy that significantly lower levels of collagen type 1 deposition has been modeled in Acomys is hypoxia during partur- than in Mus,77 with relatively higher levels of collagen ition, which can cause cerebral palsy, cognitive disabil- III and V.8 Transcriptional profiling has shown a rela- ities, impaired memory and retarded development.67 tively higher ratio of metalloproteinases to metallopro- A study on pups subjected to hypoxia at birth revealed teinase inhibitors in Acomys as compared to Mus, central nervous system (CNS) inflammation and a suggesting higher levels of ECM turnover in Acomys, range of effects in behavioral testing.67 Follow-up stu- which could facilitate cell migration.8 Initial analysis of dies suggested that treating pregnant females with the role of inflammatory cells in Acomys regeneration melatonin before asphyxia, or supplementing the suggested that, compared to Mus, Acomys skin wounds maternal diet with creatine during pregnancy, improves have greater numbers of mast cells but lower numbers survival and ameliorates effects on both the structure of mature macrophages. Interestingly, proinflamatory and function of the CNS, kidneys and dia- cytokines levels were significantly lower in Acomys, sug- phragm.4,5,68–72 The model has also been used to gesting a blunted inflammatory response in spiny study in utero CNS development.59,73,74 mice.76 However, a more recent study contradicted this view. While neutrophils and macrophages infil- Regeneration research trated both Mus and Acomys wounds, the relative amounts of cells infiltrating the region had distinct tem- A recent interesting development in spiny mouse poral profiles in both systems, with CD11bþ cells being research has been the discovery of the ability of this significantly more abundant in Mus at day 3 post- species to regenerate extensively after wounding. injury.78 Furthermore, cells showed different distribu- Regeneration occurs broadly but unevenly throughout tions, with pro-inflammatory macrophages (CD86þ) the animal kingdom; however, it is relatively rare in being excluded from the blastema in Acomys. mammals, which tend to respond to injury with fibrotic Importantly, transient local depletion of macrophages scarring rather than epimorphic regeneration.75 In by clodronate liposome injection into the ear abrogated 2012, Seifert and colleagues reported that two species blastema formation, which resumed once the depletion of spiny mice (A. kempi and A. percivali) have weak was resolved. In addition, the scarring response was skin that tears easily in response to mechanical characterized by higher myeloperoxidase activity stress.10 It is not uncommon for animals to suffer exten- while regeneration showed higher reactive oxygen spe- sive, full-thickness skin wounds. Remarkably, the cies activity.78 The exact role of the different subsets of wounds close quickly; histological analysis has shown inflammatory cells in the fibrotic vs. regenerative that bona fide epimorphic regeneration occurs, with response remains to be elucidated. An important ques- animals reconstituting the original tissue architecture, tion is whether the remarkable regenerative capabilities including adipose tissue, dermis, epidermis, sebaceous of Acomys are limited to skin and ears or extend to glands and hair. Moreover, 4-mm circular full-thick- other organs and systems. ness wounds in the ear pinna close fully within 2 months, regenerating hyaline cartilage, adipocytes, Discussion dermis, epidermis and hair follicles.10 This observation has been confirmed in a third member of the species (A. Spiny mice show a range of biological characteristics cahirinus) by our own work.9 In addition, we have worthy of study. While their interest as a model for observed extensive angiogenesis, nerve fibers and the diabetes type 2 has waned, a number of groups con- presence of muscle fibers in the regenerated region.9 tinue to study diverse aspects of their physiology, ecol- The cellular and molecular mechanisms underlying ogy and behavior. In addition to offering a system in this regenerative capacity are still poorly understood. for the study of late organogenesis and perinatal events, The nature of the cellular and molecular mechanisms they have recently been shown to possess remarkable underlying the regenerative response in Acomys are cur- regenerative abilities, the full extent of which remains to rently under investigation. Interestingly, in the ear- be determined. They are a robust genus that is relatively punch model, both Mus and Acomys are capable of easy to maintain and breed. Their main limitation is the mounting a proliferative response to wounding; how- small litter size and the longer time to sexual maturity ever, this response is abrogated in Mus and only (compared to Mus). While a long gestation period Acomys goes on to establish a proliferating blastema, offers the advantage of being able to study advanced which eventually leads to regeneration.76 In the developmental processes, it also limits their Pinheiro et al. 573 reproductive potential and building up sufficient num- 7. Bellofiore N, Ellery SJ, Mamrot J, et al. First evidence of bers of animals for research can be challenging. They a menstruating rodent: The spiny mouse (Acomys cahir- are also rather uncharacterized in terms of reproductive inus). Am J Obstet Gynecol 2017; 216: 40.e1–40.e11. biology, genetics and genomics; some initial transcrip- 8. Brant JO, Lopez M-C, Baker HV, et al. A comparative tomic data is published8,77,79 and genome sequencing is analysis of gene expression profiles during skin regener- ation in Mus and Acomys. PLoS One 2015; 10: e0142931. underway, but other techniques and reagents, particu- 9. Matias Santos D, Rita AM, Casanellas I, et al. Ear larly antibodies, remain to be developed. We hope this wound regeneration in the African spiny mouse work can ease the entry of new researchers into the field Acomys cahirinus. Regeneration (Oxf) 2016; 3: 52–61. of Acomys biology. 10. Seifert AW, Kiama SG, Seifert MG, et al. Skin shedding and tissue regeneration in African spiny mice (Acomys). Nature 2012; 489: 561–565. Acknowledgement 11. Haughton CL, Gawriluk TR and Seifert AW. The biol- The authors acknowledge funding from the Fundac¸a˜ o para a ogy and husbandry of the African spiny mouse (Acomys Cieˆncia e a Tecnologia of Portugal and from the Center for cahirinus) and the research uses of a laboratory colony. Biomedical Research, University of Algarve. J Am Assoc Lab Anim Sci 2016; 55: 9–17. 12. Barome PO, Lymberakis P, Monnerot M, et al. Declaration of Conflicting Interests Cytochrome b sequences reveal Acomys minous (Rodentia, Muridae) paraphyly and answer the question The author(s) declared no potential conflicts of interest with about the ancestral karyotype of Acomys dimidiatus. Mol respect to the research, authorship and/or publication of this Phylogenet Evol 2001; 18: 37–46. article. 13. Bonhomme F, Iskandar D, Thaler L, et al. Electromorphs and phylogeny in muroid rodents. Funding In: Luckett WP and Hartenberger JL (eds) Evolutionary relationships among rodents. Boston: Springer, 1985, The author(s) disclosed receipt of the following financial sup- pp.671–683. port for the research, authorship and/or publication of this 14. Denys C, Michaux J, Petter F, et al. Molar morphology article: the Fundac¸a˜ o para a Cieˆncia e a Tecnologia (grant as a clue to the phylogenetic relationship of Acomys to PTDC/BIA-ANM/0697/2014 and UID/BIM/04773/2013) the Murinae. Isr J Zool 1992; 38: 253–262. and by structural funds from the Center for Biomedical 15. Hutterer R, Lopez-Martinez N and Michaux J. A new Research, University of Algarve, (UID/BIM/04773/2013 rodent from quaternary deposits of the Canary Islands CBMR). and its relationships with neogene and recent murids of Europe and Africa. Palaeovertebrata 1988; 18: 241–262. 16. Jacobs LL. Fossil rodents (Rhizomyidae and Muridae) ORCID iD from Neogene Siwalik deposits, Pakistan. Flagstaff: DF Prata http://orcid.org/0000-0001-7118-559X Museum of Northern Arizona, 1978. 17. Chevret P, Denys C, Jaeger J-J, et al. Molecular evidence that the spiny mouse (Acomys) is more closely related to References gerbils (Gerbillinae) than to true mice (Murinae). Proc 1. Nowak RM and Walker EP. Walker’s mammals of the Natl Acad Sci USA 1993; 90: 3433–3436. world. Baltimore: Johns Hopkins University Press, 1999. 18. Luckett W and Hartenberger J. Evolutionary relationships 2. Wilson DE and Reeder DM. Mammal species of the world: among rodents: A multidisciplinary approach. New York: A taxonomic and geographic reference. Baltimore: Johns Plenum Press, 1985. Hopkins University Press, 2005. 19. Wilson AC, Ochman H and Prager EM. Molecular time 3. Gonet AE, Stauffacher W, Pictet R, et al. Obesity and scale for evolution. Trends Genet 1987; 3: 241–247. diabetes mellitus with striking congenital hyperplasia of 20. Agulnik SI and Silver LM. The Cairo spiny mouse the islets of langerhans in spiny mice (Acomys cahirinus). Acomys cahirinus shows a strong affinity to the Diabetologia 1966; 1: 162–171. Mongolian gerbil Meriones unguiculatus. Mol Biol Evol 4. Dickinson H, Ireland ZJ, LaRosa, et al. Maternal dietary 1996; 13: 3–6. creatine supplementation does not alter the capacity for 21. Barome P-O, Monnerot M and Gautun J-C. Intrageneric creatine synthesis in the newborn spiny mouse. Reprod phylogeny of Acomys (Rodentia, Muridae) using mito- Sci 2013; 20: 1096–1102. chondrial gene cytochrome b. Mol Phylogenet Evol 1998; 5. Ellery SJ, Ireland Z, Kett MM, et al. Creatine pretreat- 9: 560–566. ment prevents birth asphyxia-induced injury of the new- 22. Martin Y, Gerlach G, Schlo¨tterer C, et al. Molecular born spiny mouse kidney. Pediatr Res 2013; 73: 201–208. phylogeny of European muroid rodents based on com- 6. Ellery SJ, LaRosa DA, Kett MM, et al. Dietary creatine plete cytochrome b sequences. Mol Phylogenet Evol 2000; supplementation during pregnancy: A study on the effects 16: 37–47. of creatine supplementation on creatine homeostasis and 23. Fabre PH, Hautier L, Dimitrov D, et al. A glimpse on the renal excretory function in spiny mice. Amino Acids 2016; pattern of rodent diversification: A phylogenetic 48: 1819–1830. approach. BMC Evol Biol 2012; 12: 88. 574 Laboratory Animals 52(6)

24. Steppan SJ, Adkins RM, Spinks PQ, et al. Multigene 43. Levy O, Dayan T and Kronfeld-Schor N. Adaptive phylogeny of the Old World mice, Murinae, reveals dis- thermoregulation in golden spiny mice: the influence of tinct geographic lineages and the declining utility of mito- season and food availability on body temperature. chondrial genes compared to nuclear genes. Mol Physiol Biochem Zool 2011; 84: 175–184. Phylogenet Evol 2005; 37: 370–388. 44. Shkolnik A and Borut A. Temperature and water rela- 25. The IUCN Red List of Threatened Species 2017-3. http:// tions in two species of spiny mice (Acomys). J Mammal www.iuncnredlist.org (accessed March 2017). 1969; 50: 245–255. 26. Ellerman J. The families and genera of living rodents: With 45. Gutzeit A, Renold AE, Cerasi E, et al. Effect of diet- a list of named forms 1758 - 1936. Volume I Rodents other induced obesity on glucose and insulin tolerance of a than Muridae. London: British Museum of Natural rodent with a low insulin response (Acomys cahirinus). History, 1940. Diabetes 1979; 28: 777–784. 27. Harrison DL and Bates PJJ. The mammals of Arabia. 46. Shafrir E. Albert Renold memorial lecture: Molecular Sevenoaks: Harrison Zoological Museum, 1991. background of nutritionally induced insulin resistance 28. Jeremy P and Bates J. The distribution of Acomys leading to type 2 diabetes–from animal models to (Rodentia: Muridae) in Africa and Asia. Isr J Zool humans. Int J Exp Diabetes Res 2001; 2: 299–319. 1994; 40: 199–214. 47. Nova´kova´M, Vasˇ a´kova´B, Kutalova´H, et al. Secondary 29. Mendelssohn H and Yom-Tov Y. Fauna Palaestina: sex ratios do not support maternal manipulation: Mammalia of Israel. Jerusalem: Israel Academy of Extensive data from laboratory colonies of spiny mice Sciences and Humanities, 1999. (Muridae: Acomys). Behav Ecol Sociobiol 2010; 64: 30. Porter RH, Matochik JA and Makin JW. Evidence for 371–379. phenotype matching in spiny mice (Acomys cahirinus). 48. Porter RH. Sex-differences in the agonistic behavior of Animal Behav 1983; 31: 978–984. spiny-mice (Acomys cahirinus). Ethology 1976; 40: 31. Porter RH, Matochik JA and Makin JW. Discrimination 100–108. between full-sibling spiny mice (Acomys cahirinus) by 49. Gutman R and Dayan T. Temporal partitioning: An olfactory signatures. Animal Behav 1986; 34: 1182–1188. experiment with two species of spiny mice. Ecology 32. Makin JW and Porter RH. Paternal behavior in the spiny 2005; 86: 164–173. 50. Pasco R, Gardner DK, Walker DW, et al. A superovula- mouse (Acomys cahirinus). Behav Neural Biol 1984; 41: tion protocol for the spiny mouse (Acomys cahirinus). 135–151. Reproduction Fertil Dev 2012; 24: 1117–1122. 33. Frynta D, Franˇ kova´M, Cˇ ı´zˇkova´B, et al. Social and life 51. Krasnov B, Sarfati M, Arakelyan M, et al. Host specifi- history correlates of litter size in captive colonies of pre- city and foraging efficiency in blood-sucking parasite: cocial spiny mice (Acomys). Acta Theriol 2011; 56: Feeding patterns of the flea Parapulex chephrenis on 289–295. two species of desert rodents. Parasitol Res 2003; 90: 34. Montandon SA, Tzika AC, Martins AF, et al. Two waves 393–399. of anisotropic growth generate enlarged follicles in the 52. Behnke JM, Barnard CJ, Mason N, et al. Intestinal hel- spiny mouse. Evodevo 2014; 5: 33. minths of spiny mice (Acomys cahirinus dimidiatus) from 35. Peitz B. The oestrous cycle of the spiny mouse (Acomys St Katherine’s Protectorate in the Sinai, Egypt. cahirinus). J Reprod Fertil 1981; 61: 453–459. J Helminthol 2000; 74: 31–43. 36. Bodenheimer F. Ecological and physiological studies on 53. Kvicˇ erova´J, Pta´cˇ kova´P and Modry´ D. Endogenous devel- Physiol Comp Ocol Int J Comp Physiol some rodents. opment, pathogenicity and host specificity of Eimeria cahir- Ecol 1949; 1: 376–389. inensis Couch, Blaustein, Duszynski, Shenbrot and Nevo, 37. Morrison P, Dieterich R and Preston D. Longevity and 1997 (Apicomplexa: Eimeriidae) from Acomys dimidiatus mortality in 15 rodent species and subspecies maintained (Cretzschmar 1826) (Rodentia: Muridae) from the Near in laboratory colonies. Acta Theriol 1977; 22: 317–335. East. Parasitol Res 2007; 100: 219–226. 38. Weigl R. Longevity of mammals in captivity; From the 54. Alsarraf M, Bednarska M, Mohallal EM, et al. Long- living collections of the world. Frankfurt am Main: term spatiotemporal stability and dynamic changes in Kleine Senckenberg-Reihe, 2005. the haemoparasite community of spiny mice (Acomys 39. Clause AR and Capaldi EA. Caudal autotomy and regen- dimidiatus) in four montane wadis in the St. Katherine eration in lizards. J Exp Zool A Comp Exp Biol 2006; 305: Protectorate, Sinai, Egypt. Parasit Vectors 2016; 9: 195. 965–973. 55. Pictet R, Orci L, Gonet AE, et al. Ultrastructural studies 40. Gliwicz J. Niche segregation in a rodent community of of the hyperplastic islets of Langerhans of spiny mice African dry savanna. J Mammal 1987; 68: 169–172. (Acomys cahirinus) before and during the development 41. Kronfeld-Schor N and Dayan T. The dietary basis for of hyperglycemia. Diabetologia 1967; 3: 188–211. temporal partitioning: Food habits of coexisting 56. Shafrir E, Ziv E and Kalman R. Nutritionally induced Acomys species. Oecologia 1999; 121: 123–128. diabetes in desert rodents as models of type 2 diabetes: 42. Wube T, Fares F and Haim A. A differential response in Acomys cahirinus (spiny mice) and Psammomys obesus the reproductive system and energy balance of spiny (desert gerbil). ILAR J 2006; 47: 212–224. mice Acomys populations to vasopressin treatment. 57. Kumar S, Singh R, Vasudeva N, et al. Acute and chronic Comp Biochem Physiol A Mol Integr Physiol 2008; 151: animal models for the evaluation of anti-diabetic agents. 499–504. Cardiovasc Diabetol 2012; 11: 9. Pinheiro et al. 575

58. Shafrir E. Overnutrition in spiny mice (Acomys cahiri- the diaphragm of the newborn spiny mouse from intra- nus): Beta-cell expansion leading to rupture and overt partum hypoxia-induced damage. Pediatr Res 2010; 68: diabetes on fat-rich diet and protective energy-wasting 393–398. elevation in thyroid hormone on sucrose-rich diet. 69. Hutton LC, Abbass M, Dickinson H, et al. Diabetes Metab Res Rev 2000; 16: 94–105. Neuroprotective properties of melatonin in a model of 59. Brunjes P. A comparative study of prenatal development birth asphyxia in the spiny mouse (Acomys cahirinus). in the olfactory bulb, neocortex and hippocampal region Dev Neurosci 2009; 31: 437–451. of the precocial mouse Acomys cahirinus and rat. Brain 70. Ireland Z, Castillo-Melendez M, Dickinson H, et al. Res Dev Brain Res 1989; 49: 7–25. A maternal diet supplemented with creatine from mid- 60. Brunjes PC. Olfactory bulb maturation in Acomys cahir- pregnancy protects the newborn spiny mouse brain inus: Is neural growth similar in precocial and altricial from birth hypoxia. Neuroscience 2011; 194: 372–379. murids? Brain Res 1983; 8: 335–341. 71. Ireland Z, Dickinson H, Snow R, et al. Maternal creatine: 61. Brunjes PC. A stereological study of neocortical matur- does it reach the fetus and improve survival after an acute ation in the precocial mouse, Acomys cahirinus. Brain hypoxic episode in the spiny mouse (Acomys cahirinus)? Res 1985; 19: 279–287. Am J Obstet Gynecol 2008; 198: 431.e1–431.e6. 62. Dickinson H, Walker DW, Cullen-McEwen L, et al. The 72. Ireland Z, Russell AP, Wallimann T, et al. spiny mouse (Acomys cahirinus) completes nephrogenesis Developmental changes in the expression of creatine before birth. Am J Physiol Renal Physiol 2005; 289: synthesizing enzymes and creatine transporter in a preco- F273–F279. cial rodent, the spiny mouse. BMC Dev Biol 2009; 9: 39. 63. Lamers WH, Mooren PG, Graaf A, et al. Perinatal devel- 73. Brunjes PC. The precocial mouse, Acomys cahirinus. opment of the liver in rat and spiny mouse. Eur J Biochem Psychobiology 1990; 18: 339–350. 1985; 146: 475–480. 74. Brunjes PC, Korol DL and Stern KG. Prenatal neuro- 64. Oosterhuis W, Mooren P, Charles R, et al. Perinatal genesis in the telencephalon of the precocial mouse development of the lung in rat and spiny mouse: its rela- Acomys cahirinus. Neurosci Lett 1989; 107: 114–119. tion to altricial and precocial timing of birth. Biol 75. Sanchez-Alvarado A. Regeneration in the metazoans: Neonate 1984; 45: 236–243. Why does it happen? Bioessays 2000; 22: 578–590. 65. O’Connell B, Moritz K, Walker D, et al. Treatment of 76. Gawriluk TR, Simkin J, Thompson KL, et al. pregnant spiny mice at mid gestation with a synthetic Comparative analysis of ear-hole closure identifies epi- glucocorticoid has sex-dependent effects on placental morphic regeneration as a discrete trait in mammals. glycogen stores. Placenta 2013; 34: 932–940. Nature Commun 2016; 7: 11164. 66. O’Connell BA, Moritz KM, Roberts CT, et al. The pla- 77. Brant JO, Yoon JH, Polvadore T, et al. Cellular events cental response to excess maternal glucocorticoid expos- during scar-free skin regeneration in the spiny mouse, ure differs between the male and female conceptus in Acomys. Wound Repair Regen 2016; 24: 75–88. spiny mice. Biol Reprod 2011; 85: 1040–1047. 78. Simkin J, Gawriluk TR, Gensel JC, et al. Macrophages 67. Hutton LC, Ratnayake U, Shields A, et al. are necessary for epimorphic regeneration in African Neuropathology and functional deficits in a model of spiny mice. Elife 2017; 6: e24623. birth asphyxia in the precocial spiny mouse (Acomys 79. Li K, Wang H, Cai Z, et al. Sympatric speciation of spiny cahirinus). Dev Neurosci 2009; 31: 523–535. mice, Acomys, unfolded transcriptomically at Evolution 68. Cannata DJ, Ireland Z, Dickinson H, et al. Maternal Canyon, Israel. Proc Natl Acad Sci USA 2016; 113: creatine supplementation from mid-pregnancy protects 8254–8259.

Re´sume´ La souris e´pineuse d’Afrique (Acomys) est un nouveau mode`le animal dote´ de remarquables caracte´ristiques biologiques qui en font un sujet d’inte´reˆt pour un large e´ventail de domaines de recherche. Les souris e´pineuses appartiennent a` une espe`ce adapte´e au milieu de´sertique et a` un re´gime alimentaire faible en calories et de´veloppent des symptoˆmes lie´s au diabe`te lorsqu’elles passent a` des re´gimes alimentaires riches en e´nergie. Les souris e´pineuses subissent des pe´riodes de gestation relativement longues et ont de petites porte´es de souriceaux tre`sde´veloppe´s, ce qui en fait un mode`le ade´quat pour la fin de l’organo- ge´ne`se et la biologie pe´rinatale. Il a re´cemment e´te´ de´montre´ qu’elles posse´dent des capacite´sdere´ge´ne´ra- tion et de gue´rison remarquables qui les rendent uniques parmi les mammife`res. Dans ce travail, nous de´crivons notre expe´rience d’he´bergement d’une colonie de souris e´pineuses africaines et couvrons tous les aspects fondamentaux de l’alimentation, de la maintenance et de la reproduction a` des fins de recherche.

Abstract Die afrikanische Stachelmaus (Acomys sp) ist ein vielversprechendes Tiermodell mit bemerkenswerten bio- logischen Eigenschaften, die sie fu¨r verschiedenste Forschungsgebiete interessant machen. Diese typische Wu¨stenspezies, die an eine kalorienarme Erna¨hrung angepasst ist, entwickelt diabetesbedingte Symptome, 576 Laboratory Animals 52(6) wenn sie auf eine energiereiche Erna¨hrung umgestellt werden. Stachelma¨use durchlaufen relativ lange Tra¨chtigkeitszeiten und haben kleine Wu¨rfe hoch entwickelter Jungtiere, sodass sie ein ada¨quates Modell fu¨r die spa¨te Organogenese und Perinatalbiologie darstellen. In ju¨ngster Zeit wurden bei ihnen bemerkens- werte Heil- und Regenerationsfa¨higkeiten festgestellt, die sie unter den Sa¨ugetieren einzigartig machen. In dieser Arbeit beschreiben wir unsere Erfahrungen, die wir bei der Unterbringung einer Kolonie afrikanischer Stachelma¨use gesammelt haben, und behandeln alle grundlegenden Aspekte der Fu¨tterung, Pflege und Zucht zu Forschungszwecken.

Resumen El rato´n africano Spiny (Acomys sp) es un modelo animal emergente con caracterı´sticas biolo´gicas remarc- ables que hace que sea sujeto de intere´s para una amplia gama de campos de investigacio´n. Normalmente se trata de una especie de desierto adaptada a una dieta baja en calorı´as, los ratones Spiny desarrollan sı´ntomas relacionados con la diabetes al pasar a tomar una dieta alta en energı´a. El rato´n Spiny pasa por periodos de gestacio´n relativamente largos y tiene pequen˜as camadas de crı´as muy desarrolladas, lo que hace que sea un modelo adecuado para la biologı´a perinatal y organoge´nesis tardı´a. Recientemente ha demostrado tener una gran capacidad de regeneracio´n y curacio´n lo que hace que sea una especie u´nica entre los mamı´feros. En este estudio describimos nuestra experiencia al enjaular una colonia de ratones africanos Spiny y al cubrir todos los aspectos ba´sicos de alimentacio´n, mantenimiento y crı´a con fines de investigacio´n. Original Article Laboratory Animals 2018, Vol. 52(6) 577–587 ! The Author(s) 2018 Diurnally active rodents for Article reuse guidelines: sagepub.com/journals- laboratory research permissions DOI: 10.1177/0023677218771720 journals.sagepub.com/home/lan Roberto Refinetti1 and GJ Kenagy2

Abstract Although inbred domesticated strains of rats and mice serve as traditional mammalian animal models in biomedical research, the nocturnal habits of these rodents make them inappropriate for research that requires a model with human-like diurnal activity rhythms. We conducted a literature review and recorded locomotor activity data from four rodent species that are generally considered to be diurnally active, the Mongolian gerbil (Meriones unguiculatus), the degu (Octodon degus), the African (Nile) grass rat (Arvicanthis niloticus), and the antelope ground squirrel (Ammospermophilus leucurus). Our data collected under 12-hour light/dark cycles confirmed and expanded the existing literature in showing that the activity rhythms of antelope ground squirrels and African grass rats are stronger and more concentrated in the light phase of the light/dark cycle than the activity rhythms of Mongolian gerbils and degus, making the former two species preferable and more reliable as models of consistent diurnal activity in the laboratory. Among the two more strongly diurnal species, antelope ground squirrels are more exclusively diurnal and have more robust activity rhythms than African grass rats. Although animals of these two species are not currently available from commercial suppliers, African grass rats are indigenous to a wide area across the north of Africa and thus available to researchers in the eastern hemisphere, whereas antelope ground squirrels can be found throughout much of western North America’s desert country and, therefore, are more easily accessible to North American researchers.

Keywords rodent, behavior, diurnal, nocturnal, locomotion

Date received: 11 January 2018; accepted: 27 March 2018

Domesticated and inbred strains of rats (Rattus norve- rats or mice. Thus, substitution of diurnal rodents for gicus) and mice (Mus musculus) are the most extensively nocturnal rodents may constitute a significant advance used mammalian animal models in biomedical research. in the contribution of animal research to human health. These standard laboratory rodents are considered to be Several rodent species that are reportedly diurnal in suitable animal models because they share with humans the field have been tested in the laboratory and found many aspects of mammalian physiology. They are con- to show alternative patterns of activity. These species venient substitutes for humans or other primates include the fat sand rat (Psammomys obesus),4 the because of their small body size and the ease and low Catamarca tuco-tuco (Ctenomys knighti),5 and the cost of their housing and maintenance. Nonetheless, rats and mice are nocturnal (active at night), whereas 1Circadian Rhythm Laboratory, Department of Psychological humans are diurnal (active during the day). Science, Boise State University, USA Every biological process shows a pattern of circadian 2Department of Biology and Burke Museum, University of rhythmicity, with organisms exhibiting clear differences Washington, USA in their physiology between day and night.1–3 For this Corresponding author: reason and because humans are diurnal, diurnally Roberto Refinetti, Department of Psychological Science, Boise active rodent species should potentially be more appro- State University, 1910 University Drive, Boise, ID 83725, USA. priate models for biomedical laboratory research than Email: [email protected] 578 Laboratory Animals 52(6) golden spiny mouse (Acomys russatus).6 We are aware of four rodent species that are diurnally active in the field and have been described as diurnal in the labora- tory. These are the Mongolian gerbil (Meriones ungui- culatus), the degu (Octodon degus), the African grass rat (Arvicanthis niloticus), and the antelope ground squirrel (Ammospermophilus leucurus). Here, we review the lit- erature on field and laboratory studies on these species and present original experimental data comparing their daily temporal patterns of locomotor activity.

Materials and methods Bibliographic research Literature searches were conducted in October 2017 in two main databases: BIOSIS (Clarivate Analytics, formerly Thomson Reuters) and PubMed (United States National Library of Medicine). The two data- bases contain approximately 26 million records each, with BIOSIS focusing on field life-science studies and PubMed focusing on biomedical studies. Searches were conducted for the species name (genus plus species). In BIOSIS, the search was restricted to Taxonomic Data so as to avoid spurious retrieval of species names appearing in reference lists rather than being the subject of a study. PubMed does not index the references cited in an article, so we were able to include All Fields in the PubMed search. PubMed was queried also for species name AND ‘‘circadian’’, so that articles dealing specifically with daily/circadian rhythmicity could be retrieved separately from articles dealing with all other biological processes such as seed dispersal or snake predation.

Animals The experimental part of the study involved 24 individ- uals (12 males and 12 females) of each species, all adults between the ages of 3 and 5 months. This number of subjects was calculated to be sufficient to attain statis- tical power of 0.8, based on the variances of the means in preliminary studies. Mongolian gerbils (Figure 1a) were purchased from Charles River Laboratories (Wilmington, MA), with a mean body mass of 88 g. Degus (Figure 1b) were pur- chased from Sandy’s Lakeside (Gaffney, SC), with a mean body mass of 237 g. African grass rats, also called Nile grass rats (Figure 1c), were bred in a local 7 Figure 1. The four species compared in the present study. colony with founders originally trapped in Kenya and (a) Mongolian gerbil (Meriones unguiculatus) photographed weighed 119 g. Antelope ground squirrels, occasionally by Stefan Ko¨der. (b) Degu (Octodon degus) photographed by also called white-tailed antelope squirrels (Figure 1d), Arjan Haverkamp. (c) African grass rat (Arvicanthis niloti- were born in captivity from pregnant females trapped cus) photographed by Roberto Refinetti. (d) Antelope by us in the field in Owyhee County, Idaho, and ground squirrel (Ammospermophilus leucurus) photo- weighed 121 g. graphed by Roberto Refinetti. Refinetti and Kenagy 579

The conservation status listed by the International activity onsets), 10 day segments of the activity records Union for Conservation of Nature for all four species of each animal were selected for analysis. The time is ‘‘Least Concern.’’ Mongolian gerbils are indigenous to series were analyzed with computer programs written China and Mongolia,8 but are easily available from com- specifically for this study or with standard programs mercial suppliers. Degus are indigenous to Chile9 and, from the Circadian Physiology software package.3 although not available through major animal suppliers, Analysis involved the computation of six parameters: are available from pet stores in many countries. African diurnality, acrophase, daily onset, alpha, robustness, grass rats are indigenous to much of northern Africa10 and distance traveled. and are not bred commercially. Antelope ground squir- Diurnality was computed as the number of wheel rels are also unavailable commercially but are indigen- revolutions during the light phase of the light/dark ous to much of western North America,11 which makes cycle divided by the total number of revolutions for them accessible with relative ease to researchers in North the whole day. The acrophase (center of gravity of America. Most ground squirrel species are hibernators,12 the activity rhythm) was computed by the single cosinor but antelope ground squirrels remain active on the sur- procedure.18,19 The daily onset of activity (initiation of face in the field throughout the entire year.13–16 running-wheel activity) was computed by an algorithm that smoothed the daily rhythm with an 8 hour moving- Procedures window filter and then identified the onset of activity as the time when the smoothed curve rose above the daily The experimental procedures were approved by the mean. Alpha (duration of the activity phase of the daily institutional animal care and use committees of the cycle) was calculated as the difference between the end University of South Carolina (Mongolian gerbils, (offset) and the beginning (onset) of activity, with degus, and African grass rats) and of Boise State the end computed similarly to the beginning but using University (antelope ground squirrels) in accordance the time of the descent of the smoothed curve below the with the guidelines of the United States National daily mean. Robustness (strength of rhythmicity) was Research Council’s Guide for the Care and Use of computed as the QP value of the chi-squared period- Laboratory Animals. ogram statistic as a percentage of the maximal QP value All animals were housed individually in polypropyl- for the dataset.20 The distance traveled (km/day) was ene cages (36 cm length, 24 cm width, 19 cm height) computed from the number of wheel revolutions and lined with wood shavings, and kept at 24C under a the circumference of the wheel. Differences between light/dark cycle with 12 hours of light per day group means were evaluated by factorial analyses of (360:0 lux). For reference, the illuminance provided by variances (ANOVAs), with species and sex as the fac- a full moon is about 0.1 lux, that of the average human tors, followed by pairwise comparisons with Tukey’s indoor working space is 200 lux, and outdoor daylight test, using OpenStat.21 exceeds 1000 lux.17 Purina rodent chow was provided ad libitum on the metal cage top, which also held a water bottle with a sipping tube. Results Each cage was equipped with a metallic running Literature review wheel (18 cm diameter for degus or 15 cm for the smal- ler species). A small magnet attached to the wheel The quantitative results of the literature search are activated a magnetic switch affixed to the top of the shown in Table 1. In both BIOSIS and PubMed, the cage and connected to data acquisition boards, and search for M. unguiculatus yielded four times as many activity counts were saved at 0.1 hour intervals. We documents as for the next most-represented species, monitored activity with running wheels because they O. degus. Searches in BIOSIS consistently retrieved are a traditional apparatus in research on circadian more documents than in PubMed. When the PubMed rhythms and also because they provide the animals search was restricted to documents containing the word with the opportunity to engage in physical exercise. circadian, the number of retrieved documents was The raw data reported in this article were greatly reduced. The fraction of documents relevant deposited in the Open Science Framework archives at to circadian biology was 2% for M. unguiculatus, osf.io/a7kjc. 20% for O. degus, and 40% for A. niloticus and A. leucurus. Data analysis The literature indicates that Mongolian gerbils are active during the day in the wild.22,23 Some laboratory After stable synchronization of the activity rhythm studies reported the activity rhythm as crepuscular, to the light/dark cycle (defined as 14 consecutive days with peaks at dawn and dusk and perhaps a slight diur- with less than 30 minutes of variability of daily nal preponderance of activity,24,25 whereas others 580 Laboratory Animals 52(6)

Table 1. Quantitative results (number of published studies) from the literature search.

BIOSIS PubMed A PubMed B

Meriones unguiculatus (Mongolian gerbil) 1841 1441 31 Octodon degus (Degu) 465 296 57 Arvicanthis niloticus (African grass rat) 240 147 64 Ammospermophilus leucurus 33 10 4 (Antelope ground squirrel)

PubMed A: search for species name only; PubMed B: search for species name AND circadian. described gerbil activity in the laboratory as noc- Experimental findings turnal.26,27 One study showed that the activity pattern of gerbils was diurnal when measured with motion All four species were easily maintained in the labora- detectors, but nocturnal when measured with running tory. The animals could be routinely moved and wheels.28 Two other investigations using running weighed with the help of a 600 ml plastic cup. Manual wheels exclusively found that some individual gerbils handling of individually housed rodents is often diffi- were diurnal while others were nocturnal.29,30 It cult, although even species not traditionally kept as seems, therefore, that the activity rhythm of the pets, such as the African grass rat, can be handled if Mongolian gerbil in the laboratory is not robust and suitably trained.63 may be expressed as either diurnal or nocturnal, The most compact and exclusively diurnal running depending on the method used to record activity, and pattern among the four species was that of the antelope furthermore that different individuals may simply ground squirrel, followed in rank by the African grass express either diurnal or nocturnal activity as a rat, the Mongolian gerbil, and the degu (Figure 2). matter of individual peculiarity. The selected actograms show the records of strong Degus in the wild are diurnal, exhibiting a bimodal and consistent runners that exhibited robust rhythmi- pattern of intensity with prominent peaks at dawn and city, rather than ‘‘average’’ runners. Most individuals dusk during the summer and with a continuous pattern of all four species ran on their wheels more than 1 km throughout the midday during the shorter days per day. Although the actograms of gerbils and degus of winter when environmental temperature does not showed a strong component of daytime activity, they inhibit midday activity.31,32 Laboratory studies show also showed considerable nighttime activity (Figure 2), similar patterns,33–38 with some studies demonstrating resulting in a lower fidelity of their overall activity to bimodality39 and others reporting predominantly unim- the daytime. odality.40 Similar to observations on Mongolian Antelope ground squirrels and African grass rats gerbils, at least two investigations have reported showed the smallest range of variability in the distribu- degus to become nocturnal in the laboratory simply tion of individual daily onsets of activity (Figure 3). upon gaining access to a running wheel.41,42 However, Individuals of these two species consistently started approximately half of the degus in two other studies activity about the time of lights-on, for a uniform diur- were diurnal and half were nocturnal regardless of the nal pattern. In contrast, about half of the gerbils presence or absence of running wheels, apparently as a and degus started activity about the time of lights-on matter of individual peculiarity.30,41 Thus, although (consistent with diurnal activity) and the other half degus appear to be more diurnal than Mongolian ger- about the time of lights-off (consistent with nocturnal bils, they do not seem to be reliably and consistently activity). diurnal under laboratory conditions. We compared six aspects of daily running-wheel African grass rats are diurnal in the wild,43 and most activity in the four species (Figure 4). Diurnality laboratory studies have shown this species to be pre- varied significantly among species, with antelope dominantly diurnal.44–49 Two studies reported a switch ground squirrels and African grass rats showing the to nocturnality in African grass rats with access to run- greatest diurnality, Mongolian gerbils intermediate ning wheels,50,51 but animals studied with running diurnality, and degus the least diurnal (Figure 4a; wheels in two other laboratories were found to be pre- ANOVA effect of species: F3, 88 ¼ 69.889, p < 0.001). 52–54 dominantly diurnal. There were no significant effects of sex (F1, 88 ¼ 2.004, Antelope ground squirrels are diurnal in the p ¼ 0.157) or the interaction of species and sex 55 wild. Eight investigations all demonstrate that the (F3, 88 ¼ 1.671, p ¼ 0.1776). species is unquestionably diurnal under laboratory As expected for diurnal animals, the acrophase of conditions.56–62 the activity rhythm occurred on average during the Refinetti and Kenagy 581

The duration of the active phase of the circadian cycle (alpha) was similar across the species, with an overall mean of approximately 13 hours (Figure 4d). Activity duration was shortest in antelope ground squirrels (mean: 10.5 hours), which is reflected by a significant effect of species in the ANOVA (F3, 88 ¼ 13.727, p < 0.001). There were no significant effects of sex (F1, 88 ¼ 0.533, p ¼ 0.526) or the inter- action of species and sex (F3, 88 ¼ 2.023, p ¼ 0.115). All individuals of all species showed statistically sig- nificant robustness of the activity rhythm, and antelope ground squirrels showed the greatest robustness of all four species (Figure 4e). ANOVA confirmed a signifi- cant effect of species (F3, 88 ¼ 18.264, p < 0.001) but no significant effects of sex (F1, 88 ¼ 1.276, p ¼ 0.260) or the interaction of species and sex (F3, 88 ¼ 0.683, p ¼ 0.568). Distance traveled (Figure 4f) was the only parameter that showed any differences between males and females. These effects occurred in only two of the four species: female African grass rats ran less than males and female antelope ground squirrels ran slightly more than males, as reflected by a significant interaction of species and sex (F3, 88 ¼ 3.346, p ¼ 0.022) without a main effect of sex per se (F1, 88 ¼ 0.756, p ¼ 0.609). Regarding species differences in amount of running, antelope ground squirrels showed the greatest amount of running, Mongolian gerbils the least, and the other two species were intermediate (F ¼ 30.094, p < 0.001). Figure 2. Actograms of the locomotor activity records of 3, 88 representatives of each of the four species tested. Actograms are plotted with the time of day on the hori- Discussion zontal axis and successive days on the vertical axis. Our literature review revealed various degrees to which diurnality is expressed by each of the four species, and our quantitative experimental comparisons demon- light phase of the light/dark cycle in three of the species, strated marked differences among the species in their but not in degus, for which the mean acrophase suitability as model species based on substantial differ- occurred at the beginning of the dark phase ences in the consistency and relative exclusivity of their (Figure 4b). This divergence of degus from the expected daytime activity. Antelope ground squirrels showed pattern yielded a significant difference between species the strongest, most robust, and most consistent pat- means (F3, 88 ¼ 32.481, p < 0.001). There were no sig- terns of diurnal activity, followed by African grass nificant effects of sex (F1, 88 ¼ 0.014, p ¼ 0.904) or the rats, which had consistent diurnal activity, and finally interaction of species and sex (F3, 88 ¼ 0.958, p ¼ 0.582). by Mongolian gerbils and degus, both of which were Antelope ground squirrels and African grass rats only partially diurnal. showed the expected pattern for onset of activity in We suggest four primary considerations that should diurnal animals with activity beginning at or near the be taken in the choice of a diurnal laboratory animal time of lights-on (Figure 4c). Plots of the onsets of model: the degree of exclusivity of diurnality, the avail- activity for Mongolian gerbils and degus were obscured ability and cost of the animals, their laboratory suit- because about half the individuals of each of these spe- ability (particularly body size), and the breadth and cies showed diurnal activity and half showed nocturnal depth of the scientific literature about the species. activity, thus producing a later hour in the day for We have shown that the problem with Mongolian mean onset. Accordingly, ANOVA indicated a signifi- gerbils is their poor degree of diurnality, in that about cant effect of species (F3, 88 ¼ 14.693, p < 0.001), half of them are not consistently active during the day. although no significant effects of sex (F1, 88 ¼ 0.944, Mongolian gerbils are the most easily available of the p ¼ 0.665) or the interaction of species and sex four species, as they are sold by many suppliers of (F3, 88 ¼ 0.144, p ¼ 0.933). laboratory animals and are quite suitable for the 582 Laboratory Animals 52(6)

Figure 3. Frequency distributions of the daily onsets of activity for the four species (n ¼ 24 individuals per species). The value for each individual is the average onset over 10 days. laboratory, weighing about 90 g. These gerbils have Mongolian gerbils, at approximately 240 g, but still been the subject of more previous research than the small enough for convenient housing in the laboratory. other three species combined. Taking into account Degus have not been used as research subjects as much that individual gerbils may vary considerably in the as Mongolian gerbils, but considerable knowledge has degree of their diurnality,4,34,52 selective breeding for accumulated about their physiology. a diurnal disposition could be attempted, although African grass rats are consistently diurnal and exhibit there is no guarantee that selective breeding would be robust rhythmicity of activity. Considerable knowledge successful or that it would not unintentionally affect has accumulated about their anatomy and physiology, other traits. particularly neuroanatomy and neurophysiology. As with Mongolian gerbils, degus are not reliably African grass rats are smaller than laboratory rats but diurnal in the laboratory, and the robustness of their larger than laboratory mice, at approximately 120 g. activity rhythm is the lowest of the four species we The availability of African grass rats, especially for studied. Degus are not available from major suppliers researchers in North America, is not good. Unless of laboratory animals but can be obtained from pet one can obtain breeding pairs from a researcher who stores in many countries. They are larger than already has an African grass rat colony, a new Refinetti and Kenagy 583

Figure 4. (a–f) Mean values (SEM) of each of the six computed statistics for each of the four species studied. In each panel, bars with dissimilar letters are significantly different from each other, as determined by post hoc Tukey’s tests (p < 0.05). The dashed lines in (b) and (c) indicate the daily time of lights-on (08:00). investigator would need to travel to Africa to trap ani- activity duration less similar to that of humans mals for research. (a ¼ 15.5 hours). On the other hand, the free-running Antelope ground squirrels are the most strongly and period of the antelope ground squirrel ( ¼ 24.2 consistently diurnal of the four species (diurnality hours)62 matches the human free-running period index: 0.97, rhythm robustness: 41%). The duration ( ¼ 24.2 hours).64 Like African grass rats, antelope of the active phase of the daily cycle is shorter in ante- ground squirrels are smaller than laboratory rats but lope ground squirrels (a ¼ 10.5 hours) than in African larger than laboratory mice, weighing approximately grass rats (a ¼ 13.5 hours), which raises the diurnality 120 g. Although antelope ground squirrels are not cur- index of the antelope ground squirrel but makes its rently available from major suppliers of laboratory 584 Laboratory Animals 52(6) animals or from pet stores, they can be trapped in greater validity and relevance as models of the circa- the field over a large part of western North America, dian organization of physiology and behavior. The where they are often the only diurnal rodents inhabiting award of the 2017 Nobel Prize in Physiology/ desert areas.11 Additionally, although most laboratory Medicine to circadian biologists73 may have marked a rodents have a short life of only two or three years, turning point in the attention that biomedical science antelope ground squirrels have been reported to live will pay in the future to the importance of biological up to 8 years in captivity.60 Given all of these advan- clocks. tages and a recent call for increased diversity of animal 65 models, the antelope ground squirrel seems to be a Declaration of Conflicting Interests particularly valuable option for a diurnal rodent for The author(s) declared no potential conflicts of interest with laboratory research. respect to the research, authorship, and/or publication of this Although we housed all four species under the same article. conditions to avoid the confounding effects of uncon- trolled variables, we recognize that different species of laboratory rodents may thrive under distinct housing Funding conditions. We did not observe abnormal behavior in The author(s) received no financial support for the research, individuals of any of the four species, but stereotypies authorship, and/or publication of this article. have been described in laboratory-housed Mongolian gerbils66,67 and have been attributed to impoverished ORCID iD 68 housing conditions. Roberto Refinetti http://orcid.org/0000-0003-2875-2389. The matter of procurement of animals is worthy of emphasis. Although the growing use of genetically References modified rodents or rodents raised under special condi- tions has led many research facilities to breed their own 1. Dunlap JC, Loros JJ and DeCoursey PJ. Chronobiology: animals rather than buy them from commercial bree- biological timekeeping. Sunderland: Sinauer, 2004. 2. Koukkari WL and Sothern RB. Introducing biological ders, the introduction of wild rodents into a vivarium rhythms. New York: Springer, 2006. would meet with some initial challenges. Animals cap- 3. Refinetti R. Circadian physiology, 3rd ed. Boca Raton: tured in the field would have to be held in quarantine CRC Press, 2016. and bred to the next generation before they could be 4. Cohen R and Kronfeld-Schor N. Individual variability introduced into a pathogen-free facility, and protocols and photic entrainment of circadian rhythms in golden for breeding them outside the main facility to develop spiny mice. Physiol Behav 2006; 87: 563–574. clean animals would have to be developed on a case- 5. Barak O and Kronfeld-Schor N. Activity rhythms and by-case basis. Thus, the adoption of a species such as masking response in the diurnal fat sand rat under the antelope ground squirrel would require a significant laboratory conditions. Chronobiol Int 2013; 30: investment in animal facilities, presumably justified by 1123–1134. the value of its extreme and exclusive diurnal activity 6. Tachinardi P, Bicudo JEW, Oda GA, et al. Rhythmic 24 pattern. Compared to establishing a primate colony, h variation of core body temperature and locomotor activity in a subterranean rodent (Ctenomys aff. knighti), the investments required for a new rodent model species the tuco-tuco. PLOS One 2014; 9: e85674. would likely be much less. 7. Refinetti R. The Nile grass rat as a laboratory animal. Also of importance regarding the adoption of a new Lab Animal (NY) 2004; 33(9): 54–57. model species is the matter of methodological resources 8. Gulotta EF. Meriones unguiculatus. Mammalian Species for research. At the present time, methodological 1971; 3: 1–5. resources for research on diurnally active rodents are 9. Woods CA and Boraker DK. Octodon degus. Mammalian far fewer than for the nocturnally active mouse and rat. Species 1975; 67: 1–5. The availability of stereotaxic atlases for neuroscience 10. Musser GG and Carleton MD. Arvicanthis. In: Wilson research,69,70 for example, is only one such resource. DE and Reeder DAM (eds) Mammals of the world: a Sequencing of the mouse and rat genomes in the early taxonomic and geographic reference. Vol 2, 3rd ed. 2000s71,72 opened the door to unprecedented advances Baltimore: Johns Hopkins University Press, 2005, pp.1285–1292. in rodent genomics, but fortunately with ongoing 11. Koprowski JL, Goldstein EA, Bennett KR, et al. Family advances in genomics the sequencing of new genomes Sciuridae, Genus Ammospermophilus. In: Wilson DE, becomes easier with each passing year. Thus, we can be Lacher TE and Mittermeier RA (eds) Handbook of the encouraged with the possibility that these resources mammals of the world. Vol 6, Barcelona: Lynx, 2016, could also be developed in order to facilitate the pp.795–796. advances in biomedical research that could be made 12. Ruby NF. Hibernation: when good clocks go cold. J Biol with diurnally active rodents taking into account their Rhythms 2003; 18: 275–286. Refinetti and Kenagy 585

13. Hudson JW. The role of water in the biology of the ante- 33. Labyak SE and Lee TM. Estrus- and steroid-induced lope ground squirrel, Citellus leucurus. Univ Calif Publ changes in circadian rhythms in a diurnal rodent, Zool 1962; 64: 1–56. Octodon degus. Physiol Behav 1995; 58: 573–585. 14. Chappell MA and Bartholomew GA. Activity and 34. Labyak SE, Lee TM and Goel N. Rhythm chronotypes in thermoregulation of the antelope ground squirrel a diurnal rodent, Octodon degus. Am J Physiol 1997; 273: Ammospermophilus leucurus in winter and summer. R1058–R1066. Physiol Zool 1981; 54: 215–223. 35. Kas MJH and Edgar DM. Photic phase response curve in 15. Karasov WH. Wintertime energy conservation by hud- Octodon degus: assessment as a function of activity phase dling in antelope ground squirrels (Ammospermophilus preference. Am J Physiol 200; 278: R1385–R1389. leucurus). J Mammal 1983; 64: 341–345. 36. Mohawk JA and Lee TM. Restraint stress delays reen- 16. Kenagy GJ and Bartholomew GA. Seasonal reproductive trainment in male and female diurnal and nocturnal patterns in five coexisting California desert rodent spe- rodents. J Biol Rhythms 2005; 20: 245–256. cies. Ecol Monogr 1985; 55: 371–397. 37. Refinetti R. Enhanced circadian photoresponsiveness 17. Rea MS (ed.). The IESNA lighting handbook. 9th ed. after prolonged dark adaptation in seven species of diur- New York: Illuminating Engineering Society of North nal and nocturnal rodents. Physiol Behav 2007; 90: America, 2000. 431–437. 18. Nelson W, Tong YL, Lee JK, et al. Methods for cosinor 38. Ashkenazy-Frolinger T, Einat H, et al. Diurnal rodents rhythmometry. Chronobiologia 1979; 6: 305–323. as an advantageous model for affective disorders: novel 19. Refinetti R, Corne´lissen G and Halberg F. Procedures for data from diurnal rodents. J Neural Transm 2015; 122: numerical analysis of circadian rhythms. Biol Rhythm Res S35–S45. 2007; 38: 275–325. 39. Garcı´a-Allegue R, Lax P, Madariaga AM, et al. 20. Refinetti R. Non-stationary time series and the robustness Locomotor and feeding activity rhythms in a light- of circadian rhythms. J Theor Biol 2004; 227: 571–581. entrained diurnal rodent, Octodon degus. Am J Physiol 21. Miller W. OpenStat reference manual. New York: 1999; 277: R523–R531. Springer, 2013. 40. Refinetti R. Rhythms of body temperature and tempera- 22. Agren G, Zhou Q and Zhong W. Ecology and social ture selection are out of phase in a diurnal rodent, behaviour of Mongolian gerbils, Meriones unguiculatus, Octodon degus. Physiol Behav 1996; 60: 959–961. at Xilinhot, Inner Mongolia, China. Anim Behav 1989; 41. Kas MJH and Edgar DM. A nonphotic stimulus inverts 37: 11–27. the diurnal-nocturnal phase preference in Octodon degus. 23. Liu W, Wan X and Zhong W. Population dynamics of J Neurosci 1999; 19: 328–333. the Mongolian gerbils: seasonal patterns and interactions 42. Ocampo-Garce´s A, Herna´ndez F, Mena W, et al. Wheel- among density, reproduction and climate. J Arid running and rest activity pattern interaction in two octo- Environm 2007; 68: 383–397. dontids (Octodon degus, Octodon bridgesi). Biol Res 2005; 24. Stutz AM. Diurnal rhythms of spontaneous activity in 38: 299–305. the Mongolian gerbil. Physiol Zool 1972; 45: 325–334. 43. Blanchong JA and Smale L. Temporal patterns of activ- 25. Pietrewicz AT, Hoff MP and Higgins SA. Activity ity of the unstriped Nile rat, Arvicanthis niloticus. rhythms in the Mongolian gerbil under natural light con- ditions. Physiol Behav 1982; 29: 377–380. J Mammal 2000; 81: 595–599. 26. Nelissen M and Nelissen-Joris N. On the diurnal rhythm 44. McElhinny TL, Smale L and Holekamp KE. Patterns of of activity of Meriones unguiculatus (Milne-Edwards, body temperature, activity, and reproductive behavior in 1867). Acta Zool Patol Antverp 1975; 61: 25–30. a tropical murid rodent, Arvicanthis niloticus. Physiol 27. Gattermann R and Weinandy R. Lack of social entrain- Behav 1997; 62: 91–96. ment of circadian activity rhythms in the solitary golden 45. Novak CM and Albers HE. N-methyl-D-aspartate hamster and in the highly social Mongolian gerbil. Biol microinjected into the suprachiasmatic nucleus mimics Rhythm Res 1997; 28(S): 85–93. the phase-shifting effects of light in the diurnal Nile 28. Weinert D, Weinandy R and Gattermann R. Photic and grass rat (Arvicanthis niloticus). Brain Res 2002; 951: non-photic effects on the daily activity pattern of 255–263. Mongolian gerbils. Physiol Behav 2007; 90: 325–333. 46. Refinetti R. Absence of circadian and photoperiodic con- 29. Umezu T, Kuribara H and Tadokoro S. Characteristics servation of energy expenditure in three rodent species. of circadian rhythm of wheel-running activity and drink- J Comp Physiol B 2007; 177: 309–318. ing behavior in Mongolian gerbils. Yakubutsu Seishin 47. Schrader JA, Walaszczyk EJ and Smale L. Changing pat- Kodo 1989; 9: 369–373. terns of daily rhythmicity across reproductive states in 30. Refinetti R. Variability of diurnality in laboratory diurnal female Nile grass rats (Arvicanthis niloticus). rodents. J Comp Physiol A 2006; 192: 701–714. Physiol Behav 2009; 98: 547–556. 31. Kenagy GJ, Nespolo RF, Va´squez RA, et al. Daily and 48. Gall AJ, Smale L, Yan L, et al. Lesions of the intergeni- seasonal limits of time and temperature to activity of culate leaflet lead to a reorganization in circadian regu- degus. Rev Chil Hist Nat 2002; 75: 567–581. lation and a reversal in masking responses to photic 32. Ocampo-Garce´s A, Mena W, Herna´ndez F, et al. stimuli in the Nile grass rat. PLoS One 2013; 8: e67387. Circadian chronotypes among wild-captured west 49. Gall AJ, Shuboni DD, Yan L, et al. Suprachiasmatic Andean octodontids. Biol Res 2006; 39: 209–220. nucleus and subparaventricular zone lesions disrupt 586 Laboratory Animals 52(6)

circadian rhythmicity but not light-induced masking antelope ground squirrel and heterothermic Syrian ham- behavior in Nile grass rats. J Biol Rhythms 2016; 31: ster. Biol Rhythm Res 1998; 29: 311–325. 170–181. 62. Refinetti R and Kenagy GJ. Circadian rhythms of body 50. Blanchong JA, McElhinny TL, Mahoney MM, et al. temperature and locomotor activity in the antelope Nocturnal and diurnal rhythms in the unstriped Nile ground squirrel. J Therm Biol 2018; 72: 67–72. rat, Arvicanthis niloticus. J Biol Rhythms 1999; 14: 63. Soler JE, Robison AJ, Nunez AA, et al. Light 364–377. modulates hippocampal function and spatial learning 51. Schwartz MD and Smale L. Individual differences in in a diurnal rodent species: a study using male rhythms of behavioral sleep and its neural substrates in Nile grass rat (Arvicanthis niloticus). Hippocampus 2018; Nile grass rats. J Biol Rhythms 2005; 20: 526–537. 28: 189–200. 52. Refinetti R. Parameters of photic resetting of the circa- 64. Czeisler CA, Duffy JF, Shanahan TL, et al. Stability, dian system of a diurnal rodent, the Nile grass rat. Acta precision, and near-24-hour period of the human circa- Sci Vet 2004; 32: 1–6. dian pacemaker. Science 1999; 284: 2177–2181. 53. Refinetti R. Daily activity patterns of a nocturnal and a 65. Yartsev MM. The emperor’s new wardrobe: rebalancing diurnal rodent in a seminatural environment. Physiol diversity of animal models in neuroscience research. Behav 2004; 82: 285–294. Science 2017; 358: 466–469. 54. Redlin U and Mrosovsky N. Nocturnal activity in a diur- 66. Stuermer IW, Plotz K, Leybold A, et al. Intraspecific nal rodent (Arvicanthis niloticus): the importance of allometric comparison of laboratory gerbils with masking. J Biol Rhythms 2004; 19: 58–67. Mongolian gerbils trapped in the wild indicates domesti- 55. Karasov WH. Daily energy expenditure and the cost of cation in Meriones unguiculatus (Milne-Edwards, 1867) activity in a free-living mammal. Oecologia 1981; 51: (Rodentia: Gerbillinae). Zool Anz 2003; 242: 249–266. 253–259. 67. Moons CP, Breugelmans S, Cassiman N, et al. The effect 56. Swade RH and Pittendrigh CS. Circadian locomotor of different working definitions on behavioral research rhythms of rodents in the Arctic. Am Nat 1967; 101: involving stereotypies in Mongolian gerbils (Meriones 431–466. unguiculatus). J Am Assoc Lab Anim Sci 2012; 51: 57. DeCoursey PJ. Free-running rhythms and patterns of cir- 170–176. cadian entrainment in three species of diurnal rodents. 68. Waiblinger E and Ko¨nig B. Refinement of gerbil housing J Interd Cycle Res 1973; 4: 67–77. and husbandry in the laboratory. Altern Lab Anim 2004; 58. Kenagy GJ. Seasonality of endogenous circadian 32(Suppl 1A): 163–169. rhythms in a diurnal rodent Ammospermophilus leucurus 69. Paxinos G and Franklin K. The mouse brain in stereotaxic and a nocturnal rodent Dipodomys merriami. J Comp coordinates, 4th ed. New York: Academic Press, 2012. Physiol A 1978; 128: 21–36. 70. Paxinos G and Watson C. The rat brain in stereotaxic 59. Pohl H. Light pulses entrain the circadian activity rhythm coordinates, 6th ed. New York: Academic Press, 2007. of a diurnal rodent (Ammospermophilus leucurus). Comp 71. Mouse Genome Sequencing Consortium. Initial sequen- Biochem Physiol B 1983; 76: 723–729. cing and comparative analysis of the mouse genome. 60. DeCoursey PJ, Krulas JR, Mele G, et al. Circadian per- Nature 2002; 420: 520–562. formance of suprachiasmatic nuclei (SCN)-lesioned ante- 72. Rat Genome Sequencing Project Consortium. Genome lope ground squirrels in a desert enclosure. Physiol Behav sequence of the Brown Norway rat yields insights into 1997; 62: 1099–1108. mammalian evolution. Nature 2004; 428: 493–521. 61. Pohl H. Temperature cycles as zeitgeber for the circadian 73. Ledford H and Callaway E. Circadian clocks scoop clock of two burrowing rodents, the normothermic Nobel prize. Nature 2017; 550: 18.

Re´sume´ Bien que les souches domestique´es de ligne´es de rats et de souris traditionnelles servent de mode`les d’animaux mammife`res dans la recherche biome´dicale, les habitudes nocturnes de ces rongeurs les rendent inapproprie´s pour la recherche ne´cessitant un mode`le ayant des rythmes d’activite´ diurne. Nous avons effectue´ une revue de la litte´rature et enregistre´ les donne´es d’activite´ locomotrice provenant de quatre espe`ces de rongeurs ge´ne´ralement conside´re´s comme actifs au cours de la journe´e, a` savoir la gerbille de Mongolie (Meriones unguiculatus), le de`gue du Chili (Octodon degus), le rat du Nil (Arvicanthis niloticus), et l’e´cureuil-antilope (Ammospermophilus leucurus). Nos donne´es recueillies pendant 12 L:12O cycles lumie`re- obscurite´ ont confirme´ et e´largi la litte´rature existante en montrant que les rythmes d’activite´ des e´cureuils- antilopes et des rats du Nil sont plus fortement exprime´s et plus concentre´s dans la phase lumineuse du cycle lumie`re-obscurite´ que les rythmes d’activite´ des gerbilles de Mongolie et du de`gue du Chili, rendant les deux premie`res espe`ces pre´fe´rables et plus fiables comme mode`les d’activite´ diurne constante en labora- toire. Parmi les deux espe`ces plus fortement diurnes, les e´cureuil-antilopes sont plus exclusivement diurnes et pre´sentent des rythmes d’activite´ plus solides que les rats du Nil. Bien que les animaux de ces deux espe`ces ne soient pas actuellement disponibles aupre`s de fournisseurs commerciaux, les rats du Nil sont Refinetti and Kenagy 587 indige`nes dans une vaste zone du nord de l’Afrique et sont donc disponibles pour les chercheurs de l’he´mi- sphe`re orientale, tandis que les e´cureuils-antilopes peuvent eˆtre trouve´s dans une grande partie des de´serts d’Ame´rique du Nord et sont par conse´quent plus facilement accessibles aux chercheurs nord-ame´ricains.

Abstract Domestizierte Inzucht-Ratten- und Ma¨usesta¨mme dienen in der biomedizinischen Forschung zwar traditio- nell als Sa¨ugetier-Modelle, aufgrund der na¨chtlichen Gewohnheiten sind diese Nagetiere jedoch fu¨r die Forschung ungeeignet, wenn es um Modelle mit menschena¨hnlichen Tagesaktivita¨tsrhythmen geht. Wir fu¨hrten eine Literaturrecherche durch und erfassten lokomotorische Aktivita¨tsdaten von vier Nagetierarten, die allgemein als tagaktiv betrachtet sind: die Mongolische Rennratte (Meriones unguiculatus), der Degu (Octodon degus), die Afrikanische (Nil-) Grasratte (Arvicanthis niloticus) und das Antilopenho¨rnchen (Ammospermophilus leucurus). Unsere unter 12L:12D Licht-Dunkel-Zyklen gesammelten Daten besta¨tigen und vertiefen die vorhandene Literatur insofern, dass sie zeigen, dass die Aktivita¨tsrhythmen von Antilopenho¨rnchen und Afrikanischen Grasratten in der Licht-Phase des Licht-Dunkel-Zyklus sta¨rker und konzentrierter sind als die Aktivita¨tsrhythmen von Mongolischen Rennratten und Degus, weshalb die beiden erstgenannten Arten als Modelle fu¨r eine konsistente Tagesaktivita¨t im Labor die zu bevorzugenden und zuverla¨ssigeren Alternativen sind. Unter den beiden sta¨rker tagaktiven Arten sind Antilopenho¨rnchen ausschließlich tagaktiv und haben einen robusteren Aktivita¨tsrhythmus als Afrikanische Grasratten. Tiere dieser beiden Arten sind derzeit zwar nicht kommerziell erha¨ltlich, doch Afrikanische Grasratten sind in weiten Teilen Nordafrikas beheimatet und somit fu¨r Forscher in der o¨stlichen Hemispha¨re verfu¨gbar, wa¨hrend Antilopenho¨rnchen stark im Westen der nordamerikanischen Wu¨sten verbreitet und daher fu¨r nordamerikanische Forscher relativ problemlos zuga¨nglich sind.

Resumen A pesar de que distintas cepas de ratas y ratones domesticados y consanguı´neos sirven como modelos tradicionales de animales mamı´feros en la investigacio´n biome´dica, los ha´bitos nocturnos de estos roedores hacen que no sean apropriados para la investigacio´n que requiera un modelo con ritmos de actividad diurnos parecidos a los de los humanos. Llevamos a cabo una revisio´n documental y registramos datos de actividades locomotores de cuatro especies de roedores distintas que por lo general se consideran como activas durante el dı´a, el jerbo de Mongolia (Meriones unguiculatus), el degu (Octodon degus), la rata de hierba africana (Arvicanthis niloticus) y la ardilla de tierra Juancito (Ammospermophilus leucurus). Nuestros datos recopi- laron por debajo de 12 L:12O ciclos de luz-oscuridad adema´s de confirmar y ampliar los datos existentes que muestran que los ritmos de actividad de las ardillas de tierra Juancitos y las ratas de hierba africanas son ma´s robustas y se concentran ma´s en la fase de luz del ciclo luz-oscuridad que los ritmos de actividad de los jerbos de Mongolia y los degus, haciendo que las dos primeras especies sean ma´s preferibles y ma´s fiables como modelos de actividad diurna consistente en el laboratorio. Entre las dos especies diurnas ma´sso´lidas, las ardillas de tierra Juancitos son exclusivamente ma´s diurnas y tienen ritmos de actividad ma´s robustos que las ratas de hierba africanas. Aunque los animales de estas dos especies no esta´n actualmente disponibles a trave´s de proveedores comerciales, las ratas de hierba africanas son indı´genas en una amplia zona del norte de A´frica y por tanto esta´n disponibles para investigadores en el hemisferio occidental, mientras que las ardillas de tierra Juancitos pueden encontrarse en casi todo el desierto occidental de Ame´rica del norte y, por tanto, son ma´sfa´cilmente accesibles para investigadores norteamericanos. Original Article Laboratory Animals 2018, Vol. 52(6) 588–598 ! The Author(s) 2018 Anaesthetic effects of alfaxalone Article reuse guidelines: sagepub.com/journals- administered intraperitoneally alone permissions DOI: 10.1177/0023677218764214 or combined with dexmedetomidine journals.sagepub.com/home/lan and fentanyl in the rat

Mario Arenillas and Ignacio A Gomez de Segura

Abstract Alfaxalone is a neuroactive steroid used as a general anaesthetic in several species including dogs, cats, rabbits and ferrets. It has a wide margin of safety and a similar anaesthetic profile to propofol. To increase its aqueous solubility, a new formulation with cyclodextrins has been marketed recently. The objective of this study was to evaluate the anaesthetic effect of several doses of alfaxalone alone, considering differences between sexes, and alfaxalone combined with dexmedetomidine and fentanyl in the rat administered by the intraperitoneal route. A total of 40 Sprague Dawley rats, involved in three studies, were used. Firstly, 25, 35 and 45 mg kg–1 of alfaxalone alone were tested. In a second study, alfaxalone (25 mg kg–1, females; 75 mg kg–1, males) was combined with dexmedetomidine (0.05 mg kg–1). Finally, alfaxalone (20 mg kg–1, females; 60 mg kg–1, males) was combined with dexmedetomidine (0.05 mg kg–1) and fentanyl (0.1 mg kg–1). Times of onset and duration of anaesthesia, and analgesia, deemed as losing of withdrawal pedal reflex, were recorded. Alfaxalone alone produced a 2 - to 3-fold longer time of anaesthesia in females, although surgical anaesthesia was not achieved in either sex. The addition of dexmedetomidine and fentanyl to alfaxalone pro- duced a similar time of analgesia as well as increased time of anaesthesia in both sexes. In conclusion, alfaxalone produces light anaesthesia in rats, and males required a higher dose. The combination with other sedatives or analgesics, such as dexmedetomidine or fentanyl, allows a more prolonged anaesthesia with analgesic effects, potentially suitable for invasive procedures.

Keywords alfaxalone, dexmedetomidine, fentanyl, intraperitoneal, anaesthesia, rat

Date received: 7 September 2017; accepted: 14 February 2018

Although inhalational anaesthesia has been increas- sedatives. Unfortunately, these procedures produce ingly employed in rodents, injectable anaesthetics prolonged sedation, and ketamine-based combinations should also be considered since they may provide may cause a relatively prolonged recovery.1 Another advantages over the former technique. These may short acting anaesthetic drug, the GABAA agonist pro- include avoidance of (1) lower cardiovascular depres- pofol, has been proposed as an alternative to ketamine- sion derived from high doses of inhalational anaes- based anaesthetic combinations. However, this drug thetics, (2) safety concerns due to potential environmental pollution, and (3) limitations in the use Department of Animal Medicine and Surgery, Complutense of bulky equipment. A suitable injectable anaesthetic University of Madrid, Spain combination should involve short-acting anaesthetic drugs with a wide safety margin and that are, prefer- Corresponding author: ably, reversible. Most commonly, the use of injectable Mario Arenillas, Department of Animal Medicine and Surgery, Anaesthesiology Service, Veterinary Clinical Teaching Hospital, anaesthetic procedures for rodents is based on combin- Veterinary Faculty, Complutense University of Madrid, Avda. ations of dissociative drugs such as ketamine with Puerta de Hierro s/n, 28040, Madrid, Spain. alpha-2 adrenergic agonists and/or tranquilizers or Email: [email protected] Arenillas and Gomez de Segura 589 does not provide surgical anaesthesia when adminis- bedding Ultrasorb, Panlab, Cornella, Spain) and were tered alone intraperitoneally (IP) and has been com- kept in a 12 h:12 h light:dark cycle at room temperature bined with the alpha-2 adrenergic receptor agonist between 21–24C. Food (A04 SAFE, standard diet for medetomidine and the opioid fentanyl in rodents to rodents, Panlab, Cornella, Spain), and water were pro- attain such an effect.1–3 vided ad libitum. All animals were considered healthy Alfaxalone is a neuroactive steroid with anaesthetic based on veterinary physical examination, with no his- properties and a wide safety margin, which modulates tory of health abnormalities, and were not handled the inhibitory GABAA receptors.4 This anaesthetic has before their inclusion in the study except for routine been reformulated solubilised with 2-hydroxypropyl-b- cage cleaning maintenance. All the animals were cyclodextrin (Alfaxan). The anaesthetic effects of alfax- used according to European and national regulations alone in the former commercial presentation, contain- and the study was approved by the Ethics Committee ing polyethoxylated castor oil (Cremaphor) (CEEA 001 b/14; Getafe University Hospital, Madrid, as surfactant, have been compared with the new formu- Spain). lation of alfaxalone, confirming that both presentations retain the same anaesthetic profile. The allergic sensi- Drugs tivity to the former formulation limited its use to vet- erinary anaesthesia.5 Alfaxalone (Alfaxan 10 mg mL–1,Ve´toquinol, Lure, Alfaxan has been licensed for use in cats and France); Dexmedetomidine (Dexdomitor 0.1 mg mL–1, dogs and reported recently in rats, where relevant dif- Esteve, Barcelona, Spain), Fentanyl (Fentanest ferences between the sexes have been observed.6–8 0.05 mg mL–1, Kern Pharma, Terrasa, Spain). These differences were observed previously with the Dilutions were made with saline 0.9 % (Grifols, former formulation of alfaxalone and alphadalone.9 Parets del Valle` s, Spain). The new formulation has also been employed in rabbits and ferrets, producing light anaesthesia lasting some Study design 20–30 min, depending on the dose employed, and fast recovery with minimal or limited cardiovascular The study was designed in a stepped fashion (Table 1). effects.10, 11 However, a lack of analgesia was observed, First, a suitable dose range of Alfaxan administered and analgesic support should be provided in painful alone to both sexes was characterised (Study 1). procedures.12 In addition, alfaxalone produces hypo- Then, in a second study (2), the Alfaxan dose selected ventilation and oxygen support may be advisable. previously from study 1 for each sex was combined with The anaesthetic profile of alfaxalone resembles that dexmedetomidine. In a final study (3), fentanyl was of propofol, and its effects may be potentiated when combined with Alfaxan and dexmedetomidine in both combined with sedatives and analgesics.12 We hypoth- sexes. As an endpoint, anaesthesia duration of between esised that this anaesthetic may produce dose-related 45 and 60 min was considered clinically adequate and, anaesthesia when administered IP to rats and may be to this end, the doses of Alfaxan in the two latter stu- potentiated by the coadministration of sedatives and dies were reduced as other drugs were combined. analgesics. The aim of this study was to assess the anaesthetic and analgesic properties of Alfaxan admin- istered alone to rats IP, and to determine a suitable Table 1. dose range in this species. In addition, differences in Experimental design of the three consecutive studies indicating the doses of Alfaxan, dexmedetomidine Alfaxan anaesthesia between the sexes were evaluated. and fentanyl administered intraperitoneally to male and Finally, the combined effects of Alfaxan with dexmede- female Sprague-Dawley rats. The n value within each study tomidine and fentanyl IP were characterised. is shown in parenthesis.

Drugs Dose (mg kg–1)

Materials and methods Study 1 (n ¼ 8; 4 females, 4 males) Animals Alfaxalone 25, 35, 45 Study 2 (n ¼ 16; 8 females, 8 males) Forty Sprague-Dawley rats from our institutional Alfaxalone 25 (females), 75 (males) breeding colony (originally purchased from Harlan Dexmedetomidine 0.05 Laboratories, San Pietro Al Natisone, Udine, Italy), Study 3 (n ¼ 16; 8 females, 8 males) weighing 230–300 g (females) and 300–400 g (males), Alfaxalone 20 (females), 60 (males) aged between 6 and 8 months old (both sexes), were Dexmedetomidine 0.05 housed in groups of three to four (Models 2154F and Fentanyl 0.1 1291H, Type III H, Techniplast, Buguggiate, Italy; 590 Laboratory Animals 52(6)

Randomisation was performed within each study, defined as the time of loss of pedal withdrawal reflex with the same number of male and female rats being (lack of movement), assessed by digital pinching of the used with a random generated table (Excel, Microsoft). toes with the nail; and Duration of Anaesthesia, defined Randomisation was blocked by sex, variables were rec- as the total time of loss of the righting reflex. orded in a non-blinded fashion, and the experimental unit was the individual rat. Rats were anaesthetised Anaesthesia between 10.00 AM and 8.00 PM. No previous acclimatisation to the procedure was Determination of a suitable dose range of performed, and rats were weighed and individually Alfaxan. A crossover study (randomised block design) transferred to a single cage with bedding material just was performed administering three doses of Alfaxan IP before the drugs were administered. Volumes of (25, 35 and 45 mg kg–1) to eight rats (four females and Alfaxan alone or combined with dexmedetomidine or four males), at least at 1-week intervals. The lowest dexmedetomidine and fentanyl were calculated accord- dose was based on a previous study.4 Although relevant ing to the doses, freshly prepared and adjusted to equal differences between the sexes were found, the results volumes within each study with sterile saline, and were considered consistent enough (duration of anaes- administered IP into the right caudal quadrant of the thesia) to choose a suitable dose of Alfaxan to be com- ventral abdomen by an experienced researcher (MA). bined with dexmedetomidine in Study 2. A cloth was employed to reduce the stress on handling. After injection, the rat was monitored closely in the Combination of Alfaxan with dexmedetomidine. Since cage until loss of the righting reflex was achieved. differences in the response between sexes were Then, it was placed on a warming blanket (Aposan, determined, a dose of Alfaxan of 25 mg kg–1 was Cofares, Madrid, Spain) to maintain body temperature selected for females (n ¼ 8) and 75 mg kg–1 for males between 37.0C and 38.5C, and a protective ointment (n ¼ 8), and were combined with dexmedetomidine (Lubrithal 10 g, Dechra Veterinary products, LLC, 0.05 mg kg–1 and administered IP in random order. Northwich, UK) was applied to the eyes. The pedal A total of 16 rats were used. withdrawal reflex was tested every 5 min alternating between limbs. Combination of Alfaxan with dexmedetomidine and The noxious stimulus was performed always by fentanyl. Alfaxan (20 and 60 mg kg–1, females (n ¼ 8) the same researcher (MA) by pinching the web of the and males (n ¼ 8), respectively), dexmedetomidine skin between the toes with the finger nail. Movement 0.05 mg kg–1 and fentanyl 0.1 mg kg–1 were combined of the hind limb was considered as a positive and administered IP in random order. A total of 16 rats response and the lack of response as negative. Once were used. the rat recovered the righting reflex, it was placed iso- lated from social group in a clean cage in a quiet and Monitoring warm area (24–28 C) and closely monitored until total recovery approximately 1 h later. The animal was trans- Once animals were induced to anaesthesia, a pulse ferred to its former cage with the social group 12 h later. oximeter probe (UT100V Veterinary Handheld Pulse Oximeter, Utech Chongqing, China) placed in a hind Statistical analysis limb was used for the continuous monitoring of the heart rate (HR) and haemoglobin oxygen saturation A hypothesis of differences between doses and sexes (SpO2), and recorded every 5 min. Observation of thor- was considered in Study 1, whereas no differences acic wall movement was employed to monitor and were considered between sexes in Studies 2 and 3 by record respiratory rate (fR) every 5 min, using a 20-s adjusting the dose. Data are expressed as mean SD. interval to count respiratory movements. A digital rectal thermometer (Henry Schein Inc., Melville, NY, Power calculation. Duration of anaesthesia was USA) was used to record body temperature (BT) every considered the principal variable for power calculation 10 min. Rats were allowed to breathe room air in study in Study 1. A power of 0.8 with an a value of 0.05, 1. However, in studies 2 and 3, 100% oxygen an effect size of 10–20 min of duration of anaesthesia, (0.5 L/min) was delivered via a face mask (Rodent and SD of 9.4 min (obtained from the first three rats, Circuit Set, Harvard Apparatus, Holliston, MA, 2-tailed), gave a sample size of eight rats per group USA) (see Study design, below). (three groups: 25, 35 and 45 mg kg–1), with both The following times were also recorded: Time to sexes balanced, in Study 1 (Experimental Design Induction, defined as the time taken from injection to Assistant, NC3Rs; https://eda.nc3rs.org.uk/eda). Rats the loss of the righting reflex; Duration of Analgesia, from this study were included in a crossover design, Arenillas and Gomez de Segura 591 Results and every animal received three doses of Alfaxan in a random fashion. In Studies 2 and 3, the duration All animals recovered fully from anaesthesia and were of analgesia was considered as the main variable included in the data analysis. with an effect size of 5–10 min and SD of 5.5 min (2-tailed) suggesting a sample size of eight rats per Study 1: Alfaxan group (two groups: eight females and eight males). Onset of anaesthesia was shorter in females compared Group comparison. To analyse data, the normality to males (RM ANOVA, all doses, p ¼ 0.018) with dif- of each group sample was ascertained with the ferences between drug doses in females (RM ANOVA, Shapiro-Wilk test. In Study 1, the repeated measures p ¼ 0.029) but not in males (RM ANOVA, p ¼ 0.661; (RM) ANOVA for repeated data was employed Table 2). Alfaxan produced a differential effect on the for monitored and recorded data during anaesthesia duration of anaesthesia between sexes (RM ANOVA, –5 (HR, fR, SpO2 and BT). Also, this RM test was all doses, p ¼ 7.8 10 , Figure 1) with differences employed to compare studied times (onset of anaesthe- between drug doses (RM ANOVA, males p ¼ 0.028, sia, duration of analgesia, duration of anaesthesia) females p ¼ 0.001). A dose-dependent effect in the where the dose was considered the repeated factor duration of anaesthesia was observed in males between within the same individual, and sex the second factor. the lowest and the two higher doses. Nevertheless, no Student’s t-test was employed to determine differences statistical differences were found between doses when between sexes when the same dose was tested. females were considered, although the most prolonged In Studies 2 and 3, Student’s t-test was employed effects were observed with the two higher doses. No for single values (weight and times) and the RM analgesic effect, determined by the response to the nox- ANOVA (HR, fR, SpO2 and BT). The Bonferroni ious stimulus, was observed with any dose in any post hoc test was employed to compare doses and animal from either sex. times when appropriate. A p value < 0.05 was con- No differences in HR, fR, SpO2 and BT were sidered as significant. observed between the studied doses. SpO2 values

Table 2. Onset of anaesthesia, defined as the time taken from injection to the loss of the righting reflex; Duration of Analgesia, defined as the time of loss of pedal withdrawal reflex (lack of movement), assessed by digital pinching of the toes with the nail; and Duration of Anaesthesia, defined as the total time of loss of the righting reflex, in rats anaesthetised with three different doses of Alfaxan administered intraperitoneally (IP) (Study 1), Alfaxan and dexmedetomidine IP (Study 2) and Alfaxan, dexmedetomidine and fentanyl IP (Study 3).

Times (min)

Onset of Duration of Duration of Dose (mg kg–1) Sex n value anaesthesia anaesthesia analgesia

Study 1. Alfaxalone 25 Females 4 4 10 73 10 – Males 4 6 17 4* – 35 Females 4 3 183 9– Males 4 5 2* 39 13* – 45 Females 4 3 1 117 14 – Males 4 5 240 24* – Study 2. Alfaxalone (below) þ dexmedetomidine (0.05 mg kg–1) 25 Females 8 6 2 117 13 36 4 75 Males 8 5 1 133 6* 24 6 Study 3. Alfaxalone (below) þ Dexmedetomidine (0.05 mg kg–1) þ Fentanyl (0.1 mg kg–1) 20 Females 8 6 1 116 850 16 60 Males 8 4 1 140 15* 51 14

* Indicates differences between sexes (Student’s t-test; see text for p-values) 592 Laboratory Animals 52(6)

51 14 min in females and males, respectively; p ¼ 0.831). 45 No differences in HR were observed between both sexes (RM ANOVA, p ¼ 0.471) although a decrease )

-1 in HR was observed with time (RM ANOVA, p ¼ 8 10–7). The fR was lower in females (RM ANOVA, 35 p ¼ 0.002) decreasing with time (p ¼ 2 10–8). All rats * received oxygen via facemask and had a SpO2 > 95%. Recovery from all alfaxalone and alfaxalone-based Alfaxalone (mg · kg Alfaxalone (mg Female combinations was smooth and uneventful. 25 Male Discussion 0 20 40 60 80 100 120 140 Time (minutes) Alfaxan produced relevant differences in the duration of anaesthesia between males and females when given Figure 1. Duration of anaesthesia from male (n ¼ 4) and alone, resulting in a more prolonged effect in females female (n ¼ 4) rats anaesthetised with three doses of when the same doses were administered IP. Compared Alfaxan (25, 35 and 45 mg kg–1) administered intraperito- to females, a three-fold higher dose of Alfaxan was neally. Significant differences were found between both required in males to attain a similar duration of anaes- sexes (RM ANOVA; p < 0.05). * indicates significant differ- thesia. This effect was observed with three different ences compared with the lowest dose, same sex doses, none of which provided analgesic effects to pre- (Bonferroni test; p < 0.05). vent the response to a standard noxious stimulation. Although alfaxalone has no known antinociceptive below 95% were observed in both males and females properties, coadministration with dexmedetomidine with all Alfaxan doses (Table 3). HR was not modified increased anaesthetic duration and produced short- by Alfaxan during the study period with any dose. BT term analgesia, which increased further when fentanyl was maintained between 37.0C and 38.5C. was also administered. Sex-related differences in the anaesthetic effects of Study 2: Alfaxan þ Dexmedetomidine alfaxalone have already been described with the former formulation with a mixture of alfaxalone and Onset of anaesthesia was similar and below 7 min alphadalone solubilised in castor oil (Althesin).9 with both sexes (t-test, p ¼ 0.378). Duration of anaes- Almost a four-fold IP higher dose was required in thesia was shorter in females (117 13 min) than males males compared to females. However, such a difference (133 6 min; t-test, p ¼ 0.006), whereas the duration of was not observed when this drug was administered by analgesia was shorter in males (24 6 min) compared the IV route.9, 13 Similarly, we found a three-fold dose to females (36 4 min; t-test, p ¼ 0.0003, Figure 2). difference between sexes with the new cyclodextrin- HR decreased over time (RM ANOVA, p ¼ 4 10–7) based formulation of alfaxalone to achieve a similar but there were no differences between sexes at the stu- effect. Differences in anaesthesia duration between the died doses (RM ANOVA, p ¼ 0.501). In addition, a former and the currently marketed formulation may reduction in fR over the time was observed with both involve the use of alphadalone in the former commer- sexes (RM ANOVA, p ¼ 1 10–5). Females presented cial presentation but also the different age and strain of lower fR over time compared to males (RM ANOVA, rats employed. Anaesthetic dose and duration of anaes- p ¼ 4 10–5) (Table 4). All rats received oxygen via thesia using the former formulation required higher facemask and had a SpO2 > 95%. BT was maintained doses of alfaxalone in female rats under 30 days old, between 37.0C and 38.5C. although no further differences were observed in rats aged above 30 days.9 However, in a study using the cyclodextrin-formulated alfaxalone in young (7–10 Study 3: Alfaxan þ Dexmedetomidine þ –1 Fentanyl weeks) Wistar female rats, a dose of 20 mg kg IP produced a shorter and highly variable duration of Onset of anaesthesia was more prolonged in females anaesthesia (42 min) compared to a slightly higher compared to males (6 1 and 4 1 min, respectively; dose of 25 mg kg–1 (73 min) employed in the present t-test, p ¼ 0.004), whereas the duration of anaesthesia study.4 The strain and age of the rats, but also absorp- was shorter (116 8 and 140 15 min, respectively; tion from the site of IP injection may account for the p ¼ 0.002). There were no differences in the duration difference, suggesting a relatively high variability in the of analgesia between both sexes (50 16 and effects of alfaxalone when administered alone.14 Table 3. Heart rate (HR) in beats per minute (bpm), respiratory rate (fR) in breaths per minute and haemoglobin oxygen saturation (SpO2) in %, from female and male Segura de Gomez and Arenillas rats anaesthetised with Alfaxan intraperitoneally at 25, 35 and 45 mg kg–1 (n value is always 4 unless otherwise indicated; see Materials and Methods for details).

Study 1. Alfaxalone

Dose Time (IP, mg kg–1) (min) Onset 5 10 15 20 25 30 45 60 75 90 105 120

HR 25 Females 408 42 394 29 385 40 377 42 372 41 364 44 355 39 350 39 351 38 340 18 (n ¼ 2) Males 363 21 350 14 339 (n ¼ 2) (n ¼ 2) (n ¼ 1) 35 Females 372 35 365 41 355 38 355 35 354 30 352 32 351 31 339 30 343 34 361 11 (n ¼ 3) Males 371 9 360 10 347 16 350 23 348 27 348 30 340 25 316 (n ¼ 3) (n ¼ 1) 45 Females 386 11 372 11 379 19 379 19 374 17 367 15 365 12 351 7 350 10 359 4 362 9 365 9 375 (n ¼ 3) (n ¼ 3) (n ¼ 1) Males 343 24 333 25 328 19 329 20 323 13 340 6 344 9 334 1 336 (n ¼ 2) (n ¼ 2) (n ¼ 2) (n ¼ 1)

SpO2 25 Females 95 294 292 293 393 293 292 192 191 192 1 (n ¼ 2) Males 89 790 9 93 (n ¼ 3) (n ¼ 1) 35 Females 91 389 492 390 391 294 590 591 390 290 4 (n ¼ 3) Males 92 392 392 191 292 292 192 2 92 (n ¼ 3) (n ¼ 1) 45 Females 91 294 495 394 394 391 191 291 292 2923 29 2 93 2 93 (n ¼ 3) (n ¼ 3) (n ¼ 1) Males 93 193 293 292 293 293 1 90 3 91 2 92 (n ¼ 2) (n ¼ 2) (n ¼ 2) (n ¼ 1) (fR) 25 Females 75 470 570 569 869 770 571 476 10 70 568 0 (n ¼ 2) Males 105 19 95 22 88 (n ¼ 3) (n ¼ 1) 35 Females 77 369 468 469 769 769 572 673 881 17 81 16 (n ¼ 3) Males 83 683 985 799 16 99 16 101 16 100 20 92 (n ¼ 3) (n ¼ 1) 45 Females 70 464 661 661 562 463 363 466 568 581 26 82 26 83 32 54 (n ¼ 3) (n ¼ 1) 593 Males 88 12 83 681 2 89 5 105 17 106 20 110 14 106 20 96 (n ¼ 3) (n ¼ 2) (n ¼ 2) (n ¼ 2) (n ¼ 1) 594 Laboratory Animals 52(6)

been reported previously. In mice, surgical anaesthesia (a) Alfaxalone + Dexmedetomidine Duraon of analgesia Duraon of anaesthesia has been determined with alfaxalone alone but also in combination with medetomidine and butorphanol, Female administered subcutaneously. Interestingly, surgical anaesthesia was not achieved with either anaesthetic regimen when given by the IP route.16 However, the definition of surgical anaesthesia was different from Male * * our study and while we evaluated the lack of pedal withdrawal reflex, other reflexes have also been con- 16 0 20 40 60 80 100 120 140 160 sidered. Withdrawal of the hind limb following toe pinching is a commonly employed method suggesting (b) Alfaxalone + Dexmedetomidine + Fentanyl surgical anaesthesia in rodents. Reduced, non-surgical, anaesthetic effects of alfaxalone may result from Female reduced drug availability due to a first pass elimination by hepatic metabolism when administered IP. In cats and rabbits, the intramuscular administration of Alfaxan alone produced moderate-to-deep sedation Male * whilst anaesthesia was achieved only in the former spe- cies when medetomidine was co-administered.11,17,18 In rats, both analgesic and anaesthetic effect potenti- 0 20 40 60 80 100 120 140 160 Time (minutes) ation was found when dexmedetomidine was combined with equipotent alfaxalone doses (3:1, males:females). Although a longer duration of anaesthesia was Figure 2. Duration of anaesthesia, defined as the total time of loss of righting reflex, and duration of analgesia, observed in males, duration of analgesia was slightly defined as the time of loss of the pedal withdrawal reflex, shorter and fR was less reduced compared with females, from male and female rats anaesthetised with a) alfaxalone suggesting that the doses of drugs providing analgesia (25 and 75 mg kg–1, eight females and eight males, such as dexmedetomidine, would need to be increased. respectively) and dexmedetomidine (0.05 mg kg–1) intra- Females showed a reduced time of anaesthesia asso- peritoneally; or b) alfaxalone (20 and 60 mg kg–1, eight ciated with a longer analgesic action. Such a finding females and eight males, respectively), dexmedetomidine was unexpected and further studies should be con- –1 –1 (0.05 mg kg ) and fentanyl (0.1 mg kg ) intraperitoneally. ducted to determine whether it was an actual, rather * indicates significant differences compared to female rats than just statistical, finding. (Student’s t-test p < 0.05). Analgesic potentiation may be produced by admin- istering opioids such as fentanyl. The combination of Furthermore, differences in the pharmacokinetics of this opioid with alfaxalone and dexmedetomidine pro- intravenous cyclodextrin-formulated alfaxalone be- duced an increase in the duration of surgical anaesthe- tween sexes have been identified in Sprague Dawley sia, deemed as the loss of pedal withdrawal reflex. rats in plasma clearance, half-life and mean residence Again, an alfaxalone dose ratio of 3:1 (males:females) time when administered as a constant rate of infusion.8 provided equivalent timing of both analgesic and A relevant clinical sex difference in the effects of anaesthetic effects. A similar propofol-based combin- Alfaxan by the IP route in rats is a limitation for ation with medetomidine and fentanyl administered its use since no other anaesthetic drug employed in IP to rats demonstrated loss of the pedal withdrawal this species produces such an effect, and doses provided reflex during at least 30 min, when the alpha-2 adrener- to male and female rats are the same. Reasons gic receptor antagonist atipamezole was administered.3 for such differences have been attributed to a As expected, this combination provided better anaes- higher concentration of the endogenous GABA modu- thetic and analgesic effects compared to propofol given lator (3a-hydroxydihydroprogesterone) in females.5 alone. Similar results were obtained when the same Alfaxalone, as well as naturally occurring steroids, combination was used in mice.1 enhance GABA-mediated inhibition in the brain, sug- Anaesthetic drugs are usually associated with cardio- gesting a potentiation of their effects.15 vascular and respiratory depression closely related Regardless of the dose or depth of anaesthesia, with the drugs involved but also their interactions. Alfaxan did not provide any analgesic effect when These effects are a safety concern during anaesthesia. administered alone, thus its combination with analgesic Dexmedetomidine has potent alpha-2 adrenoceptor drugs would be required for painful procedures. In rats, activity characterised by vasoconstriction, leading alfaxalone-based anaesthetic combinations have not to an initial increase in blood pressure followed by Arenillas and Gomez de Segura 595 31 12 19,20 hypotension, and bradycardia. Not surprisingly, when combined with alfaxalone, dexmedetomidine pro- duced bradycardia. Besides, fentanyl, a potent mu 10 21 84 10 20 316 opioid receptor agonist, also produces vagally mediated bradycardia.21,22 Thus, both alfaxalone-dexmedetomi- dine and alfaxalone-dexmedetomidine-fentanyl com- 26 961 13 72 18 303 24 297 9 6 binations produce a marked decrease in HR. Ventilatory depression, determined by marked decreases in both fR and SpO , was observed when 21 297 21 764 12 52 12 299 24 294 2 649 17 24 alfaxalone was given alone, with the latter values sug- gesting mild to moderate hypoxemia within a normal 19 273 19 277 861 14 49 7 287 22 287 741 11 38 BT range. Oxygen supplementation was then employed with the two alfaxalone combinations, where a decrease in fR was also observed. A greater decrease in fR 17 276 17 271 955 14 46 10 276 24 279 935 10 33 was observed in females, which would suggest an increased depth of anaesthesia. However, the duration of analgesia was similar between sexes. The combin- 23 277 12 267 752 12 40 8 273 21 272 430 732 ation of medetomidine and fentanyl produces respira- tory depression in rats.23 Dexmedetomidine might induce relative minor ventilatory depression in rats, 17 278 16 265 846 937 7 265 20 269 728 626 suggesting that most of the depressant effects might )

–1 be due to fentanyl alone or its combined effects with dexmedetomidine.24,25 Therefore, it is advisable to pro- 12 280 19 266 844 935 8 262 23 267 830 523 vide oxygen to anaesthetised rats when these alfaxa- lone-based combinations are employed. This study has some limitations. Complete cardio- 14 277 18 269 745 734 12 261 22 268 927 621 vascular and respiratory monitoring was not con- Fentanyl (0.1 mg kg þ )

) sidered in the present study and only basic parameters –1 –1 were assessed. Therefore, the effects on relevant param- 15 265 18 271 842 936 10 261 20 271 527 420 eters such as blood pressure or capnography should be considered to better assess the true impact of alfaxalone and alfaxalone-based combinations. Besides, pedal 18 261 19 275 642 934 13 259 21 279 929 517 withdrawal reflex was considered as a surrogate for sur- gical stimulus. Therefore, these drugs and combinations should be tested in clinical conditions where surgery is 19 262 26 276 10 41 934 10 257 19 281 724 317 performed. Although the anaesthetic and analgesic dexmedetomidine (0.05 mg kg Dexmedetomidine (0.05 mg kg

þ þ variables were not blindly assessed and may have biased the results, the same researcher (MA) performed 21 262 25 276 13 39 13 31 8 256 20 281 727 11 16 all assessments thus reducing variability. Rats were in males) in males)

–1 –1 anaesthetised between 10.00 AM and 8.00 PM. A circa- 17 265 23 270 12 39 20 34 8 257 21 281 10 25 13 13 dian effect on anaesthetic potency is unlikely to have occurred since all experiments were performed during the rats’ rest phase.26 18 273 18 275 645 15 39 5 274 19 290 16 23 19 17 In conclusion, Alfaxan is a suitable drug for light anaesthesia in rats where analgesia is not required 57 57 onset 5 10 15 20 25 30and 40 produces 50 60 minor 75 adverse 90 effects. 105 120 However, 135 150 this in females and 75 mg kg in females and 60 mg kg 8) 290 8) 289 8) 35 8) 46 anaesthetic should be combined with other sedatives –1 –1 ¼ ¼ ¼ ¼ 8) 302 8) 314 8) 8) n n n n ¼ ¼ ¼ ¼ and opioids to provide surgical anaesthesia when n n n n R in female and male rats anaesthetised with Alfaxan combined with dexmedetomidine or dexmedetomidine and fentanyl. Doses administered IP to

f administered intraperitoneally. A relevant sex differ- ence in the dose required to produce anaesthetic effects Males ( Males ( Males ( Males ( was determined, where males required a dose three HR and times higher. This may be a practical limitation for its use. Finally, clinical studies should evaluate the feasi- bility of these anaesthetic combinations in surgical R (bpm) Females ( R (bpm) Females ( HR (bpm) Females ( HR (bpm) Females ( f f Table 4. females and males are indicated (seeTime Materials (min) and Methods for details). Study 2. Alfaxalone (25 mg kg Study 3. Alfaxalone (20 mg kg procedures. 596 Laboratory Animals 52(6)

Declaration of Conflicting Interests alfaxalone – a clinical trial. Vet Anaesth Analg 2014; 41: 378–385. The author(s) declared no potential conflicts of interest with 13. Child KJ, Currie JP, Davis B, et al. The pharmacological respect to the research, authorship, and/or publication of this properties in animals of CT1341 – a new steroid anaes- article. thetic agent. Br J Anaesth 1971; 43: 2–13. 14. Chapter 5 - Intraperitoneal Drug Administration. In: Funding Claassen V (ed) Techniques in the Behavioral and Neural The author(s) received no financial support for the research, Sciences. Elsevier, 1994, pp. 46–58. authorship, and/or publication of this article. 15. Turner DM, Ransom RW, Yang JS, et al. Steroid anes- thetics and naturally occurring analogs modulate the ORCID iD gamma-aminobutyric acid receptor complex at a site dis- Ignacio A Gomez de Segura http://orcid.org/0000-0001- tinct from barbiturates. J Pharmacol Exp Ther 1989; 248: 5436-5396 960–966. 16. Higuchi S, Yamada R, Hashimoto A, et al. Evaluation of References a combination of alfaxalone with medetomidine and butorphanol for inducing surgical anesthesia in labora- 1. Alves HC, Valentim AM, Olsson IA, et al. tory mice. Jpn J Vet Res 2016; 64: 131–139. Intraperitoneal anaesthesia with propofol, medetomidine 17. Rodrigo-Mocholi D, Belda E, Bosmans T, et al. Clinical and fentanyl in mice. Lab Anim 2009; 43: 27–33. efficacy and cardiorespiratory effects of intramuscular 2. Alves HC, Valentim AM, Olsson IA, et al. administration of alfaxalone alone or in combination Intraperitoneal propofol and propofol fentanyl, sufenta- with dexmedetomidine in cats. Vet Anaesth Analg 2016; nil and remifentanil combinations for mouse anaesthesia. Lab Anim 2007; 41: 329–336. 43: 291–300. 3. Alves HN, da Silva AL, Olsson IA, et al. Anesthesia with 18. Grubb TL, Greene SA and Perez TE. Cardiovascular and intraperitoneal propofol, medetomidine, and fentanyl in respiratory effects, and quality of anesthesia produced by rats. J Am Assoc Lab Anim Sci 2010; 49: 454–459. alfaxalone administered intramuscularly to cats sedated 4. Lau C, Ranasinghe MG, Shiels I, et al. Plasma pharma- with dexmedetomidine and hydromorphone. J Feline cokinetics of alfaxalone after a single intraperitoneal or Med Surg 2013; 15: 858–865. intravenous injection of Alfaxan((R)) in rats. J Vet 19. Virtanen R. Pharmacological profiles of medetomidine Pharmacol Ther 2013; 36: 516–520. and its antagonist, atipamezole. Acta Vet Scand Suppl 5. Estes KS, Brewster ME, Webb AI, et al. A non-surfac- 1989; 85: 29–37. tant formulation for alfaxalone based on an amorphous 20. Savola JM. Cardiovascular actions of medetomidine and cyclodextrin: activity studies in rats and dogs. Int J their reversal by atipamezole. Acta Vet Scand Suppl 1989; Pharm 1990; 65: 101–107. 85: 39–47. 6. Muir W, Lerche P, Wiese A, et al. The cardiorespiratory 21. Reitan JA, Stengert KB, Wymore ML, et al. Central and anesthetic effects of clinical and supraclinical doses vagal control of fentanyl-induced bradycardia during of alfaxalone in cats. Vet Anaesth Analg 2009; 36: 42–54. halothane anesthesia. Anesth Analg 1978; 57: 31–36. 7. Muir W, Lerche P, Wiese A, et al. Cardiorespiratory and 22. Laubie M, Schmitt H, Canellas J, et al. anesthetic effects of clinical and supraclinical doses of Centrally mediated bradycardia and hypotension induced alfaxalone in dogs. Vet Anaesth Analg 2008; 35: 451–462. by narcotic analgesics: dextromoramide and fentanyl. Eur 8. White KL, Paine S and Harris J. A clinical evaluation of J Pharmacol 1974; 28: 66–75. the pharmacokinetics and pharmacodynamics of intra- 23. Hu C, Flecknell PA and Liles JH. Fentanyl and medeto- venous alfaxalone in cyclodextrin in male and female midine anaesthesia in the rat and its reversal using atipa- rats following a loading dose and constant rate infusion. mazole and either nalbuphine or butorphanol. Lab Anim Vet Anaesth Analg 2017; 44: 865–875. 1992; 26: 15–22. 9. Fink G, Sarkar DK, Dow RC, et al. Sex difference in 24. Fernandes FC, Ferreira HC, Cagido VR, et al. Effects of response to alphaxalone anaesthesia may be oestrogen dexmedetomidine on respiratory mechanics and control dependent. Nature 1982; 298: 270–272. of breathing in normal rats. Respir Physiol Neurobiol 10. Giral M, Garcia-Olmo DC, Gomez-Juarez M, et al. 2006; 154: 342–350. Anaesthetic effects in the ferret of alfaxalone alone and 25. Bol CJ, Vogelaar JP and Mandema JW. Anesthetic pro- in combination with medetomidine or tramadol: a pilot file of dexmedetomidine identified by stimulus-response study. Lab Anim 2014; 48: 313–320. and continuous measurements in rats. J Pharmacol Exp 11. Huynh M, Poumeyrol S, Pignon C, et al. Intramuscular Ther 1999; 291: 153–160. administration of alfaxalone for sedation in rabbits. 26. Munson ES, Martucci RW and Smith RE. Circadian Vet Rec 2015; 176: 255. variations in anesthetic requirement and toxicity in rats. 12. Pinelas R, Alibhai HI, Mathis A, et al. Effects of different Anesthesiology 1970; 32: 507–514. doses of dexmedetomidine on anaesthetic induction with Arenillas and Gomez de Segura 597

Re´sume´ L’alfaxalone est un ste´roı¨de neuroactif utilise´ comme anesthe´sique ge´ne´ral chez plusieurs espe`ces, notam- ment chez les chiens, les chats, les lapins ou les furets. Elle est dote´e d’une grande marge de se´curite´ et d’un profil anesthe´sique similaire au propofol. Pour augmenter sa solubilite´ dans l’eau, une nouvelle formulation avec cyclodextrines a e´te´ re´cemment commercialise´e. L’objectif de cette e´tude e´tait d’e´valuer l’effet anesthe´- siant de plusieurs doses d’alfaxalone seule, compte tenu des diffe´rences entre les sexes, et de l’alfaxalone associe´ea` la dexme´de´tomidine et au fentanyl, par voie intrape´ritone´ale chez le rat. Un total de 40 rats Sprague Dawley, implique´s dans trois e´tudes, ont e´te´ utilise´s. Premie`rement, 25, 35 et 45 mg kg-1 d’alfaxalone seule ont e´te´ teste´s. Dans une seconde e´tude, l’alfaxalone (25 mg kg-1, femelles, 75 mg kg-1,maˆles) a e´te´ combine´ea` la dexme´de´tomidine (0.05 mg kg-1). Enfin, l’alfaxalone (20 mg kg-1, femelles, 60 mg kg-1,maˆles) a e´te´ associe´ea` la dexme´de´tomidine (0.05 mg kg-1) et au fentanyl (0.1 mg kg-1). Le de´lai de de´clenchement et la dure´e de l’anesthe´sie, ainsi que l’analge´sie, conside´re´e comme la perte du re´flexe podal, ont e´te´ enregistre´s. L’alfaxalone produit a` elle seule une dure´e d’anesthe´sie 2 a` 3 fois plus longue chez les femelles bien que l’anesthe´sie chirurgicale n’ait pas e´te´ atteinte dans aucun des deux sexes. L’ajout de la dexme´de´tomidine et du fentanyl a` l’alfaxalone a produit une dure´e similaire d’analge´sie ainsi qu’une augmentation de la dure´e d’anesthe´sie chez les deux sexes. En conclusion, l’anesthe´sie par alfaxalone produit une anesthe´sie le´ge`re chez les rats, bien qu’elle ne´cessite une dose plus e´leve´e chez les maˆles. Son association a` d’autres se´datifs ou analge´siques, tels que la dexme´de´tomidine ou le fentanyl, permet une plus longue dure´e d’anesthe´sie accompagne´e d’effets analge´siques, qui pourrait convenir pour les proce´dures invasives.

Abstract Alfaxalon ist ein neuroaktives Steroid, das zur Vollnarkose bei verschiedenen Arten wie Hunden, Katzen, Kaninchen oder Frettchen eingesetzt wird. Es hat einen großen Sicherheitsspielraum und ein a¨hnliches Narkoseprofil wie Propofol. Um seine Wasserlo¨slichkeit zu erho¨hen, wurde vor kurzem eine neue Formulierung mit Cyclodextrinen auf den Markt gebracht. Ziel der vorliegenden Studie war es, die beta¨ubende Wirkung von mehreren Dosen von Alfaxalon allein unter Beru¨cksichtigung der Unterschiede zwischen den Geschlechtern zu evaluieren und in Kombinierung mit Dexmedetomidin und Fentanyl bei der Ratte durch intraperitoneale Verabreichung. Insgesamt wurden 40 Sprague- Dawley-Ratten fu¨r drei Studien verwendet. Zuna¨chst wurden 25, 35 und 45 mg/kg Alfaxalon allein getestet. In einer zweiten Studie wurde Alfaxalon (25 mg/kg, weiblich; 75 mg/kg, ma¨nnlich) mit Dexmedetomidin (0.05 mg/kg) kombiniert. Schließlich wurde Alfaxalon (20 mg/kg, weiblich; 60 mg /kg, ma¨nnlich) mit Dexmedetomidin (0.05 mg/kg) und Fentanyl (0.1 mg/kg) kombiniert. Die Zeiten des Wirkungsbeginns und der Dauer von Ana¨sthesie und Analgesie, die fu¨r den Verlust des Ru¨ckzugsreflexes gelten, wurden aufgezeichnet. Alfaxalon allein fu¨hrte zu einer 2-3-mal la¨ngeren Narkosezeit bei weiblichen Ratten, wenngleich eine chirurgische Beta¨ubung bei keinem Geschlecht erreicht wurde. Der Zusatz von Dexmedetomidin und Fentanyl zu Alfaxalon ergab eine a¨hnliche Zeit fu¨r Analgesie sowie eine la¨ngere Zeit der Ana¨sthesie bei beiden Geschlechtern. Zusammenfassend la¨sst sich sagen, dass Alfaxalon bei Ratten eine leichte Beta¨ubung erzeugt, wenngleich ma¨nnliche Tiere eine ho¨here Dosis beno¨tigten. Die Kombination mit anderen Beta¨ubungsmitteln oder Analgetika, wie z. B. Dexmedetomidin oder Fentanyl, ermo¨glicht eine la¨ngere Ana¨sthesie mit schmerzstillender Wirkung, die mo¨glicherweise fu¨r invasive Verfahren geeignet ist.

Resumen La alfaxalona es un esteroide neuroactivo utilizado como aneste´sico general en varias especies como perros, gatos, conejos o hurones. Tiene un amplio margen de seguridad y un perfil aneste´sico similar al propofol. Para aumentar su solubilidad en agua, se ha comercializado recientemente una nueva fo´rmula con ciclodex- trinas. El objetivo de este estudio fue evaluar el efecto aneste´sico de varias dosis de alfaxalona, considerando las diferencias entre sexos, y en combinacio´n con dexmedetomidina y fentanilo en ratas tras su administra- cio´n intraperitoneal. Se utilizaron un total de 40 ratas Sprague Dawley en tres estudios. En primer lugar, se administraron 25, 35 y 45 mg kg-1 de alfaxalona. En un segundo estudio, se combino´ alfaxalona (25 mg kg-1, hembras; 75 mg kg-1, machos) y dexmedetomidina (0,05 mg kg-1). Finalmente, se combino´ alfaxalona (20 mg kg-1, hembras; 60 mg kg-1, machos), dexmedetomidina (0,05 mg kg-1) y fentanilo (0,1 mg kg-1). Se registraron los tiempos de inicio y duracio´n de anestesia y analgesia (considerada como una pe´rdida del reflejo de retirada podal). La alfaxalona produjo del doble al triple de duracio´n de anestesia en hembras, aunque la 598 Laboratory Animals 52(6) anestesia quiru´rgica no se consiguio´ ni en machos ni en hembras. La adicio´n de dexmedetomidina y fentanilo a la alfaxalona produjo un tiempo de anestesia parecido, adema´s de aumentar el tiempo en anestesia de ambos sexos. En conclusio´n, la alfaxalona produce una anestesia superficial en ratas aunque los machos requirieron una dosis mayor. La combinacio´n con otros sedantes o analge´sicos, como la dexmedetomidina o el fentanilo, permite una anestesia ma´s prolongada con efectos analge´sicos, potencialmente adecuada para procedimientos invasivos. Original Article Laboratory Animals 2018, Vol. 52(6) 599–610 ! The Author(s) 2018 Refinement and reduction through Article reuse guidelines: sagepub.com/journals- application of a quantitative score system permissions DOI: 10.1177/0023677218757815 for estimation of TB-induced disease journals.sagepub.com/home/lan burden using computed tomography

Sally A Sharpe1, Donna Smyth1, Anthony McIntyre2, Fergus Gleeson2 and Mike J Dennis1

Abstract Until validated correlates of protection are identified, animal models remain the only way to test the efficacy of the new vaccines and drugs urgently needed to fight the global epidemic caused by infection with Mycobacterium tuberculosis. Non-human primates (NHP) offer the most relevant models of human tuber- culosis (TB) and are central to the development process for new interventions. Efficacy evaluations are dependent on the capability of the test model to discriminate improved outcomes between treated groups after experimental exposure to M. tuberculosis and therefore the ability to measure TB-induced disease burden is central to the process. We have developed a score system that allows us to quantify the disease burden induced in macaques by infection with M. tuberculosis, based on the extent and features of disease visible on computed tomography (CT) images. The CT determined disease burden was then verified against that obtained using an established pathology-based approach. Trials of the system as a tool to measure disease burden have shown the approach capable of revealing differences between treatment groups in order to: (a) characterise outcome of infection and enable model refinement; (b) demonstrate the efficacy of drug treatment regimens by showing differences in outcome between test groups. Initial trials suggest that the imaging-based score system provides a valuable additional tool for the measurement of TB-induced disease burden that offers the opportunity to apply both refinement and reduction within studies.

Keywords computed tomography, tuberculosis, non-human primate

Date received: 29 September 2017; accepted: 15 January 2018

Introduction models of human TB because of the close similarities in their anatomy, physiology and immune systems,3–5 Mycobacterium tuberculosis, the causative agent of which lead to the development of comparable clinical tuberculosis (TB), is an intracellular pathogen that signs, disease pathology and immune features following infects 10 million people worldwide and kills approxi- M. tuberculosis infection. mately 1.4 million people every year.1 The resultant The ability to quantify the disease burden associated global health crisis has been intensified by emergence with M. tuberculosis infection in the animals under of drug-resistant M. tuberculosis strains, co-infection study is essential but these measures need to be accurate with HIV and the poor efficacy of the current vaccine, 2 BCG. New treatments including a new vaccine are 1National Infection Service, Public Health England, Porton Down, urgently needed and until validated correlates of pro- Salisbury, UK tection are identified, animal models remain fundamen- 2The Churchill Hospital, Headington, Oxford, UK tal to the development process, as they provide a Corresponding author: platform in which the efficacy of new interventions SA Sharpe, Public Health England, Porton Down, Salisbury, SP4 can be tested against infectious challenge. Non-human 0JG, UK. primates (NHP) offer the most clinically relevant Email: [email protected] 600 Laboratory Animals 52(6) and sensitive enough to distinguish subtle differences characterised breeding colonies managed by Public between treatment groups. Quantitative gross path- Health England that have been established in the UK ology score systems based on the number and size of for more than 30 years. Genetic analysis of these colo- lesions in pulmonary and extra-pulmonary tissues are nies has previously confirmed the rhesus macaques to established and widely used in the macaque models of be of the Indian genotype and cynomolgus macaques of TB. Their application, which has demonstrated vaccine Indonesian genotype.18 Absence of previous exposure efficacy,6–11 has also assisted model characterisation by to mycobacterial antigens was confirmed: by a tubercu- revealing the effect on disease burden caused by param- lin skin test conducted whilst the animals were still in eters such as disease status (active or latent) or study their original breeding colony; and just prior to study duration.11,12 Thus gross pathology scoring systems start, using an ex-vivo IFN-g ELISPOT (MabTech, provide valuable measures of total disease burden. Nacka, Sweden) to measure responses to PPD (SSI, However, as determination of the scores depends on Copenhagen, Denmark), and pooled 15-mer peptides information gathered during necropsy, the approach of ESAT6 and CFP10 (Peptide Protein Research Ltd, cannot be used to determine the in life evolution of Fareham, UK). disease within individual animals. Animals were housed in compatible social groups, in Serial images can be collected from live animals accordance with the Home Office (UK) Code of using medical imaging systems such as chest radio- Practice for the Housing and Care of Animals Bred, graphs (CXR), computed tomography (CT) and posi- Supplied or Used for Scientific Purposes, December tron emission tomography – computed tomography 2014, and the National Committee for Refinement, (PET-CT), all of which have been established for use Reduction and Replacement (NC3Rs), Guidelines on with the TB-macaque model and applied to visualise Primate Accommodation, Care and Use, August the development of pulmonary disease.13–15 The 2006. The housing environment was maintained approaches described to quantify TB-induced pulmon- within the temperature range 18–24C and a relative ary disease burden from the images collected vary humidity range 40–70%. Cages were constructed with according to the imaging modality. Qualitative score high level observation balconies and enrichment was systems have been defined for the evaluation of CXRs afforded by the provision of toys, swings, feeding puz- based on the occurrence, position and extent of pul- zles and DVDs for visual stimulation. Banks of cages monary infiltrate and have been proven capable of were placed in directional airflow containment systems demonstrating improved outcomes associated with vac- that allowed group housing and environmental control cination,6–11 although their sensitivity is limited because whilst providing a continuous, standardised inward of the relativity low resolution achievable with X-ray. flow of fully conditioned fresh air identical for all The more detailed images obtained using CT and groups. In addition to standard old world primate pel- PET-CT have enabled the development of quantitative lets (Primate Mazuri Expanded MP(E), Dietex measures of pulmonary disease based on the number of International, Witham, UK) further food was provided individual nodules,10,16 changes in lung attenuation,17 from a selection of vegetables and fruit that were varied and increase in fluorodeoxyglucose (FDG) avidity.16 on a day to day basis. Animals were sedated by intra- CT images also enable the detection and characterisa- muscular (IM) injection with ketamine hydrochloride tion of lymph nodes and extra-pulmonary tissues (Ketaset, 100 mg/ml, Fort Dodge Animal Health Ltd, including the spleen, liver and kidneys for the presence Southampton, UK; 10 mg/kg) for simple procedures of TB-induced disease. Therefore, CT images have the such as blood sampling that required removal from power to provide an overall representation of the extent their housing. None of the animals had been used pre- of TB disease within the host in a similar manner to viously for experimental procedures. In order to meet that provided by the established systems based on gross welfare requirements it is imperative that macaques are pathology changes. housed together in single gender, behaviourally harmo- The work described in this report aimed to assess the nious groups and therefore treatment groups were ran- validity and utility of a new scoring system applied to domly assigned to socially compatible cohorts using information obtained from CT images of TB-infected software generated random number allocations macaques for the quantification of disease burden. (Microsoft Excel) As gender has not been reported to influence the outcome of infection with M. tubercu- losis in macaques, males were selected for these studies. Materials and methods All animal procedures and study designs were approved Experimental animals by the Public Health England Animal Welfare and Ethical Review Body, Porton Down, UK, and Rhesus macaques and cynomolgus macaques, aged 3–4 authorised under an appropriate UK Home Office pro- years, were obtained for these studies from the ject licence. Sharpe et al. 601 M. tuberculosis challenge strains retained dose (the number of organisms assumed The M. tuberculosis strain with the characteristics most to be retained in the lung) have been described appropriate to meet the aims of each in vivo study was previously.10,11,23 selected for challenge. The Erdman strain of M. tuber- culosis is well characterised in the macaque model and Bronchoscopic instillation widely used in studies to assess the efficacy of new inter- ventions against TB. The M. tuberculosis Erdman K01 The bacterial inoculum for intrabronchial delivery was stock (BEI Resources, HPA-Sept 2011) was used for prepared by dilution of an aliquot from the same sus- challenge, as previously described.11 A stock suspension pension used for aerosol challenge into sterile PBS that was initially prepared from a 5 ml starter culture ori- provided a suspension whereby 2 ml contained approxi- ginally generated from colonies grown on Middlebrook mately 35 CFU, which was verified by retrospective 7H11 selective agar supplemented with oleic acid, albu- analysis.24 min, dextrose and catalase (OADC) (BioMerieux, UK). A liquid batch culture was then grown to logarithmic Clinical assessment phase in 7H9 medium (Sigma-Aldrich, UK) supple- mented with 0.05% (v/v) Tween 80 (Sigma-Aldrich, Animals were monitored daily for behavioural abnorm- UK). Aliquots were stored at 80C. Titre of the alities including depression, withdrawal from the stock suspension was determined from thawed aliquots group, aggression, and changes in feeding patterns, res- by enumeration of colony forming units (CFU) cul- piration rate and the occurrence of cough. Animals tured onto Middlebrook 7H11 OADC selective agar. were weighed, rectal temperature measured and exam- The H37Rv strain has previously been reported to be ined for gross abnormalities on each occasion that less virulent in macaques than the Erdman strain,19 and required blood sample collection, aerosol challenge or we have used this for specific studies where this feature euthanasia. Red blood cell (RBC) haemoglobin levels is beneficial. The M. tuberculosis H37Rv (National were measured using a HaemaCue haemoglobinometer Collection of Type Cultures (NCTC) 7416) challenge (Haemacue Ltd, Dronfield, UK) to identify the stock was generated from a chemostat grown to presence of anaemia, and erythrocyte sedimentation steady state under controlled conditions at 37C 0.1, rates (ESR) were measured using the Sediplast system pH 7.0 0.1 and a dissolved oxygen tension of (Guest Medical, Edenbridge, UK) to detect and 10% 0.1, in a defined medium, the details of which monitor inflammation induced by infection with have been previously described,20,21 High-concentra- M. tuberculosis. tion culture (approximately 2 108 CFU ml1 was ali- The time of necropsy, if prior to the end of the quoted and frozen at 80C. Titre of the stock planned study period, was determined by experienced suspension was determined from thawed aliquots primatology staff and based on a combination of the by enumeration of CFU cultured onto Middlebrook following adverse indicators: depression or withdrawn 7H11 supplemented with OADC selective agar behaviour, abnormal respiration (dyspnoea), loss of (BioMerieux, UK). Aliquots were stored at 80C. 20% of peak post-challenge weight, ESR levels elevated For challenge, sufficient vials of the selected strain above normal (>20 mm), haemoglobin level below were thawed and diluted appropriately, in sterile dis- normal limits (<100 g/dl), increased temperature tilled water. (>41C) and abnormal thoracic radiograph. The range of adverse indicators used allowed application Aerosol exposure of humane intervention when individuals had pro- gressed to moderate disease. Macaques were challenged by exposure to aerosols of 11,21 M. tuberculosis, as previously described. Mono- Drug treatment dispersed bacteria in particles were generated using a 3-jet Collison nebuliser and, in conjunction with a mod- Eight cynomolgus macaques infected with M. tubercu- ified Henderson apparatus,22 delivered to the nares of losis Erdman strain following exposure to aerosols esti- each sedated primate via a modified veterinary anaes- mated to contain presented doses of between 192 and thetic mask. Challenge was performed on sedated ani- 271 CFU were enrolled in a study to assess the efficacy mals placed within a ‘head-out’, plethysmography of two anti-TB therapy regimens. Drug treatment was chamber (Buxco, Wilmington, NC, USA) to enable initiated eight weeks after infection with M. tubercu- the aerosol to be delivered simultaneously with the losis. The four macaques in group A received eight measurement of respiration rate and respired volume. weeks of treatment with a combination of isoniazid The calculations to derive the presented dose (PD) (the (Sigma-Aldrich) dose 15 mg/kg, rifampicin (Sigma- number of organisms that the animals inhale) and the Aldrich)15 mg/kg and pyrazinamide (Sigma-Aldrich) 602 Laboratory Animals 52(6)

200 mg/kg, followed by eight weeks of treatment with judged to cause additional debilitating effects on the isoniazid (Sigma-Aldrich) dose 15 mg/kg and etham- host. The scores attributed to each tissue (lung lobe, butol (Sigma-Aldrich) 75 mg/kg. The same anti-TB organ, lymph node) were summed to give the total combination therapies were provided in the reverse CT score. order using the same schedule to the four macaques The images used to develop and test the CT score in group B. During each treatment block animals system were collected from M. tuberculosis infected received a daily oral dose of the appropriate drug com- macaques enrolled in several studies performed at bination formulated in a four ml volume of fruit puree. PHE Porton to establish new aerosol challenge CT scans were collected prior to the start of treatment models for the evaluation of new TB therapeutics. at week eight and at the end of each drug treatment These studies were aimed to evaluate, M. tuberculosis phase at weeks 16 and 24. dose, strain and delivery route. Images from similarly treated individuals across studies, that is, the same time Computed tomography (CT) imaging after exposure, to the same strain and dose of M. tuber- culosis were combined for each analysis. The images CT scans were collected from sedated animals using a from 14 rhesus macaques and 18 cynomolgus macaques 16 slice Lightspeed CT scanner (General Electric collected less than six days before necropsy were used Healthcare, Milwaukee, WI, USA) 3, 8 and 11 to evaluate the correlation between the CT score and weeks after aerosol exposure to M. tuberculosis total pathology score. The number of images available (Figure 1), as described previously.11 In order to for each assessment of the CT score system utility is improve the characterisation of lesions and lymph shown in Table 2. nodes Niopam 300 (Bracco, Milan, Italy), a non- ionic, iodinated contrast medium, was administered Necropsy intravenously (IV) at 2 ml/kg body weight. Scans were evaluated by an expert radiologist with 30 Animals were anaesthetised and blood samples and years’ experience of interpreting chest CT scans clinical data collected prior to euthanasia by intra-car- (FG) blinded to the animal’s treatment and clinical diac injection of a lethal dose of anaesthetic status, for the number and distribution of pulmonary (Dolelethal, Ve´toquinol UK Ltd, 140 mg/kg). A post- lesions across lung lobes and the presence of nodules mortem examination was performed immediately and (Figure 1(a) and (b)) cavitation (Figure 1(c)), con- gross pathological changes were scored using an estab- glomeration, consolidation (Figure 1(d)) as an indica- lished system based on the number and extent of lesions tor of alveolar pneumonia, a ‘tree-in-bud’ pattern present in the lungs, spleen, liver, kidney and lymph (Figure 1(e)) as an indicator of bronchocentric pneu- nodes, as described previously.10 Procedures were con- monia and lobular collapse (Figure 1(f)). The airways ducted by personnel blinded to treatment group were evaluated for the occurrence of wall thickening allocation. (Figure 1(g)) and presence of bronchoceles. The lymph nodes were assessed for enlargement (Figure Statistical analyses 1(h)) and the presence of necrosis (Figure 1(i)). Extra-pulmonary tissues including liver (Figure 1(j)), The CT scans examined for this work were collected kidneys (Figure 1(k)) and spleen (Figure 1(l)) were from macaques enrolled in studies designed in accord- examined for the presence of single or multiple foci ance with statistical advice such that the minimum of disease, cavitation or necrosis. numbers of animals were used in group sizes sufficient to robustly address the specific aims of each study. CT score system Differences in the CT score were compared between species and test groups using the non-parametric The disease burden attributable to infection with Mann–Whitney U test. The Spearman correlation test M. tuberculosis was scored using a relative scoring was used to determine the level of correlation between system (Table 1) based on the number of lesions present qualitative pathology scores and CT scores. The distri- in lungs, spleen, liver, kidney and lymph nodes and the butions of each data set were compared using the presence and extent of TB-induced structural abnorm- D’Agostino and Pearson’ omnibus normality test, alities. The scores attributed, based on the number of which indicated a normal distribution for all data sets nodules present in each tissue were stratified using with sufficient data points. Non-parametric tests were ranges in line with those described in systems used to selected for these analyses because of the relatively quantify TB-induced gross pathology in macaques.12 small data sets. All analyses were performed using Scores were also applied for the presence of disease GraphPad Prism, version 5.01 (GraphPad Software features, such as cavitation or necrosis, as these were Inc, La Jolla, CA, USA). Sharpe et al. 603

Figure 1. Examples of TB-induced disease features: (a) well-defined pulmonary nodule; (b) widespread discrete pul- monary nodules; (c) cavitating pulmonary nodule; (d) discrete area of consolidation; (e) tree-in-bud pattern; (f) lobar collapse and consolidation; (g) airway disease; (h) enlarged precarinal lymph node; (i) necrotic paratracheal lymph node; (j) liver: multiple small abscesses; (k) bilateral renal abscess; (l) multiple splenic abscesses. Arrows indicate features.

Results the CT score system should be comparable to that mea- TB-induced disease burden assessed from sured using established approaches. To evaluate the CT scans collected in life is comparable to legitimacy of the CT score system, the disease burden disease burden determined at necropsy scores from CT images were compared to scores deter- based on gross pathology mined by application of an established gross pathology- based system. Disease burden was measured using both In order to provide a valid tool for the quantification of systems in 14 rhesus macaques and 18 cynomolgus TB-induced disease burden, the disease measured using macaques that had been infected with M. tuberculosis 604 Laboratory Animals 52(6)

Table 1. CT disease burden scoring system.

Tissue Clinical feature Measure Score

Lung lobe Number of nodules none 0 1–3 1 4–10 2 10–20 3 20–50 4 >50 5 Conglomerate nodules Presence of conglomerate nodules 1 Cavitation Presence of cavities 1 Consolidation Presence of consolidation as a discrete 1 area within the lobe Whole lobe consolidated 5 Tree-in- bud Presence 1 Lobular collapse Presence 5 Airway Bronchi/ole wall thickening Presence of bronchi/ole wall thickening 1 Bronchoceles Presence of bronchoceles 1 Lymph node Enlargement and necrosis Visible not enlarged 0 Enlarged 1 Enlarged and necrotic 2 Extensive necrosis 3 Extra-pulmonary organ Number of disease foci No visible lesions 0 One small focus of disease 1 Several foci of infection 5 Extensive necrosis 1

Table 2. Number and origin of CT scan images analysed to assess CT score system utility.

M. tuberculosis Number of CT scan images for analysis Drug treatment Time of image collection post exposure (week)

Strain Dose range Route 3 8 11 16 24 Erdman Ultra-low Aerosol No 15 11 11 na na Erdman Medium Aerosol No 11 5 na na na Rhesus Erdman Medium Aerosol No 2 2 na na na Erdman Medium Intra bronchial No 2 2 na na na Erdman Ultra-low Aerosol No 17 13 4 na na Erdman High Aerosol No 17 17 na na na Cynomolgus Erdman Medium Aerosol No 9 na na na na H37Rv Medium Aerosol No 9 na na na na Erdman Medium Aerosol Yes na 8 na 8 8 na: not available.

Erdman strain for between three and 32 weeks at the determined using the CT system and the gross path- time of euthanasia. The CT score system was applied to ology system when the results from both macaque spe- images collected from subjects less than six days before cies were combined (p ¼ <0.0001, r ¼ 0.7405), or as euthanasia and the pathology score system was applied single species (rhesus: p ¼ 0.0178, r ¼ 0.6305; cynomol- during necropsy. Spearman’s correlation analysis gus macaques: p ¼ 0.0026, r ¼ 0.6649) (Figure 2). This revealed significant correlations between scores indicated that the disease burden measured using the Sharpe et al. 605

p ¼ 0.0096; cynomolgus, week 3 p 0.0001, week 8 p ¼ 0.0049). The median scores showed the disease burden to increase significantly between weeks three and eight in the rhesus macaques exposed to either, ULD (p ¼ 0.0275) or medium doses, (p ¼ 0.0039) of M. tuberculosis and in the cynomolgus macaques fol- lowing high dose (p ¼ 0.0049) but not ULD exposure (p ¼ 0.0804). Further comparison of the median CT scores determined from scans collected three, eight and eleven weeks after ULD exposure in rhesus (week 11, n ¼ 11) and cynomolgus macaques (week 11, n ¼ 4) (Figure 3(c)) revealed the disease burden to continue to increase significantly from week three to week eleven (p ¼ 0.004) in rhesus macaques, in contrast the disease burden in cynomolgus macaques did not show signifi- Figure 2. Correlation plot comparing TB-induced disease cant increases during the same period (p ¼ 0.2854). burden in rhesus and cynomolgus macaques scored using Furthermore, disease burden was significantly greater an established system based on gross pathological in rhesus macaques than cynomolgus macaques at all changes applied at necropsy, with disease burden scored three time points assessed (week 3: p ¼ 0.0014, week 8: using the CT score system applied to images collected less p ¼ 0.0007, week 11: p ¼ 0.0168). than 6 days before necropsy. Data points represent indi- Next the CT score system was applied to evaluate vidual animals, and the macaque species are indicated by the impact of M. tuberculosis strain on the disease symbol shape and fill colour. Spearman’s correlation burden induced. Data were scored from CT scans col- coefficient (r) and significance values (p) are indicated. lected from two groups of nine cynomolgus macaques, three weeks after aerosol exposure to a presented dose CT score system was directly comparable to the burden of between 192 and 271 CFU of M. tuberculosis strain measured by examination of gross pathology changes. Erdman or M. tuberculosis strain H37Rv. The scores Therefore, disease burden determined from CT scans did identify a significantly higher level of disease burden provides a relevant measure of disease burden that in cynomolgus macaques following aerosol exposure to can be applied ‘in life’. M. tuberculosis Erdman strain compared to H37Rv (p ¼ 0.0028) (Figure 2(d)). Application of the CT score system for The influence of the route of M. tuberculosis expos- model characterisation and refinement ure on disease burden was investigated through appli- cation of the CT scoring system to data collected three Following validation of the disease scores derived from and eight weeks after exposure from in four rhesus CT image analysis as a measure of the pathological macaques in a small pilot study (Figure 2(e)).25 changes caused by M. tuberculosis infection, the next Whilst the small number of macaques in the study pro- step was to evaluate the utility this approach to discrim- hibited the demonstration of significant differences inate differences in disease outcome. The CT score between delivery routes, the scores identified a trend system was applied to evaluate whether changes to spe- for higher disease burden in rhesus macaques following cific parameters impact on the outcome of infectious aerosol exposure to M. tuberculosis compared to challenge in the TB macaque model. CT scores were bronchoscopic placement of the inoculum. calculated from images collected three and eight weeks after exposure to either an ultra-low dose Application of the CT score system (ULD), or higher dose of M. tuberculosis strain for evaluation of the efficacy of Erdman from rhesus (ULD < 40 CFU: week 3 n ¼ 15, interventions against TB week 8 n ¼ 11; medium dose range 46–250 CFU: week 3 n ¼ 8, week 8 n ¼ 5) (Figure 3(a)) and cynomolgus If a new tool for measuring disease burden is to be macaques (ULD < 40 CFU: week 3 n ¼ 17, week 8 useful for evaluating the efficacy of new interventions n ¼ 13, high dose range 100–500 CFU: week 3 n ¼ 13, against TB, it needs to possess the power to show that week 8 n ¼ 13) (Figure 3(b)). Comparison of the median treatment corresponds to a reduction in the disease scores for each group revealed significantly greater dis- burden induced after infection. To evaluate the poten- ease burden in the macaques of both species, following tial of the CT score system to provide a measure of an exposure to the higher dose of M. tuberculosis, at both overall treatment effect, the extent of TB disease burden time points (rhesus, week 3 p ¼ 0.0059, week 8 was scored from CT scans collected from eight 606 Laboratory Animals 52(6)

Figure 3. CT score system for model characterisation and refinement. The effect of aerosol dose of M. tuberculosis Erdman strain on disease burden assessed three and eight weeks after exposure: (a) CT scores calculated from CT images of rhesus macaques following exposure to either an ultra-low (PD < 40 CFU) or a medium dose (PD 46–250 CFU); (b) CT scores calculated from CT images of cynomolgus macaques following exposure to either an ultra-low (PD < 40 CFU) or a high dose (PD 100–500 CFU). (c) Influence of macaque species on outcome of exposure to an ultra-low dose (PD < 40 CFU) of M. tuberculosis Erdman strain measured by CT score. (d) Pulmonary disease burden measured by CT score induced three weeks after aerosol exposure to a presented dose of between 192 and 271 CFU of M. tuberculosis strain Erdman or M. tuberculosis strain H37Rv in cynomolgus macaques. (e) Influence of route of M. tuberculosis Erdman strain delivery on the outcome of exposure measured by CT score in rhesus macaques. Data points represent individual animals with medians indicated. Horizontal bars represent significant differences between test groups as determined by Mann–Whitney test with significance values (p) indicated above. cynomolgus macaques enrolled in a trial of two anti-TB regimens. Whilst both treatment regimens caused a combination therapies prior to, at the mid-point of, and non-significant reduction in burden over the 16 weeks at the end of drug treatment. Comparison of the of treatment, the reduction was more pronounced in median scores calculated from scans from all eight ani- group B that received isoniazid/ethambutol then isonia- mals revealed a significant reduction in disease burden zid rifampicin/pyrazinamide (Figure 4(b)) (Non-signifi- following treatment with combinations of isoniazid cant p values from comparisons between weeks 8, 16 rifampicin/pyrazinamide and isoniazid/ethambutol and 24: group A: 0.3750, 0.8750, 0.2500; group B: relative to the burden present at the start of treatment, 0.2500, 0.1250, 0.1250.) after the first eight week phase of treatment (p ¼ 0.0469) and the full 16 weeks of treatment (p ¼ 0.0234) (Figure Discussion 4(a)). Interrogation of the CT scores calculated from each test group of four macaques separately for evi- A score system has been developed that quantifies dence of treatment effects suggested potential differ- the disease burden induced by infection with ences in the efficacy conferred by the two treatment M. tuberculosis in macaques, based on the extent and Sharpe et al. 607

Figure 4. The effect of anti-TB therapy on TB-induced disease burden scored from CT images. CT scores calculated from CT images collected from cynomolgus macaques infected with M. tuberculosis Erdman strain prior to the start of treatment with anti-TB combination therapies (week 8), at the mid-point of treatment (week 16) and the end of treatment (week 24): (a) the cumulative effect of treatment with combinations of isoniazid/rifampicin/pyrazinamide and isoniazid/ ethambutol; (b) the effect of treatment with isoniazid/rifampicin/pyrazinamide, then isoniazid/ethambutol (group A) compared with treatment with isoniazid/ethambutol then isoniazid/rifampicin/pyrazinamide (group B). Data points represent individual animals with medians indicated. Horizontal bars represent significant differences between test groups as determined by Mann–Whitney test with significance values (p) indicated above. features of disease visible on CT images captured in M. tuberculosis dose and strain. Analysis of CT scores reports of scans conducted by an experienced consult- revealed the disease burden to be higher, and increase ant radiologist blinded to treatment status. Tools that more rapidly, in rhesus macaques than in cynomolgus measure the disease burden induced following infec- macaques exposed to equivalent aerosol doses. The tious challenge in animal models are critical as cur- subsequent conclusion that cynomolgus macaques pos- rently this is the only way to demonstrate the efficacy sess a superior ability to control TB is in in line with of the new interventions being developed to combat other reports that have compared M. tuberculosis infec- TB. Score systems based on changes in gross pathology tion in the two macaque species.3,4,6,11 Findings from seen at necropsy provide an accepted and well estab- the assessment of the effects of TB exposure dose and lished approach for quantifying disease burden in the strain measured by CT score also agreed with previous macaque model of TB, that have been used to demon- reports, with increased disease burden measured fol- strate the effects of experimental treatments.6–11 lowing exposure to larger aerosol doses,10,11 or after However, the approach is limited as assessment can infection with M. tuberculosis strain Erdman as only be applied to each subject on a single occasion. opposed to the H37Rv strain.19 The parallels between We have shown that the measure of disease burden our findings based on the analysis of CT scores with provided by the new CT scoring system accurately studies reported that used other measures of disease reflects the extent of TB-induced disease determined burden support the validity of the approach. using pathology scoring in both the rhesus and cyno- The analysis of CT scores also has the potential to molgus macaque species. The strong correlation provide new insights in to the macaque TB model. We between the scores derived from the CT and pathology have previously described the qualitative difference in data give confidence that the scores derived from CT the pattern of pulmonary disease induced by M. tuber- images will provide a useful additional approach for culosis after either aerosol exposure, or intra-bronchial quantifying disease burden in macaques. placement of an equivalent dose.24 As part of the work To evaluate the utility of the CT score approach to conducted for this report, the CT score system was discriminate effects within the TB macaque model, we applied to scans collected from animals in the pilot assessed the capacity of CT scores to reveal the impact study and enabled quantification of the influence of on disease burden induced by changes to variable com- route of delivery on disease outcome for the first ponents of the model, such as the macaque species, time. Results revealed the development of an increased 608 Laboratory Animals 52(6) level of disease following aerosol exposure compared to further assessment of the utility and value of this that induced after intra-bronchial placement. Although approach. It will be particularly important to trial the the numbers of animals in this pilot study are small, the system in studies designed to assess vaccine efficacy as clear difference in the disease is suggestive of an influ- this is an arena which needs approaches capable of ence of delivery route on infection outcome that war- demonstrating treatment effects. rants further investigation. Thus the information provided by the CT score system can be used to refine Acknowledgements the models applied in future studies allowing selection This work was supported by the Department of Health, UK. of the most appropriate model parameters to meet the The views expressed in this publication are those of the needs of the experimental hypothesis under test. The authors and not necessarily those of the Department of development of new interventions against TB relies on Health. We thank the staff of the Biological Investigations the provision of refined TB macaque models that Group at PHE Porton for assistance in conducting studies, include sensitive measures of disease burden capable and Ann Williams, Simon Clark and Faye Lanni for aero- of discriminating differences between treatment biology and bacteriology support. groups to test the efficacy of new vaccines and drugs. The drug treatment study reported here shows that CT Declaration of Conflicting Interests scores not only provide a measure that can demonstrate The author(s) declared no potential conflicts of interest with a statistically significant effect on disease burden, but respect to the research, authorship, and/or publication of this also highlight differences in the effects of the two treat- article. ment regimens. Quantification of disease burden from CT images Funding provides an attractive approach for gathering data on The author(s) disclosed receipt of the following financial sup- disease burden as images can be collected using non- port for the research, authorship, and/or publication of this invasive procedures from live animals. In contrast, the article: from The Department of Health, UK. pathology score system can only be applied after the death of the subject, and therefore disease burden can References only be measured on a single occasion. Consequently, 1. World Health Organization. Global tuberculosis report for studies, such as those to characterise disease caused 2016, http://www.who.int/tb/publications/global_report/ by a new agents or to define optimal treatment effects, gtbr2016_executive_summary.pdf where there is a need to evaluate disease burden on 2. Colditz GA, Brewer TF, Berkey CS, et al. Efficacy of BCG multiple occasions, the serial sacrifice of groups of simi- vaccine in the prevention of tuberculosis. Meta-analysis of larly treated animals has been used to enable pathology the published literature. JAMA 1994; 271: 698–702. to be determined over time. As CT images can be col- 3. Scanga CA and Flynn JL. Modelling tuberculosis in non- lected from the same individual on regular occasions human primates. Cold Spring Harbor Perspect Med 2014; after experimental infection there is an opportunity to 4: a018564. evaluate the evolution of disease or treatment effect. 4. Pen˜ a JC and Ho WZ. Monkey models of tuberculosis: Using this approach the CT score system significantly lessons learned. Infect Immun 2015; 83: 852–862. 5. Kaushal D, Mehra S, Didier PJ, et al. The non-human reduces the number of animals needed per study, for primate model of tuberculosis. J Med Primatol 2012; 41: example, a study requiring the evaluation of disease 191–201. burden on five occasions in a group of six subjects 6. Langermans JAM, Anderson P, van Soolingen D, et al. using CT, would require 30 individuals to perform the Divergent effect of bacillus Calmette-Guerin (BCG) vaccin- same evaluation using gross pathology scores. The use ation on Mycobacterium tuberculosis infection in highly of serial imaging over serial sacrifice, not only enables related macaque species: implications for primate models in reduction, but also provides refinement, as change can tuberculosis vaccine research. PNAS 2000; 98: 11497–11502. be evaluated within the same individual reducing the 7. Langermans JA, Doherty TM, Vervenne RA, et al. opportunity for inter-animal variation to mask experi- Protection of macaques against Mycobacterium tubercu- mental effects and therefore improving data quality. losis infection by a subunit vaccine based on a fusion pro- The ability to quantify subtle pathological changes tein of antigen 85B and ESAT-6. Vaccine 2005; 23: through imaging also has the potential to reduce the 2740–2750. 8. Verreck FAW, Vervenne RAW, Kondova I, et al. need to progress to higher levels of disease and thus MVA.85A boosting of BCG and an attenuated phoP defi- to reduce the severity experienced during a study. cient M. tuberculosis vaccine both show protective efficacy Our initial trials suggest that the CT score system against tuberculosis in rhesus macaques. PLoS One 2009; developed provides a valuable system for the measure- 4: e5264. ment of disease burden. Application of the score system 9. Verreck FAW, Tchilian EZ, Vervenne RAW, et al. to CT scans from more studies is now required to allow Variable BCG efficacy in rhesus populations: pulmonary Sharpe et al. 609

BCG provides protection where standard intra-dermal tuberculosis chemotherapy regimens. Antimicrob Agents vaccination fails. Tuberculosis 2017; 104: 46–57. Chemother 2013; 57: 4237–4244. 10. Sharpe SA, McShane H, Dennis MJ, et al. Establishment 18. Mitchell JL, Mee ET, Almond NM, et al. of an aerosol challenge model of tuberculosis in rhesus Characterisation of MHC haplotypes in a breeding macaques and an evaluation of endpoints for vaccine colony of Indonesian cynomolgus macaques reveals a testing. Clin Vaccine Immunol 2010; 17: 1170–1182. high level of diversity. Immunogenetics 2012; 64: 123–129. 11. Sharpe S, White A, Gleeson F, et al. Ultra-low dose aero- 19. Gormus BJ, Blanchard JL, Alvarez XH, et al. Evidence sol challenge with Mycobacterium tuberculosis leads to for a rhesus monkey model of asymptomatic tuberculosis. divergent outcomes in rhesus and cynomolgus macaques. J Med Primatol 2004; 33: 134–145. Tuberculosis 2016; 96: 1–12. 20. James BW, Williams A and Marsh PD. The physiology 12. Lin PL, Rodgers M, Smith L, et al. Quantitative com- and pathogenicity of Mycobacterium tuberculosis grown parison of active and latent tuberculosis in the cynomol- under controlled conditions in a defined medium. J Appl gus macaque model. Infect Immun 2009; 77: 4631–4642. Microbiol 2000; 88: 669–677. 13. Lewinsohn DM, Tydeman IS, Frieder M, et al. High 21. Clark SO, Hall Y, Kelly DL, et al. Survival of resolution radiographic and fine immunologic definition Mycobacterium tuberculosis during experimental aeroso- of TB disease progression in the rhesus macaque. lization and implications for aerosol challenge models. Microbes Infect 2006; 8: 2587–2598. J Appl Microbiol 2011; 111: 350–359. 14. Dennis M, Parks S, Bell G, et al. A flexible approach to 22. Druett HA. A mobile form of the Henderson apparatus. imaging in ABSL-3 laboratories. Appl Biosaf 2015; 20: J Hyg 1969; 67: 437–448. 89–98. 23. Harper GJ and Morton JD. The respiratory retention of 15. Scanga CA, Lopresti BJ, Tomko J, et al. In vivo imaging bacterial aerosols: experiments with radioactive spores. in an ABSL-3 regional biocontainment laboratory. Epidemiol Infect 1953; 51: 372–385. Pathog Dis 2014; 71: 207–212. 24. Sibley L, Dennis M, Sarfas C, et al. Route of delivery to 16. Lin PL, Maiello P, Gideon HP, et al. PET CT identifies the airway influences the distribution of pulmonary disease reactivation risk in cynomolgus macaques with latent but not the outcome of Mycobacterium tuberculosis infec- M. tuberculosis. PLoS Pathog 2016; 12: e1005739. tion in rhesus macaques. Tuberculosis 2016; 96: 141–149. 17. Lin PL, Coleman T, Carney JP, et al. Radiologic responses in cynomolgus macaques for assessing

Re´sume´ Jusqu’a` ce que des corre´lats de protection valide´s soient identifie´s, les mode`les animaux restent la seule fac¸on de tester l’efficacite´ de nouveaux vaccins et me´dicaments rapidement ne´cessaires pour lutter contre l’e´pide´mie mondiale cause´e par l’infection a` Mycobacterium tuberculosis. Les primates non humains (PNH) offrent les mode`les les plus pertinents de tuberculose (TB) humaine et sont au cœur du processus de de´veloppement de nouvelles interventions. L’efficacite´ des e´valuations sont tributaires de la capacite´ du mode`le d’essai a` de´terminer les re´sultats ame´liore´s entre les groupes traite´s apre`s une exposition expe´ri- mentale a` la M. tuberculosis. La capacite´ de mesurer le fardeau de la maladie induit par la tuberculose s’ave`re par conse´quent essentielle a` ce processus. Nous avons de´veloppe´ un syste`me de score qui nous permet de quantifier le fardeau de la maladie induit chez le macaque par l’infection a` M. tuberculosis,en fonction de l’e´tendue et des caracte´ristiques de la maladie visible sur les images de tomodensitome´trie (TDM). Le fardeau de la maladie a e´te´ de´termine´ par TDM puis ve´rifie ´ par rapport a` celui obtenu a` l’aide d’une approche fonde´e sur la pathologie. Les essais du syste`me comme outil de mesure du fardeau de la maladie ont montre´ que l’approche e´tait susceptible de re´ve´ler les diffe´rences entre les groupes de traite- ment afin de (a) caracte´riser les re´sultats de l’infection et de permettre l’ame´lioration du mode`le ; (b) de´montrer l’efficacite´ des traitements me´dicamenteux en pre´sentant des diffe´rences de re´sultats entre les groupes expe´rimentaux et te´moins. Les essais pre´liminaires indiquent que le syste`me de score base´ sur l’imagerie constitue un pre´cieux outil supple´mentaire pour mesurer le fardeau de la maladie induit par la tuberculose en offrant la possibilite´ d’appliquer a` la fois le raffinement et la re´duction dans les e´tudes.

Abstract Bis validierte Schutzkorrelate identifiziert sind, bleiben Tiermodelle die einzige Mo¨glichkeit, die Wirksamkeit der neuen Impfstoffe und Medikamente zu testen, die dringend beno¨tigt werden, um die globale, durch Infektion mit Mycobacterium tuberculosis verursachte Epidemie zu beka¨mpfen. Nichtmenschliche Primaten (NHP) bieten die zweckdienlichsten Modelle der menschlichen Tuberkulose (TB) und sind fu¨r den Entwicklungsprozess neuer Maßnahmen von zentraler Bedeutung. Wirksamkeitsbewertungen ha¨ngen von 610 Laboratory Animals 52(6) der Fa¨higkeit des Versuchsmodells ab, verbesserte Ergebnisse zwischen behandelten Gruppen nach experi- menteller Exposition gegenu¨ber M. tuberculosis zu unterscheiden, und daher ist die Fa¨higkeit, TB-induzierte Krankheitsbelastung zu messen, von zentraler Bedeutung fu¨r den Prozess. Wir haben ein Score-System entwickelt, das uns erlaubt, die Krankheitsbelastung zu quantifizieren, die bei Makaken durch eine Infektion mit M. tuberculosis induziert wird, basierend auf dem Ausmaß und den Merkmalen der Krankheit, die auf Computertomographie-Bildern (CT) sichtbar sind. Die per CT ermittelte Krankheitsbelastung wurde mit derjenigen verglichen, die mit Hilfe eines etablierten pathologischen Ansatzes ermittelt wurde. Versuche mit dem System als Instrument zur Messung der Krankheitsbelastung haben gezeigt, dass es mit diesem Ansatz mo¨glich ist, Unterschiede zwischen den Behandlungsgruppen aufzudecken, um (a) das Infektionsergebnis zu charakterisieren und eine Modellverbesserung zu ermo¨gli- chen; (b) die Wirksamkeit von Behandlungsschemata fu¨r Arzneimittel zu demonstrieren, indem Unterschiede im Ergebnis zwischen Test- und Kontrollgruppen aufgezeigt werden. Erste Studien deuten darauf hin, dass das bildgebende Score-System ein wertvolles zusa¨tzliches Instrument zur Messung der TB-induzierten Krankheitsbelastung darstellt, das die Mo¨glichkeit bietet, sowohl Verbesserung als auch Reduktion bei Studien zum Tragen zu bringen.

Resumen Hasta que se identifique una correlacio´n validada de proteccio´n, los modelos animales siguen siendo el u´nico modo de probar la eficacia de nuevas vacunas y fa´rmacos requeridos urgentemente para luchar contra la epidemia global causada por la infeccio´ndeMycobacterium tuberculosis. Los primates no humanos repre- sentan los modelos ma´s relevantes de tuberculosis humana (TB) y son esenciales para el proceso de desarrollo de nuevas intervenciones. Las evaluaciones de eficacia dependen de la capacidad del modelo de prueba a la hora de discriminar resultados mejorados entre grupos tratados despue´s de una exposicio´n experimental a la M. tuberculosis y, por tanto, la capacidad de medir la carga de TB inducida es primordial para el proceso. Hemos creado un sistema de puntuacio´n que nos permite cuantificar la carga de enfermedad inducida en macacos con M. tuberculosis, en base al alcance y las caracterı´sticas de los aspectos visibles de la enfermedad en ima´genes de tomografı´a computerizada (CT). La carga de enfermedad determinada por la CT fue entonces verificada frente a los datos obtenidos utilizando un me´todo probado basado en la patologı´a. Distintas pruebas del sistema como herramienta para medir la carga de enfermedad han mostrado el me´todo capaz de establecer diferencias entre grupos de tratamientos para (a) caracterizar resultados de infeccio´ny permitir el refinamiento del modelo; (b) demostrar la eficacia de los regı´menes de tratamiento con fa´rmacos revelando las diferencias de resultados entre grupos de prueba y de control. Las pruebas iniciales sugieren que el sistema de puntuacio´n basado en ima´genes representa una herramienta adicional valiosa para medir la carga de TB inducida que ofrece la oportunidad de aplicar refinamiento y reduccio´n en los estudios. Original Article Laboratory Animals 2018, Vol. 52(6) 611–620 ! The Author(s) 2018 Color matters: They would choose if Article reuse guidelines: sagepub.com/journals- they could (see)! permissions DOI: 10.1177/0023677218766370 journals.sagepub.com/home/lan Tina Nitezki1, Nadja Schulz2 and Stephanie Kra¨mer1,3

Abstract Concerning standardization of laboratory animal husbandry, only exiguous changes of habitat can potentially influence animal physiology or results of behavioral tests. Routinely, mice chow is dyed when different types of diets are dispensed. Given the fact that the dye itself has no effects on food odor or flavor, we wanted to test the hypothesis that the color of chow has an impact on food uptake in mice. Twelve-week-old male mice of different strains (C57BL/6J, DBA/2J, C3H/HeJ, BALB/cJ; n ¼ 12/strain) were single-housed in PhenoMasterÕ cages. After acclimatization standard mice chow in different colors was administered. Food intake was monitored as a two-alternative choice test of different color combinations. All animals had an average food intake of 3 g/d and no preferences were observed when a combination of identically colored food was offered. Preference tests yielded significant aversion to blue food and significant attraction to yellow and green food in C57BL/6 and DBA/2J mice. In C3H/HeJ and BALB/cJ mice no color- related pattern occurred. Selected mice strains have known differences concerning functionality of their visual sense. C57BL/6 and DBA/2 mice are considered to be normal sighted at testing age, BALB/c is representative for albino strains and C3H mice carry mutations resulting in retinal alterations. Results suggesting that normal-sighted mice would be selective concerning food color when given the choice. Nevertheless, this does not influence overall quantity of food intake when animals were provided solely with food colored with a single dye. Moreover, visually impaired mice showed no color-related food preferences.

Keywords refinement, color vision, food choice, color preference, eating

Date received: 18 August 2017; accepted: 27 February 2018

Introduction regarded as crepuscular creatures in general exhibiting activity peaks at the beginning and at the end of the In the second half of the last century much effort was 12-hour nocturnal phase, colors simply were neglected, spent on standardizing experimental animal husbandry as the spectral composition of light is not sufficient aiming to create a robust rationale, as it became enough for adequate visual discrimination.5 increasingly obvious, that animal care and housing itself has potential impact on in vivo studies.1,2 For 1 example, in animal facilities the implementation of an Department of Toxicology, Institute of Nutritional Science, University of Potsdam, Germany artificial 12-hour-light and 12-hour-dark regimen is 2German Institute of Human Nutrition, Germany acknowledged worldwide. The idea of installing this 3Department of Laboratory Animal Science and Animal Welfare, circadian light rhythm is based on the observation Justus-Liebig University Giessen, Germany that most organisms on earth feature a biological clock running a 24-hour cycle.3,4 Corresponding author: Tina Nitezki, Department of Toxicology, Institute of Nutritional Despite providing experimental animals with a con- Science, University of Potsdam, Arthur-Scheunert-Allee 114-116, trolled light system, not much is known about the D-14558 Potsdam, Germany. impact of colors in this setting. As mice and rats are Email: [email protected] 612 Laboratory Animals 52(6)

The ability to discriminate colors is due to the indi- development in mice, even comparable to genotypic vidual’s biological endowments of the visual system, manipulations. To facilitate the discrimination for the thereby recording changes in the spectral composition. user between different diets administered to different We know that vertebrates exhibit different color vision experimental groups, diets are dyed, presupposing systems depending on the evolutionary configuration of that the color of the provided chow is without impact their class of cone receptors.6 As diurnal creatures like on food intake. The dyes belong to the class of food humans and primates belong to the class of trichromates dyes and are expected to fulfil the requirements of being harboring three cone pigments, in contrast rodents com- neutral in taste, non-toxic and having no effects on monly have only two.7,8 Like most mammals, mice metabolism. Whether the color has a possible impact therefore belong to the dichromates whose retina con- on behavior associated with preferences or aversions tains two classes of cone photopigments with absorption resulting in alterations of chow consumption has not maxima at approximately 360 nm (ultraviolet sensitive) been investigated before. Therefore, the aim of this and 510 nm (green sensitive), red cones being absent.9,10 study was to solely investigate the impact of color on Thus, normal-sighted laboratory mouse strains like the feeding habits of different strains of laboratory mice C57BL/6 are expected to detect short-wave light, even depending on their eyesight. in ultraviolet ranges, but are unable to see colors with long wavelengths, like red. Moreover, there are several mouse strains (e.g. DBA/2, C3H) that develop visual Material and methods impairments with time or are albinotic (like BALB/c). Animals DBA/2 mice develop progressive eye abnormalities that closely mimic human hereditary glaucoma, having its All mice were bred in house in the animal facility onset between 3 and 4 months of age. Clinical signs of the German Institute for Human Nutrition include iris pigment dispersion, associated with iris atro- (Nuthetal, Germany). Mice were reared in individually phy, synechia, and elevated intraocular pressure.11,12 ventilated cages (SealSafe-IVC blue line, Tecniplast, C3H mice are homozygous for the Pde6b rd1 allele. Hohenpeißenberg, Germany) under specific, patho- Retinal degeneration 1 (Pde6brd1) is a spontaneous gen-free conditions and housed in groups (three to mutation that is present in multiple inbred strains, five animals per cage) under standard laboratory hous- including the C3H substrains. Strains that carry rd1 ing conditions (temperature: 22 0.5C, humidity: lack rod cells, becoming blind by weaning age.13,14 50 10%) with a diurnal 12-hour light and dark cycle BALB/c mice, comparable to other albinotic animals (lights on at 6:30 a.m.). Animals had free access to fresh of other species, have a significantly lower visual per- tap water and food. Cages were equipped with aspen formance than C57BL/6 mice, due to albinism.15 wood chips (LTE E-001; ABEDD Vertrieb GmbH, Albinism is a mutation in the tyronase gene, the rate- Wien, Austria), a paper towel and cotton fiber limiting enzyme in melanin synthesis (brown-black (Alpha-Nest; LBS Serving Biotechnology, Horley, eumelanin, red-yellow pheomelanin), resulting in hypo- UK) as nesting material and cardboard houses pigmentation of the retinal pigment epithelium.16 (Des.ResTM; LBS Serving Biotechnology, Horley, Interestingly, in mice, a correlation exists between the UK) as shelter. For this study, surplus animals from levels of melanin and rod numbers, so that entirely pig- in-house breeding were used. Since only male mice of ment-deficient mice have the lowest number of rods.17 the mouse strains of interest were available in sufficient In addition to normal-sighted mouse strains, albin- quantity, female mice were excluded. otic animals and others with visual impairment are All animal experiments were performed in compliance implemented in scientific animal research. It can be with the German animal protection law (TierSchG). hypothesized that the sensation of color is eminently Mice were housed and handled in accordance with different in varying mice strains, thereby accounting good animal practice as defined by Federation of for a possible variability in in vivo experiments. European Laboratory Animal Science Associations Routinely, in biomedical research special diets are (FELASA) (www.felasa.eu/guidelines.php) and the administered to mice, as this enables researchers to pro- national animal welfare body GV-SOLAS (www.gv- vide (or undersupply) animals with elected components, solas.de/index.html). The local authorities approved all featuring low stress, compared to application methods animal experiments (LAGV Brandenburg, Germany; like oral gavage or injections. These diets are strictly permit number: V3-2347-03-2011). controlled under the aspects of fulfilling the nutritional needs of the animals, palatability and stability in Diets and study design order to prevent random variations concerning the phenotypic expressions. There is strong evidence At the age of 12 weeks male mice of four different strains that nutritional factors can implicate behavioral (C57BL/6J, DBA/2J, C3H/HeJ, BALB/cJ; n ¼ 12/strain) Nitezki et al. 613 were randomized into different study groups. For assess- checked for normal distribution (Shapiro–Wilk test) and ment of detailed food intake animals were transferred for for homogeneity of variances (Levene’s test for equality of single-housing into PhenoMasterÕ cages (TSE-Systems, variances) and the appropriate statistical test was per- Bad Homburg, Germany). Housing conditions, equip- formed. To reveal differences of three or more independent ment and level of enrichment in PhenoMasterÕ cages samples (comparison of mouse strains), one-way analysis were equivalent to rearing cages. Animals were allowed of variance (ANOVA) was used for interval-scaled data, to acclimatize for one week on standard mice chow (SC which are normally distributed, and the Kruskal–Wallis 1320; Altromin, Lage, Germany) and food consumption test for non-normally distributed data. Since all according was assessed dependent on time of day. PhenoMasterÕ data were normally distributed, for comparison of food cages are equipped with two measuring stations for moni- consumption per mouse strain (e.g. left compared to toring food intake and a single station for water. Food right food rack) unpaired T test was applied. P values containers are placed oppositely in the short corners of below 0.05 were considered significant for all analyses the cage and were changed daily in order to exclude side (*p 0.05, **p 0.01, ***p 0.005). All data are pre- preferences of the animals. After acclimatization and per- sented as mean standard deviation (SD). formance of visual placing response (see below) chow of different colors was administered at the age of 13 weeks. The food color green (g) corresponds to the ‘‘natural’’ Results color of the SC 1320. Red (r), blue (b) and yellow (y) Visual placing response food was dyed on the basis of the SC 1320. Chow was prepared by the manufacturer Altromin, Lage, Germany) Performance in visual placing response differed signifi- simultaneously from one charge in the mill and was cantly between mouse strains (Kruskal–Wallis test; stored at –20C afterward in the animal facility until p 0.005). Approaching the cage grid, C57BL/6J (reac- usage. Before feeding portions of food were defrosted at tion score: 1.45 0.51) and DBA/2J (1.30 0.47) mice room temperature. This regimen was performed in order responded immediately to the visual stimulus. Mouse to prevent time-dependent effects on the food ingredients strains being potentially blind at testing age, C3H/HeJ due to storage. Food intake was monitored as a two- (0.10 0.31) and BALB/cJ (0.10 0.31), failed signifi- alternative choice test of different color combinations cantly to react appropriately (Figure 1). (g/b; g/r; g/y; y/r; y/b; r/b; y/y; r/r; b/b; g/g); measuring period: seven days per combination). Color combinations Average food consumption were provided in randomized order between animals of each mouse strain to exclude age-related color prefer- Average food consumption was assessed by monitoring ences. Before a new combination was applied, animals intake of SC at the beginning of the experiment. were allowed to equilibrate on SC for one day.

Visual placing response To assess the function of the visual system before begin- ning the two-alternative choice test, a visual placing response modified from Metz and Schwab18 and Pinto and Enroth-Cugell19 was performed. At the age of 13 weeks each mouse was suspended by holding its tail and then lowered toward a solid object (a fixed external cage grid) for one to two seconds without any contact with the vibrissae. Normally, when the head of the mouse was moved near the grid, it raised its head and extended the forelimbs to place them onto the grid. The procedure was conducted three times per animal and averaged. Reaction was rated with a scoring system: 0 represents no observable placing response, 1 indicates a weak and 2 a clear visual placing reaction. Figure 1. Performance in visual placing response. Graph shows reaction scores in the performance of visual placing Statistical analysis response of C57BL/6J, DBA/2J, C3H/HeJ and BALB/cJ mice (n ¼ 12 per group): Asterisks indicate significances: All statistical analyses were performed using IBM SPSS 23 *p 0.05, **p 0.01, ***p 0.005; Kruskal–Wallis test. (IBM Deutschland GmbH, Germany). All data were Data presented as means (SD). 614 Laboratory Animals 52(6)

The different mice strains showed some individualities (one-way ANOVA; p 0.005), but showed a strong concerning the quantity (C57BL/6J 18.2 1.36, DBA/2J preference for green food. Thereby, they always pre- 20.5 2.43, C3H/HeJ 22.1 1.84 and BALB/cJ ferred green to yellow, red and blue and preferred 18.2 1.82 g/week; one-way ANOVA; no significance), yellow to red and blue. Again blue was the color that but had in common that most food was eaten in the was preferred less. In total the consumption of green dark phase (D) compared to the light phase (L) food dominated with 40 4.0%, followed by yellow (C57BL/6J L:4.9 0.17, D:13.3 1.29 g/week; DBA/2J with 27 3.4%, red with 20 9.4% and blue with L:3.8 0.24, D:16.6 2.30 g/week; C3H/HeJ L:7.5 13 4.2% (one-way ANOVA; p 0.005; Figure 3). 0.36, D:14.6 1.78 g/week; BALB/cJ L:4.8 0.31, D:13.4 1.73 g/week; unpaired T test; p 0.05; Figure 2). Relative food consumption of C3H/HeJ Relative food consumption of C57BL/6J In contrast to C57BL/6J and DBA/2J, no specific pref- erence pattern could be observed in the C3H/HeJ mice, Concerning the two-alternative choice test, normal- even if a significantly higher consumption of green food sighted C57BL/6J mice discriminated significantly directly compared to yellow (unpaired T test; p 0.005) between offered food options (one-way ANOVA; and yellow food directly compared to red (unpaired p 0.005). They always preferred yellow to green, red T test; p 0.05) could be detected. Nevertheless, con- and blue and preferred green to red and blue. Blue was cerning the overall consumption, no significance could the color that they preferred less. In total the consump- be detected and the intake of green food preferences tion of yellow food dominated with 36 2.3%, were 26 6.3%, yellow at 24 5.0%, red at followed by green with 25 3.6%, red with 24 7.4% 22 4.0% and blue at 28 2.6% (Figure 4). and blue with 15 3.4% (one-way ANOVA; p 0.005; Figure 3). Relative food consumption of BALB/cJ

Relative food consumption of DBA/2J As the albinotic mouse strain BALB/cJ is considered to be visually impaired, as expected no specific attraction Comparable to the C57BL/6J mice, DBA/2J discrimi- concerning food color could be observed. In total nated significantly between offered food options BALB/cJ mice consumed 24 6.2% of green food, 26 3.3% of yellow food, 22 3.3% of red food and 28 1.6% of blue food (Figure 4).

Cumulated food consumption At the end of the experiment an additional side-by-side comparison was performed in order to exclude side preferences. Therefore, food of the same color was pro- vided in the two-alternative choice test, whereas the choice that had to be made was not the color of food, as it was identical, but the side of offered food in the cage. For all mice strains no side preferences could be detected (Figure 5). When animals were provided with food of the same color, no significant side preferences could be observed (unpaired T test). Furthermore, no significant color- dependent food attraction or aversion could be moni- Figure 2. Average food intake per animal per week tored (one-way ANOVA). Compared to the initial ana- (standard chow). Graph shows the average food intake lysis of average food consumption at the beginning of (standard chow) per animal per week (g/week) monitored our observations, all animals ranged near these results depending on the light and dark phase during husbandry of (Table 1). C57BL/6J, DBA/2J, C3H/HeJ and BALB/cJ mice (n ¼ 12 per group) at the beginning of the experiment. One-way ana- lysis of variance revealed no significant differences in Discussion overall food consumption. Unpaired T test revealed sig- nificant differences depending on light or dark phase for all In many animal studies food choice is correlated to mouse strains; ***p 0.005. Data presented as means chemosensory or cognitive aspects, but not toward (SD). the appearance of food. However, there are mouse Nitezki et al. 615

Figure 3. Average food intake of C57BL/6J and DBA/2J mice per week (experimental chow). Graphs show the average food intake of green- (dashed), blue- (horizontal lines), yellow- (uniform) and red- (vertical lines) stained food for each combination (E1–E6) of (a) C57BL/6J mice (n ¼ 12) and (b) DBA/2J mice (n ¼ 12); per animal per week (g/week) assessed with a two-choice preference test; unpaired T test; ***p 0.005 and relative intake of administered stained food (%) averaged per color over all time points for (c) C57BL/6J and (d) DBA/2J mice.

Figure 4. Average food intake of C3H/HeJ and BALB/cJ mice per week (experimental chow). Graphs show the average food intake of green- (dashed), blue- (horizontal lines), yellow- (uniform) and red- (vertical lines) stained food for each combination (E1–E6) of (a) C3H/HeJ mice (n ¼ 12) and (b) BALB/cJ mice (n ¼ 12); per animal per week (g/week) assessed with a two-choice preference test; unpaired T test; *p 0.05; ***p 0.005 and relative uptake of administered stained food (%) averaged per color over all time points for (c) C3H/HeJ and (d) BALB/cJ mice. 616 Laboratory Animals 52(6)

Figure 5. Assessment of possible side preferences for food consumption. Graphs show (a) average food uptake of food of the same color: green (dashed), blue (horizontal lines), yellow (uniform) and red (vertical lines) of C57BL/6J, DBA/2J, C3H/ HeJ and BALB/cJ mice (n ¼ 12 per group) per animal per week (%/week) assessed with a two-choice preference test (line between same-colored bar indicates left or right food rack) and (b) relative overall intake of administered stained food (%). models that are inevitably based on high food intake in colors. In general, mice are accepted as animal models a short time, such as models of diet-induced obesity.20 for investigating the visual sense, as they show major Colored foods could probably enhance or inhibit food congruity, sharing similar cell types and structural fea- intake, potentially influencing the development of the tures.10,21 For our study we selected the C57BL/6J model and thus affect the results obtained. mouse, as being representative for a normal-sighted In this study, we documented that mice of different mouse (close to its wild living relatives), the DBA/2J strains, providing variations in their genetic ability of mouse, as a representative for being born normal eyesight, indeed showed preferences for distinct food sighted, but developing impaired visual capacity with Nitezki et al. 617

Table 1. Two-alternative-choice-test for assessment of side preferences and average food uptake.

Yellow vs. yellow Green vs. green Red vs. red Blue vs. blue

C57BL/6J Left 10.1 1.1 9.8 1.0 7.9 1.2 10.1 1.8 Right 9.2 1.3 9.4 0.8 9.1 1.2 9.2 1.9 Total 19.3 1.1 19.2 0.9 17.9 1.2 19.3 1.8 DBA/2J Left 9.5 1.8 8.9 1.5 10.1 3.4 10.1 0.9 Right 11.5 2.3 12.1 2.2 14.9 2.9 10.0 1.2 Total 21.0 2.2 21.0 1.8 25.0 3.1 20.1 0.9 C3H/HeJ Left 7.9 1.7 10.5 0.8 9.4 2.2 8.0 1.4 Right 10.9 0.9 10.1 1.3 10.7 1.7 10.8 1.9 Total 18.8 1.4 20.6 1.1 20.1 1.8 18.8 1.4 BALB/cJ Left 9.7 1.1 9.3 0.8 9.0 1.6 14.1 3.1 Right 11.5 0.6 12.0 1.5 12.2 0.9 11.8 2.2 Total 21.2 0.8 21.3 1.2 21.2 1.4 25.9 2.9

Food of the same color was provided in two feeding tubes located on the left and right corner of the PhenoMasterÕ cage. Food consumption was monitored for one week (g/week) after equilibration on standard diet for one day. Data presented as means SD. Unpaired T test revealed no significant differences depending on position of food rack. age, the C3H/HeJ mouse, which becomes visually pointing at attraction or dislike could be detected. Both impaired at an early age, and the BALB/cJ mouse, a mice strains took in physiological amounts of food classic albinotic mouse strain.10 without any preference for yellow, green, red, or blue. Visual orienting tests of all tested mice verified the This fact is important as it documents that the chosen assumptions concerning the visual capacities and accur- food colors or the process of manufacturing and stor- acy of being designated into the testing groups. At the age had no effect on taste perception, or an olfactory age of 13 weeks DBA/2J mice performed well in visual impact, respectively. placing response, so it can be assumed that they were Furthermore, when not given the choice, mice of all normal sighted at the beginning of the two-choice alter- mouse strains (independent of visual capacity) took in a native tests. Mice of this strain develop retinal degen- physiological amount of colored food, comparable to eration at the cellular level at the age of 4 months, the food intake on SC at the beginning of the study. becoming worse with time and leading to complete As a retrospective consideration, these findings are of blindness approximately at the age of 9 to 12 great relevance as the outcome of many studies using months.22 It has to be mentioned that the visual placing differently dyed diets must not be doubted. Our response is classifiable as an orientation test and allows research supports the assumption that the net quantity the analysis of whether an animal is able to recognize of food intake seems to be primarily dependent on more- or less-tangible objects.23 Indeed, no statement metabolic influencing factors (e.g. caloric intake).25,26 can be made about possible deviations in terms of the In recent years the implementation of the replace- animals’ color perception. ment, reduction and refinement (the 3Rs) concept has Our results show that the normal-sighted C57BL/6J been enforced intensively. Within that, numerous inves- mice preferred chow tinted yellow or as an alternative, tigations concerning influences of enrichment materials, green. These colors are obviously close to the natural cage conditions, embedding material and others have situation. In a study conducted in wild house mice the been performed. So we know, for example, that the favored food was found to be whole canary seeds, selection of a home cage is not only due to the cage which are a yellowish-greenish color. Also, pinhead color, but also to the structure and color of the supplied oatmeal and wheat were well accepted.24 Comparably, embedding material. If mice are given a choice, they DBA/2J mice tested at an age of visual competence, would prefer embedding material similar to their own preferred food that was green, followed by yellow com- fur coat color.27 Noteworthy, this behavior might pared to red and blue food pellets. In both sighted mice implicate a kind of self-reflection in addition to color strains blue was the color preferred less. This might be preferences and leads to the assumption that colors explained by the rare occurrence of blue food in nature, play a key role in the well-being of mice. Focusing on except for some berries, potatoes, eggplant or cabbage. refinement aspects, using dyed diets in a preferred color As we expected, the visual sense at the testing time might be of importance under pathological conditions. point of C3H/HeJ mice and BALB/cJ was strongly Further studies have to be conducted in order to ana- impaired. For that reason, it is valid that no pattern lyze whether food color preferences in combination 618 Laboratory Animals 52(6) with enhanced palatability can help to prevent excessive 4. Karatsoreos IN, Bhagat S, Bloss EB, et al. Disruption of weight loss and thereby support animal well-being. circadian clocks has ramifications for metabolism, brain, With respect to conducting experiments associated and behavior. Proc Natl Acad Sci U S A 2011; 108: with the administration of differently dyed diets, it is 1657–1662. important to mention that, given a choice, normal- 5. Jacobs GH, Williams GA, Cahill H, et al. Emergence of sighted mice prefer one color over another, but if novel color vision in mice engineered to express a human cone photopigment. Science 2007; 315: 1723–1725. there is no such option, the intake of preferred colored 6. Jacobs GH. Evolution of colour vision in mammals. food to disliked colored food is comparable. In sum, Philos Trans R Soc Lond B Biol Sci 2009; 364: 2957–2967. whenever the purpose of a study is to directly compare 7. Kalloniatis M and Luu C. The perception of color. 2005 the intake of different foods, diets should not be col- (updated 9 July 2007). In: Kolb H, Fernandez E and ored with various dyes because (at least normal- Nelson R (eds) Webvision: The organization of the retina sighted) mice obviously choose when given a choice. and visual system. Salt Lake City (UT): University of In conclusion, the results of this study point out that Utah Health Sciences Center, 1995, https://www.ncbi. standardization is still of high importance concerning nlm.nih.gov/books/NBK11530/ (accessed 23 June 2017). conducting experimental studies in mice and helps us to 8. Jacobs GH. Losses of functional opsin genes, short-wave- fill in another stone into the big mosaic ‘‘mouse.’’ In length cone photopigments, and color vision—A signifi- order to fulfil the requirements of the 3Rs, all research- cant trend in the evolution of mammalian vision. Vis ers in this field are summoned to contribute, in order to Neurosci 2013; 30: 39–53. enhance the comfort for the used animals and within 9. Smallwood PM, Olveczky BP, Williams GL, et al. that to satisfy not only the needs but also the demands Genetically engineered mice with an additional class of cone photoreceptors: Implications for the evolution of of experimental animals. To fundamentally contribute color vision. Proc Natl Acad Sci U S A 2003; 100: to refinement of laboratory animal experimentation, 11706–11711. the characterization of laboratory rodents with specific 10. Brown RE and Wong AA. The influence of visual ability relations to experimental applications should be pur- on learning and memory performance in 13 strains of sued and in light of ‘‘good scientific practice,’’ future mice. Learn Mem 2007; 14: 134–144. protocols (such as Animal Research: Reporting of In 11. Anderson MG, Libby RT, Mao M, et al. Genetic context Vivo Experiments checklists) should integrate meth- determines susceptibility to intraocular pressure elevation odological aspects like the food color provided. in a mouse pigmentary glaucoma. BMC Biol 2006; 4: 20. 12. Moon JI, Kim IB, Gwon JS, et al. Changes in retinal neuronal populations in the DBA/2J mouse. Cell Tissue Acknowledgment Res 2005; 320: 51–59. The authors are thankful to Prof A. Schu¨ rmann, Department 13. Gime´ nez E and Montoliu L. A simple polymerase chain of Experimental Diabetology, German Institute of Human reaction assay for genotyping the retinal degeneration Nutrition, for providing the necessary support and measuring mutation (Pdeb(rd1)) in FVB/N-derived transgenic devices for this project. mice. Lab Anim 2001; 35: 153–156. 14. Hart AW, McKie L, Morgan JE, et al. Genotype-pheno- Declaration of Conflicting Interest type correlation of mouse pde6b mutations. Invest Ophthalmol Vis Sci 2005; 46: 3443–3450. The author(s) declared no potential conflicts of interest with 15. Yeritsyan N, Lehmann K, Puk O, et al. Visual capabil- respect to the research, authorship, and/or publication of this ities and cortical maps in BALB/c mice. Eur J Neurosci article. 2012; 36: 2801–2811. 16. Beermann F, Orlow SJ and Lamoreux ML. The Tyr Funding (albino) locus of the laboratory mouse. Mamm Genome The author(s) received no financial support for the research, 2004; 15: 749–758. authorship, and/or publication of this article. 17. Donatien P and Jeffery G. Correlation between rod photoreceptor numbers and levels of ocular pigmentation. Invest Ophthalmol Vis Sci 2002; 43: References 1198–1203. 1. Castelhano-Carlos MJ and Baumans V. The impact of 18. Metz GA and Schwab ME. Behavioral characterization light, noise, cage cleaning and in-house transport on wel- in a comprehensive mouse test battery reveals motor and fare and stress of laboratory rats. Lab Anim 2009; 43: sensory impairments in growth-associated protein-43 null 311–327. mutant mice. Neuroscience 2004; 129: 563–574. 2. Streba CT, Vere CC, Miscu C, et al. Review of mice and 19. Pinto LH and Enroth-Cugell C. Tests of the mouse visual ethics. Curr Heal Sci J 2012; 38: 5–8. system. Mamm Genome 2000; 11: 531–536. 3. Merrow M, Spoelstra K and Roenneberg T. The circadian 20. Wang CH and Liao JK. A mouse model of diet-induced cycle: Daily rhythms from behaviour to genes. EMBO Rep obesity and insulin resistance. Methods Mol Biol 2012; 2005; 6: 930–935. 821: 421–433. Nitezki et al. 619

21. Won J, Shi LY, Hicks W, et al. Mouse model resources 25. Woods SC, Seeley RJ, Porte D Jr, et al. Signals that for vision research. J Ophthalmol 2011; 2011: 391384. regulate food intake and energy homeostasis. Science 22. Wong AA and Brown RE. Age-related changes in visual 1998; 280: 1378–1383. acuity, learning and memory in C57BL/6J and DBA/2J 26. Woods SC. Signals that influence food intake and body mice. Neurobiol Aging 2007; 28: 1577–1593. weight. Physiol Behav 2005; 86: 709–716. 23. Fox WM. Reflex-ontogeny and behavioural development 27. Kawakami K, Xiao B, Ohno R, et al. Color preferences of the mouse. Anim Behav 1965; 13: 234–241. of laboratory mice for bedding materials: Evaluation 24. Rowe FP, Bradfield A and Redfern R. Food preferences using radiotelemetry. Exp Anim 2012; 61: 109–117. of wild house-mice (Mus musculus). J Hyg (Lond) 1974; 73: 473–478.

Re´sume´ Concernant la normalisation des e´levages d’animaux de laboratoire, seuls des changements mineurs de leur habitat peuvent potentiellement influencer la physiologie des animaux ou les re´sultats des tests comporte- mentaux. Habituellement, la nourriture des souris show est colore´e en fonction des diffe´rents types de re´gimes administre´s. E´tant donne´ que la couleur n’a aucun effet sur l’odeur ou le gouˆt des aliments, nous avons souhaite´ ve´rifier l’hypothe`se selon laquelle la couleur des aliments a un impact sur la quantite´ consomme´e par les souris. Des souris maˆles aˆge´s de 12 semaines issus de diffe´rentes souches (C57BL/6J, DBA/2J, C3H/HeJ, BALB/ cJ; n ¼ 12/souche) ont e´te´ he´berge´s individuellement dans des cages PhenoMasterÕ. Apre`s une phase d’accli- matation, des aliments normaux de couleurs diffe´rentes ont e´te´ administre´s. La consommation alimentaire a e´te´ mesure´e dans le cadre d’un test permettant aux souris de choisir entre deux combinaisons de couleurs diffe´rentes. Tous les animaux ont consomme´ en moyenne 3 g de nourriture par jour et aucune pre´fe´rence n’a e´te´ remarque´e lorsqu’une combinaison d’aliments de couleur identique e´tait offerte. Les tests de pre´fe´rence ont re´ve´le´ une forte aversion aux aliments de couleur bleue et une attirance importante envers les aliments de couleurs jaune et verte chez les souris C57BL/6 et DBA/2J. Chez les souris C3H/HeJ et BALB/cJ, aucune pre´fe´rence base´e sur les couleurs n’a e´te´ observe´e. Les ligne´es de souris se´lectionne´es pre´sentent des diffe´rences connues en ce qui concerne la fonctionna- lite´ de leur sens visuel. Il est conside´re´ que les souris C57BL/6 et DBA/2 posse`dent une vue normale au moment du test. La ligne´e BALB/c repre´sente les souches de souris albinos et les souris C3H sont porteuses de mutations entraıˆnant des modifications de la re´tine. Les re´sultats sugge`rent que les souris posse´dant une vue normale sont se´lectives en ce qui concerne la couleur des aliments lorsqu’on leur donne le choix. De manie`re ge´ne´rale, ceci n’influence toutefois pas la quantite´ de nourriture consomme´e lorsque les animaux rec¸oivent uniquement des aliments ne pre´sentant qu’une seule couleur. Par ailleurs, les souris malvoyantes n’ont affiche´ aucune pre´fe´rence alimentaire associe´e aux couleurs.

Abstract Bei der Standardisierung der Labortierhaltung ko¨nnen schon geringfu¨gige Vera¨nderungen des Habitats die Physiologie des Tieres oder die Ergebnisse von Verhaltenstests beeinflussen. Routinema¨ßig wird das Futter von Ma¨usen gefa¨rbt, wenn verschiedene Arten von Dia¨ten verabreicht werden. Angesichts der Tatsache, dass der Farbstoff selbst keine Auswirkungen auf den Lebensmittelgeruch oder -geschmack hat, wollten wir die Hypothese testen, dass die Futterfarbe einen Einfluss auf die Nahrungsaufnahme bei Ma¨usen hat. 12 Wochen alte ma¨nnliche Ma¨use verschiedener Sta¨mme (C57BL/6J, DBA/2J, C3H/HeJ, BALB/cJ; n ¼ 12/ Stamm) wurden einzeln in PhenoMasterÕ Ka¨figen untergebracht. Nach der Akklimatisierung wurde Standard- Ma¨usefutter in verschiedenen Farben verabreicht. Die Nahrungsaufnahme wurde als ein Zwei-Alternativen- Wahltest verschiedener Farbkombinationen u¨berwacht. Alle Tiere nahmen durchschnittlich 3 g/Tag Nahrung auf und es wurden keine Pra¨ferenzen beobachtet, wenn eine Kombination von gleichfarbigen Futtermitteln angeboten wurde. Pra¨ferenztests ergaben eine signifikante Abneigung gegen blaues Futter und eine signifikante Vorliebe fu¨r gelbes und gru¨nes Futter bei C57BL/6- und DBA/2J-Ma¨usen. Bei C3H/HeJ- und BALB/cJ-Ma¨usen waren keine farbbezogenen Muster erkennbar. Ausgewa¨hlte Sta¨mme von Ma¨usen weisen bekanntermaßen Unterschiede in der Funktionalita¨t ihres Sehsinns auf. C57BL/6- und DBA/2-Ma¨use gelten im Testalter als normalsichtig, BALB/c sind repra¨sentativ 620 Laboratory Animals 52(6) fu¨r Albino-Sta¨mme und C3H-Ma¨use sind von Mutationen betroffen, die zu Netzhautvera¨nderungen fu¨hren. Die Ergebnisse legen nahe, dass normalsichtige Ma¨use selektiv in Bezug auf die Futterfarbe sein du¨rften, sofern sie die Wahl haben. Dies hat jedoch keinen Einfluss auf die Gesamtmenge der Nahrungsaufnahme, wenn die Tiere ausschließlich mit durch einen einzigen Farbstoff gefa¨rbtem Futter versorgt wurden. Außerdem zeigten sehbehinderte Ma¨use keine farbbezogenen Futtervorlieben.

Resumen Respecto a la estandarizacio´n de la crı´a de animales de laboratorio u´nicamente los cambios exiguos del ha´bitat pueden influir potencialmente en la fisiologı´a del animales o en los resultados de los tests de comportamiento. De forma rutinaria, se tin˜e la comida de los ratones al suministrar distintos tipos de dietas. Considerando el hecho que el tinte en sı´ no tiene ningu´n efecto en el sabor o color de la comida, quisimos probar la hipo´tesis de que el color de la comida tiene un impacto en la ingesta de comida entre los roedores. Ratones macho de 12 semanas de diferentes cepas (C57BL/6J, DBA/2J, C3H/HeJ, BALB/cJ; n ¼ 12/cepa) fueron enjaulados en jaulas PhenoMasterÕ. Tras la aclimatacio´n, se administro´ comida esta´ndar para roe- dores de distintos colores. Se controlo´ la ingesta de comida como una prueba de eleccio´n entre dos alter- nativas de distintas combinaciones de colores. Todos los animales tenı´an una ingesta de comida media de 3 g/d y no se observo´ ninguna preferencia al ofrecer una combinacio´n de distintos alimentos del mismo color. Los tests de preferencia demostraron una gran aversio´n por la comida de color azul y una atraccio´n significativa hacia alimentos amarillos y verdes en los ratones C57BL/6 y DBA/2J. En ratones C3H/HeJ y BALB/cJ no ocurrio´ ningu´n patro´n relacionado con el color. Distintas cepas seleccionadas de ratones han experimentado diferencias en cuanto a la funcionalidad de su sentido visual. Los ratones C57BL/6 y DBA/2 se consideraron como con vista normal en la edad de prueba, BALB/c representa cepas albinas y C3H ratones con mutaciones que provocan alternaciones de la retina. Los resultados sugieren que los ratones con vista normal seleccionarı´an la comida segu´n el color si tuvieran la oportunidad de hacerlo. De todos modos, esto no influye en la cantidad total de ingesta de comida cuando se dio a los animales comida de un mismo color. Asimismo, los ratones con problemas de visio´n no mostraron ninguna preferencia relacionada con el color de la comida. Original Article Laboratory Animals 2018, Vol. 52(6) 621–629 ! The Author(s) 2018 Four simple ways to increase power Article reuse guidelines: sagepub.com/journals- without increasing the sample size permissions DOI: 10.1177/0023677218767478 journals.sagepub.com/home/lan Stanley E Lazic

Abstract Underpowered experiments have three problems: true effects are harder to detect, the true effects that are detected tend to have inflated effect sizes and as power decreases so does the probability that a statistically significant result represents a true effect. Many biology experiments are underpowered and recent calls to change the traditional 0.05 significance threshold to a more stringent value of 0.005 will further reduce the power of the average experiment. Increasing power by increasing the sample size is often the only option considered, but more samples increases costs, makes the experiment harder to conduct and is contrary to the 3Rs principles for animal research. We show how the design of an experiment and some analytical decisions can have a surprisingly large effect on power.

Keywords experimental design, power, reproducibility, sample size, statistics

Date received: 26 October 2017; accepted: 5 March 2018

Introduction 0.005 because the traditional 0.05 threshold provides little evidence against the null hypothesis.9–13 If jour- Statistical power is the probability of detecting a true nals start requiring a lower threshold for significance, effect, and therefore when experiments have low power, the power of all experiments will be further reduced, a true effect or association is hard to find.1,2 In add- exacerbating the above problems. By way of example, ition, with low power, small effects will only be statis- suppose we are conducting a two-group experiment tically significant when the effect size is overestimated with independent samples in each group and want (or if the within-group SD is underestimated).3–5 In 80% power to detect a difference of 1.25 (arbitrary) other words, the correct qualitative conclusion might units. Assume the within-group SD is 1 and the signifi- be reached (there is an effect) but the magnitude or cance threshold is 0.05. We require 11 samples per strength of the effect is overestimated.4 A final and group, or 22 in total. If the significance threshold is less appreciated consequence of lower power is that as reduced to 0.005, 38 total samples are required: a power decreases, so does the probability that a statis- 71% increase in sample size. If only 22 samples are tically significant result represents a true effect.2,6–8 The used with the more stringent 0.005 significance thresh- probability that a significant result is a true effect is old, the power of the experiment is only 44%. often called the positive predictive value and is calcu- To increase power, many researchers only consider lated as the number of true positive results divided by increasing the sample size (N). Indeed, standard power the total number of positive results (the sum of true and sample size calculations in textbooks and review positive and false positive results). Since true positives are harder to detect as power decreases, the probability Innovative Medicines and Early Development Biotech Unit, that p < 0.05 represents a true positive also decreases AstraZeneca, UK (see Wacholder et al. and Ioannidis for further details).6,7 Thus, lower power has a triple-negative Corresponding author: Stanley E Lazic, Quantitative Biology, Discovery Sciences, effect on statistical inference. Innovative Medicines and Early Development Biotech Unit, Statisticians and scientists have recently been AstraZeneca, Cambridge, CB4 0WG, UK. arguing for a more stringent significance threshold of Email: [email protected] 622 Laboratory Animals 52(6) articles suggest that the only option to increase power is Trade-offs are inevitably required when planning an to increase the number of samples. The structure of the experiment, and defining a key question or hypothesis design of the experiment is taken as given, the probabil- enables the experiment to be designed to maximise the ity of a Type I error (false positive) is set to ¼ 0:05, chance of success, but at the cost of being unable to and the power is usually fixed to 80% or 90%. The address other questions. Hence, the first step is to effect size is the smallest effect that is biologically or clearly define the key question that the experiment clinically meaningful and that the researcher would will answer. This may sound trite, but consider an like to detect, and the within-group SD is derived experiment testing a new compound in an animal from pilot studies, published results or an educated model of a disease. The key questions could be: guess (in practice, and not recommended, the effect size and SD are often adjusted to give the sample size 1. Is the compound active (does it work at all)? that investigators were planning to use all along).14,15 2. Is the compound active at a specific dose? This only leaves N to be adjusted to meet the desired 3. What is the minimum effective dose? power. We assume, as is standard for most traditional 4. What dose gives half of the maximum response (the frequentist analyses, that the sample size is determined Effective Dose 50% or ED50)? before the experiment is conducted and cannot be chan- 5. What is the relationship between dose and response? ged during the experiment. This inflexible approach is 6. Is one dose better than another? inefficient because experiments cannot be stopped early if a definitive conclusion is reached based on a prede- No design is optimal for answering all of these ques- fined criterion. Although methods have been developed tions; some designs are better for some questions and for the analysis of such ‘adaptive’ designs, they are not other designs are better for other questions. Once the considered here.16–18 question or hypothesis is defined, the four points below Increasing the sample size makes the experiment can be used to plan the experiment. more expensive, harder to conduct, and has ethical implications for animal experiments. Also, it is often Use fewer factor levels for continuous difficult to hold everything else constant while increas- predictors ing N. A small experiment could be run all at once, but a larger experiment may need to be run in smaller In experiments with continuous predictors such as dose, batches: perhaps on two separate days, or by two concentration, time, pH, temperature, pressure, illu- researchers. ‘Day’ and ‘researcher’ are technical factors mination level and so on, how do we decide on the that could influence the outcome and may need to be minimum and maximum levels, the number of points included as variables in the analysis. If so, the design of in between, and the specific levels where observations the experiment now differs from the design used for the will be made or samples allocated? Some choices are power calculation. Similarly, data from a small experi- easy, for example, when a minimum value of zero ment may be collected over a short period of time (e.g. serves as a control condition. The maximum value is 1 or 2 hours), making circadian effects negligible. A also often easy to choose as it is the upper limit that we larger experiment may need to collect data over a are interested in testing or learning about. But what longer period, and now circadian effects may become about the points between the minimum and maximum? more pronounced. The design of the experiment now The places where observations are made (minimum and needs to change to accommodate the circadian effects; maximum plus intermediate points) are called the for example, by using time as a covariate or blocking design points.20 factor.19 This is again a different experiment to the one To illustrate how the number of design points affects used for the power calculation. the power, we compare four designs with 2–5 equally Simple options are available to increase power – spaced design points (experimental groups) and a fixed often dramatically – while keeping the sample size total sample size of 20. Assume that the dose of a com- fixed. Or coming from the other direction, certain pound is the factor of interest, which ranges from min- design and analytical options should be avoided to pre- imum ¼ 0 to maximum ¼ 100. 10,000 data sets were vent loss of power. Four options are described below simulated from the true model (Figure 1(a)), which that apply to many biological experiments and other has a maximum response of 40 at dose ¼ 0, and a min- options for increasing power are described in Lazic;19 imum response of 27 at dose ¼ 100. The variability of for example, ensuring that predictor variables are the data is shown by the scatter of points around the uncorrelated, optimising the spatial and temporal spa- dose-response line (SD ¼ 9). Data were simulated under cing of observations, and using blocking and covariates each design and analysed with a one-way analysis of to reduce noise. The R code for the examples shown in variance (ANOVA), testing the first question above: this paper is provided as supplementary material. ‘is the compound active at all’? Continuous predictor Lazic 623

Figure 1. Effect of increasing the number of groups. A fictitious example of the relationship between an outcome and the dose of a compound (a). Four designs with the same sample size but with observations taken at different doses are indicated with black circles. For example, design 1 has 10 samples at doses 0 and 100. Increasing the number of groups from 2 to 5 decreases the power for the overall analysis of variance (ANOVA) from 84 to 40% (b). Testing a more specific hypothesis for a trend has improved power compared with an ANOVA analysis, but still loses power with more groups. variables such as dose are often better analysed with be detected, it lowers the probability of detecting a linear or nonlinear regression methods,21 especially linear relationship if that is indeed the correct one. when the form of the relationship is known from If the aim of the study is to answer the second ques- theory or background knowledge. A one-way tion (‘is the compound active at a specific dose?’), then ANOVA is used instead because it is the most only the doses in question and the control group are common approach in many fields and because it can required. Given this research question, a small increase be used for all designs, including those not considered in power can be achieved by shifting a few samples here. from the treated groups to the control group, compared The power for each design is calculated as the pro- with an equal number of samples in each group (see portion of significant results from the ANOVA F-test. Bate and Karp22). Questions 3–5 are better addressed Despite the same sample size, the power of these experi- with a linear or nonlinear regression model as the min- ments differs greatly (Figure 1(b), ‘ANOVA line’). The imum effective dose or the ED50 are unlikely to be at power of design 1 with two groups is 84% and steadily one of the doses used in the experiment. As nonlinear decreases to 40% with design 4. models interpolate between the observed design points, It is clear that using two groups maximises power. they can provide an estimated value anywhere in the But what if the true relationship is not sigmoidal but a design space. Nonlinear models can benefit from having ‘U’, inverted-‘U’ or some other complex shape? A more than the five design points used here and are often design with two groups would be unable to detect in the 8–10 range.23,24 Addressing question 6 (‘is one such relationships, but if a linear or monotonic rela- dose better than another?’) does not even require the tionship is expected, then one additional design point control group, although controls inform us about the can allow departures from the assumed relationship to state of the experimental system and should be be detected (e.g. design 2 in Figure 1).20 Trade-offs are retained.25 Nevertheless, the control group could have always necessary when designing experiments; while an fewer samples, which are then transferred to the treated additional group allows a more complex relationship to groups to increase the power of the comparisons 624 Laboratory Animals 52(6) between the treated groups. This decreases the power of high). This practice is common, despite the many treated versus control group comparisons, illustrating papers warning against it. Dichotomising variables again how the aim of the experiment or the question reduces power, can bias estimates and can increase asked influences the design. Berger and Wong discuss a false positives.21,26–39 wider range of designs and how they affect power for This recommendation does not contradict the first different types of relationships.20 point about using fewer factor levels for continuous ANOVA analyses are often followed by post hoc variables because, when designing an experiment, we tests comparing group means. Power also decreases can only take observations at discrete points within for post hoc comparisons as the number of groups the design space, and so there will be some discretising. increases. For example, the power for the 0 versus high- But a continuous variable that we record – either an est dose group comparison in design 5 is only 45%, and outcome or a predictor – can take any value. The argu- this does not adjust for multiple comparisons, which ment here is that if you have the continuous values, the would reduce power further. The power for the 0 costs outweigh the benefits of converting it into a cat- versus highest dose group in design 1 is 85%, the egorical variable. same as the overall ANOVA, because there are only To illustrate this point, Figure 2 shows data for 50 two groups and so a post hoc test is unnecessary. samples that have a correlation of 0.4. These are nat- A final point to maximise power is ensure that the urally analysed with a Pearson correlation or linear predictor variable covers a wide enough range; for regression (solid line), both of which give p ¼ 0.002 example, the compound would appear to have no for the association. If the variables are dichotomised effect if the maximum dose tested in the experiment at their median (dashed lines), the number of data was 30. points falling in each quadrant can be counted (num- bers in grey boxes), forming a 2 2 table. This Use a focused hypothesis test doubled-dichotomised data is commonly analysed with a 2 test or Fisher’s exact test, and the 2 test A second way to increase power is to test a specific gives p ¼ 0.024, almost 10 times larger than the regres- hypothesis instead of a general ‘are there any differ- sion analysis. To estimate the reduction power when ences?’ hypothesis. For the simulated example above, using the 2 test, 10,000 data sets like the one in the ANOVA analysis can detect any pattern of differ- ences between groups, but the trade-off is that it has less power to detect any specific pattern. If we expect the outcome to either steadily increase or decrease as the dose increases, then a focused test of this pattern is more powerful (Figure 1(b), ‘trend test’ line). With two groups, the power of the trend test is iden- tical to the ANOVA analysis (both are equivalent to a t-test), but with five groups (design 4) the power of the trend test is 66%, compared with 40% for the ANOVA analysis. Trend tests are available in most statistical packages (even if they are not labelled as such) and usually require that the predictor variable – dose in this case – is treated as an ordered categorical factor. The output from such an analysis should include the test for a linear trend, often called the ‘linear contrast’ or similar. Alternatively, treating dose as a continuous variable and analysing the data with a regression ana- lysis instead of an ANOVA is another option that has high power as well as other advantages.21

Don’t dichotomise or bin continuous Figure 2. Analysis of continuous and binned data. A variables regression analysis (black line) gives p ¼ 0.002. Binning the data with a bivariate median split (dashed lines) and ana- Dichotomising or binning refers to taking a continuous lysing the number of samples in each quadrant (grey variable – either an outcome or predictor variable – and boxes) with a 2 test gives p ¼ 0.024. The power of the 2 reducing it to a categorical variable based on a thresh- test is only 34%, compared with 84% for the regression old such as the median (e.g. low/high or low/medium/ analysis. Lazic 625

Figure 2 were simulated with N ¼ 50 and a correlation Suppose that we have 16 mice (assume all male for of 0.4. Analysing the continuous values with a regres- simplicity) from four litters (A–D) and we want to sion analysis had 84% power, while binning and using test the effect of a compound at a single dose. the 2 test reduced power to 34%. Assume that each mouse can be randomly and indi- Data are sometimes binned because the relationship vidually assigned to one of the two treatment groups. between the predictor and the outcome variable is non- Although this is a simple two-group design, we might linear and a simple regression or correlation analysis expect differences between litters and want to take this would be unsuitable. Binning the data is still a poor into account in the design. option and frequently a simple alternative is available. When all animals from a litter are in the same con- For example, Spearman’s rank correlation tests for a dition, the factor litter is said to be nested under the monotonic relationship between two variables, and factor treatment (Figure 3, left). When animals from a there is no requirement for a linear relationship or litter are spread across both treatment groups, the assumptions about the distribution of the variables. treatment and litter factors are crossed (Figure 3, right). Alternatively, a ‘U’, inverted-‘U’ relationship between The main point here is that the nested arrangement the variables might be modelled with a quadratic term is much less powerful than the crossed one. Figure 4 in a regression model, which is available in all statistical illustrates the difference in power between the two software. The take home message of this section is that designs. The litter means are drawn from a normal dis- it is almost always preferable to analyse a continuous tribution with an SD of 3, and the values for the indi- variable directly instead of reducing it to a categorical vidual animals are drawn from a normal distribution variable. with an SD of 0.5. Thus, the litter-to-litter variation is large relative to the variation between animals within a Cross your factors, don’t nest them litter. Figure 4(a) shows the data before the application of a treatment, and an immediate danger can be seen The levels of an experimental factor are either (a) set by with a nested design: if litters A and B end up in one the researcher, (b) a property of the samples or (c) a group and C and D in the other, large differences technical aspect of how the experiment is conducted. between groups exist at baseline. Hence, nested designs For example, the dose of a drug that an animal receives can also potentially lead to more false positive is controlled by the researcher, while the sex of the findings.40,41 animal is a property of the animal. If the experiment Figure 4(b) shows one possible randomisation of a is conducted over multiple days or by multiple research- nested design, where litters A and D end up in the con- ers, then day and researcher are technical factors. trol group and B and C in the treated group. The effect Factor arrangement refers to how multiple factors of the treatment in this example is to decrease the out- relate to each other within the experimental design, come by 2 units (note how the control group values of and there are three possibilities. black O’s and green ’s are identical in Figure 4(a) and When two factors are completely confounded, levels (b)). Analysing the data with a t-test gives a p-value of of one factor always co-occur with the same levels of 0.14. However, since the litters were randomised to the the other factor; for example, if all the control animals treatment conditions and not the individual animals, are run on the first day and all the treated animals are the litter is a more appropriate experimental unit, that run on the second day. Confounding a treatment effect is, the sample size is four litters, not 16 animals.19,42 that we are interested in testing with an uninteresting One way to conduct this analysis is to calculate the technical effect is never a good idea because it is impos- litter average and use these values for a standard stat- sible to attribute any differences between treatment istical test, which gives p ¼ 0.57. Figure 4(c) shows one groups to the effect of the treatment; differences may possible randomisation of the crossed design, with the have arisen from the day-to-day variation. To conclude same effect size of 2 units. Analysis of this design, anything about a treatment effect we would have to which includes litter as a variable in the model, gives assume that the effect of day is zero, and therefore p ¼ 0.0007 for the effect of the treatment. this arrangement should be avoided. To calculate the power of the two designs and both The second possibility is a crossed or factorial analyses of the nested design, 10,000 data sets were arrangement, which occurs when all levels of one generated with the above characteristics. The nested factor co-occur with all levels of another factor and is design that treats animals as the experimental unit has the most common arrangement in experimental biol- 55% power, and treating the litter as the experimental ogy. The final possibility is a nested arrangement, unit (the more appropriate analysis) has only 7% where levels of one factor are grouped or nested power. The power of the crossed design is over 99%. under the levels of another factor. Figure 3 shows the These large differences in power exist because the difference between crossed and nested arrangements. nested design compares the effect size of 2 units 626 Laboratory Animals 52(6)

Figure 3. Crossed versus nested designs. An experiment with 16 mice from four litters (A–D) can be randomised to treatment groups either by litter, leading to a nested design (left), or within litter, leading to a crossed design (right).

Figure 4. Simulated data for a nested and crossed design. Natural variation between and within litters (a). The nested design compares the effect of the treatment against the high variation between litters, leading to a large p-value (0.14 or 0.57, depending on the analysis) (b). The crossed design compares the treatment effect to the smaller within-litter variation, leading to a much smaller p-value of 0.0007 (c). Lazic 627 against the high litter-to-litter variation (SD ¼ 3), Funding whereas the crossed design compares the effect size The author(s) received no financial support for the research, against the much smaller variation of animals within authorship and/or publication of this article. a litter (SD ¼ 0.5). With the crossed design, the test for a treatment effect is performed within the litters and hence large differences between litters are irrelevant ORCID iD for testing treatment effects. Litter is used here as a Stanley E Lazic http://orcid.org/0000-0002-3409-3407 ‘blocking variable’ to remove the unwanted litter-to- litter variation.40,43 References The difference in power between nested and crossed designs become less pronounced as litter effects get 1. Maxwell SE. The persistence of underpowered studies in psychological research: Causes, consequences, and reme- smaller, animal variation within a litter gets larger or dies. Psychol Methods 2004; 9: 147–163. both. The nested design should be avoided because of 2. Button KS, Ioannidis JP, Mokrysz C, et al. Power failure: low power, more false positives and ambiguity in defin- Why small sample size undermines the reliability of 19 ing the sample size . Unfortunately, nested designs are neuroscience. Nat Rev Neurosci 2013; 14: 365–376. common because they are often easier to conduct and 3. Ioannidis JPA. Why most discovered true associations manage; for example, it’s easier to randomise litters to are inflated. Epidemiology 2008; 19: 640–648. treatment groups as littermates can be housed together 4. Gelman A and Carlin J. Beyond power calculations: in the same cage. Assessing type S (Sign) and type M (Magnitude) errors. Other technical variables or properties of subjects Perspect Psychol Sci 2014; 9: 641–651. such as batches, microtitre plates, cages or other hous- 5. McShane BB and Bockenholt U. You cannot step into ing structure, body weight, day and experimenter can the same river twice: When power analyses are optimistic. also be nested or crossed with the treatment effects of Perspect Psychol Sci 2014; 9: 612–625. 6. Wacholder S, Chanock S, Garcia-Closas M, et al. interest. These factors need to be carefully arranged to Assessing the probability that a positive report is false: ensure high power and no confounding. The points dis- An approach for molecular epidemiology studies. J Natl cussed above can combine to reduce power even fur- Cancer Inst 2004; 96: 434–442. ther; for example, using a nested design, dichotomising 7. Ioannidis JPA. Why most published research findings are a variable and testing a general hypothesis will have false. PLoS Med 2005; 2: e124. a dramatic loss of power compared with a crossed 8. Nuzzo R. Scientific method: Statistical errors. Nature design, without dichotomisation and testing a focused 2014; 506: 150–152. hypothesis. 9. Berger JO and Sellke T. Testing a point null hypothesis: The irreconcilability of p values and evidence. J Am Stat Assoc 1987; 82: 112–122. Conclusion 10. Sellke T, Bayarri MJ and Berger JO. Calibration of p Low power continues to undermine many biology values for testing precise null hypotheses. Am Stat 2001; experiments, but a few simple alterations to a design 55: 62–71. 11. Johnson VE. Revised standards for statistical evidence. or analysis can dramatically increase the information Proc Natl Acad Sci USA 2013; 110: 19313–19317. obtained without increasing the sample size. In the 12. Singh Chawla D. Big names in statistics want to shake up interest of minimising animal usage and reducing much-maligned P value. Nature 2017; 548: 16–17. 44,45 waste in biomedical research, researchers should 13. Benjamin DJ, Berger JO, Johannesson M, et al. Redefine aim to maximise power by designing confirmatory statistical significance. Nat Hum Behav 2018; 2: 6–10. experiments around key questions, use focused hypo- 14. Schulz KF and Grimes DA. Sample size calculations in thesis tests, and avoid dichotomising and nesting, randomised trials: Mandatory and mystical. Lancet 2005; which ultimately reduce power and provide no other 365: 1348–1353. benefits. 15. Bacchetti P. Current sample size conventions: Flaws, harms, and alternatives. BMC Med 2010; 8: 17. Acknowledgements 16. Berry SM, Carlin BP, Lee J, et al. Bayesian adaptive methods for clinical trials. Boca Raton: CRC Press, 2010. I would like to thank Deborah Mayo, Frank Harrell and 17. Schonbrodt FD, Wagenmakers EJ, Zehetleitner M, et al. other readers who commented on the pre-print version of Sequential hypothesis testing with Bayes factors: this article on bioRxiv or social media. Efficiently testing mean differences. Psychol Methods 2017; 22: 322–339. Declaration of Conflicting Interests 18. Schonbrodt FD and Wagenmakers EJ. Bayes factor The author(s) declared no potential conflicts of interest with design analysis: Planning for compelling evidence. respect to the research, authorship and/or publication of this Psychon Bull Rev. Epub ahead of print 1 March 2017. article. DOI: 10.3758/s13423-017-1230-y. 628 Laboratory Animals 52(6)

19. Lazic SE. Experimental design for laboratory biologists: 33. Maxwell S and Delaney H. Bivariate median splits and Maximising information and improving reproducibility. spurious statistical significance. Quant Methods Psychol Cambridge: Cambridge University Press, 2016. 1993; 113: 181–190. 20. Berger MPF and Wong WK. An introduction to optimal 34. Owen SV and Froman RD. Why carve up your continu- designs for social and biomedical research. Chichester: ous data? Res Nurs Health 2005; 28: 496–503. John Wiley & Sons, 2009. 35. Royston P, Altman DG and Sauerbrei W. Dichotomizing 21. Lazic SE. Why we should use simpler models if the data continuous predictors in multiple regression: A bad idea. allow this: Relevance for ANOVA designs in experimen- Stat Med 2006; 25: 127–141. tal biology. BMC Physiol 2008; 8: 16. 36. Taylor J and Yu M. Bias and efficiency loss due to cate- 22. Bate S and Karp NA. A common control group - gorizing an explanatory variable. J Multivar Anal 2002; optimising the experiment design to maximise sensitivity. 83: 248–263. PLoS One 2014; 9: e114872. 37. van Walraven C and Hart RG. Leave ’em alone - why 23. Motulsky H and Christopoulos A. Fitting models to bio- continuous variables should be analyzed as such. logical data using linear and nonlinear regression: A prac- Neuroepidemiology 2008; 30: 138–139. tical guide to curve fitting. Oxford: Oxford University 38. Wainer H, Gessaroli M and Verdi M. Finding what is not Press, 2004. there through the unfortunate binning of results: The 24. Ritz C and Streibig JC. Nonlinear regression with R. New Mendel effect. Chance 2006; 19: 49–52. York: Springer-Verlag, 2008. 39. Kuss O. The danger of dichotomizing continuous vari- 25. Glass DJ. Experimental design for biologists. Cold Spring ables: A visualization. Teach Stat 2013; 35: 78–79. Harbor: Cold Spring Harbor Laboratory Press, 2007. 40. Lazic SE and Essioux L. Improving basic and transla- 26. Barnwell-Menard J-L, Li Q and Cohen AA. Effects of cat- tional science by accounting for litter-to-litter variation egorization method, regression type, and variable distribu- in animal models. BMC Neurosci 2013; 14: 37. tion on the inflation of type-I error rate when categorizing 41. Lazic SE. Comment on ‘‘stress in puberty unmasks latent a confounding variable. Stat Med 2015; 34: 936–949. neuropathological consequences of prenatal immune acti- 27. Bennette C and Vickers A. Against quantiles: vation in mice’’. Science 2013; 340: 811. Categorization of continuous variables in epidemiologic 42. Lazic SE, Clarke-Williams CJ and Munafo MR. What research, and its discontents. BMC Med Res Methodol exactly is ‘N’ in cell culture and animal experiments? 2012; 12: 21. bioRxiv 2018; (in press). DOI: 10.1101/183962. 28. Chen H, Cohen P and Chen S. Biased odds ratios from 43. Festing MFW. Randomized block experimental designs dichotomization of age. Stat Med 2007; 26: 3487–3497. can increase the power and reproducibility of laboratory 29. Cohen J. The cost of dichotomization. Appl Psychol Meas animal experiments. ILAR J 2014; 55: 472–476. 1983; 7: 249–253. 44. Macleod MR, Michie S, Roberts I, et al. Biomedical 30. Fedorov V, Mannino F and Zhang R. Consequences of research: Increasing value, reducing waste. Lancet 2014; dichotomization. Pharm Stat 2009; 8: 50–61. 383: 101–104. 31. Irwin J and McClelland G. Negative consequences of 45. Ioannidis JPA, Greenland S, Hlatky MA, et al. dichotomizing continuous predictor variables. J Mark Increasing value and reducing waste in research design, Res 2003; 40: 366–371. conduct, and analysis. Lancet 2014; 383: 166–175. 32. Kenny PW and Montanari CA. Inflation of correlation in the pursuit of drug-likeness. J Comput Aided Mol Des 2013; 27: 1–13.

Re´sume´ Les expe´riences dont la puissance est insuffisante posent trois proble`mes : les ve´ritables effets sont plus difficiles a` de´tecter, les ve´ritables effets de´tecte´s ont tendance a` avoir des valeurs d’effet gonfle´es, et au fur et a` mesure que la puissance diminue, la probabilite´ qu’un re´sultat statistiquement significatif repre´sente un ve´ritable effet diminue e´galement. De nombreuses expe´riences biologiques ont une puissance insuffisante et de re´cents appels a` changer le seuil de signification traditionnel de 0,05 en le remplac¸antpar une valeur plus rigoureuse de 0,005 permettra de re´duire davantage la puissance de l’expe´rience moyenne. Augmenter la puissance en augmentant la taille de l’e´chantillon est souvent la seule option envisage´e, mais augmenter les e´chantillons signifie augmenter les couˆts, rend l’expe´rience plus difficile a` mener, et est contraire aux principes des 3R en recherche animale. Nous montrons comment la conception d’une expe´rience et certaines de´cisions d’analyse peuvent avoir un effet important sur la puissance.

Abstract Ungenu¨gend leistungsstarke Experimente sind von drei Problemen gekenzeichnet: Tatsa¨chliche Effekte sind schwerer zu erkennen, die Gro¨ße tatsa¨chlicher Effekte, die erkannt werden, ist ha¨ufig u¨bertrieben, und mit abnehmender Sta¨rke nimmt auch die Wahrscheinlichkeit ab, dass ein statistisch signifikantes Ergebnis einen Lazic 629 tatsa¨chlichen Effekt darstellt. Viele Biologieexperimente verfu¨gen u¨ber ungenu¨gend Sta¨rke, und die ju¨ngsten Bestrebungen, die traditionelle Signifikanzschwelle von 0,05 auf einen strengeren Wert von 0,005 zu a¨ndern, du¨rften die Leistung durchschnittlicher Experimente weiter reduzieren. Eine Leistungssteigerung durch Erho¨hung der Stichprobengro¨ße ist oft die einzige Option, die in Betracht gezogen wird, aber mehr Stichproben erho¨hen die Kosten, erschweren die Durchfu¨hrung des Experiments und stehen im Widerspruch zu den 3R-Prinzipien fu¨r die Tierforschung. Wir zeigen, wie die Gestaltung eines Experiments und einige analytische Entscheidungen einen u¨berraschend großen Einfluss auf die Leistung haben ko¨nnen.

Resumen Los experimentos con menos poder tienen tres problemas: los efectos verdaderos son complicados de detector, los efectos verdaderos que son detectados tienden a tener un taman˜o de efecto aumentado y a medida que el poder se reduce tambie´n lo hace la probabilidad de que un resultado estadı´sticamente sig- nificativo represente un efecto verdadero. Muchos experimentos biolo´gicos tienen un poder insuficiente y recientes llamadas para cambiar el lı´mite significativo tradicional de 0,05 a un valor ma´s contundente de 0,005 reducira´ todavı´ama´s el poder del experimento medio. Aumentar el poder aumentando el taman˜odela muestra es a menudo la u´nica opcio´n considerada, pero con ma´s muestras se incrementan los costes, el experimento es ma´s difı´cil de llevarse a cabo y se contradice el principio de las 3 R para la investigacio´n animal. Mostramos co´mo el disen˜o de un experimento y algunas decisiones analı´ticas pueden tener sorpren- dentemente un gran efecto sobre el poder Original Article Laboratory Animals 2018, Vol. 52(6) 630–640 ! The Author(s) 2018 Pigs are useful for the molecular Article reuse guidelines: sagepub.com/journals- study of bone inflammation and permissions DOI: 10.1177/0023677218766391 regeneration in humans journals.sagepub.com/home/lan

Freja Lea Lu¨thje1,2, Kerstin Skovgaard2, Henrik Elvang Jensen1 and Louise Kruse Jensen1

Abstract Pigs are used with increased frequency to model different kinds of orthopedic surgical conditions. In order to show the full potential of porcine models in orthopedic research, it is therefore required to examine the expression of bone regulatory genes in pigs affected by orthopedic surgery and compare it to the expression in humans and mice as mice, are one of the most applied animal species in orthopedics today. In the present study, the local molecular response to drilling of a tibial implant cavity, and the subsequent insertion of a steel implant was examined in a porcine model. Pigs were euthanized five days after drilling of the bone. The molecular response of 73 different genes was analyzed using a high-throughput quantitative polymerase chain reaction platform and compared to histopathology. Histologically, it was found that bone remodeling was initiated on day 5 after surgery and was associated with upregulation of several genes involved in bone degradation and formation (CTSK, ACP5, IBSP, RANK, RANKL and COL1A1). Interleukin-6 and several acute- phase proteins (C3, SAA and ITIH4) were significantly upregulated, indicating their importance in the initial process of healing and osseointegration. All tested bone morphogenic proteins (BMP2, -4 and -7) including their inhibitor noggin were also significantly upregulated. Surprisingly, vascular endothelial growth factor A was not found to be regulated five days after surgery while several other vascular growth factors (ANGPT1, ANGPT2 and PTN) were upregulated. The pig was found to be a useful model for elucidation of bone regulatory genes in humans.

Keywords Porcine model, bone drilling, bone healing, orthopedic implant, aseptic inflammation, gene expression, high- throughput RT-qPCR, osseointegration, orthopedic surgery, acute-phase proteins, growth factors

Date received: 1 December 2017; accepted: 27 February 2018

Introduction more pigs are used to model different kinds of ortho- Anatomically and physiologically, pigs are similar to pedic surgical conditions, and recently several studies humans1 and have many advantages as a model for have been published highlighting the advantages of orthopedic conditions when compared to rodents.2 porcine models of osseointegration of orthopedic The size of pigs makes them suitable for using ortho- pedic devices and procedures similar to those applied in 1Department of Veterinary and Animal Science, University of humans, and in contrast to mice, it is possible to sample Copenhagen, Denmark large quantities of blood and tissue for analysis. Most 2Department of Biotechnology and Biomedicine, Technical important, the immune system and the bone matrix University of Denmark, Denmark composition of humans are more similar to pigs than to mice.3–7 However, there is no argument for a bone Corresponding author: Freja Lea Lu¨thje, Department of Veterinary and Animal Science, genetic superiority of a porcine orthopedic model over University of Copenhagen, Ridebanevej 3, 1870 Frederiksberg C, a mouse model, as the local expression of bone regula- Denmark. tory genes is non-characterized in pigs. Increasingly Email: [email protected] Lu¨thje et al. 631 implants.8,9 Therefore, it is necessary to examine the At arrival, the animals were allowed to acclimatize expression of bone regulatory genes in pigs after ortho- for one week before entering the trial. The animals pedic surgery and compare them to the expression in were housed in the same pen (3 3 m) and fed a com- humans and mice in order to evaluate the full potential mercial pig diet ad libitum and had free access to tap of pigs as experimental animal models in orthopedic water. Control pigs previously used to characterize and research. show reproducibility of a porcine implant-associated Reports on the local molecular response toward osteomyelitis model were used in the present study.25 bone damage including orthopedic implant insertion Therefore, with reference to reduction, refinement and originate mainly from fractures studied in murine replacement (the 3Rs), the present study fulfills the models.10 The best way to obtain insight into the expectation of reduction, as the material is collected human physiology is obviously to examine humans; from previously used pigs, instead of using new pigs however, for ethical reasons this is not always feasible. to generate similar results. The systemic and local response in patients with impaired fracture healing have been examined in a few studies using blood samples or specimens of bone Materials and methods 11–13 tissue, respectively. However, in order to fully Experimental design understand the healing process to bone damage, the local molecular and morphological response is opti- The surgical procedure has recently been described in mally examined in discriminative animal models. Both detail25 and was approved by the Danish Animal the local histology and gene expression after bone Experiments Inspectorate (license No. 2013/15-2934- damage are well characterized in rodents.14–17 00946) according to the Danish proclamation of law However, molecular characterization in large animal on animal experimentation (LBK No. 1306,23/11/ models like pigs and in humans8,18,19 are limited even 2007). All animals were cared for in accordance with though orthopedic conditions like joint replacement the principles laid down by the European Union and osteosynthesis are standard procedures.20,21 Directive 2010/63/EU for Protection of Animals used A major issue related to orthopedic surgery is osteo- for Scientific Purposes. Briefly, a 4 20 mm implant necrosis caused by trauma and high temperatures cavity was drilled with a steel k-wire in the trabecular during drilling in bones,22 as well as the response to a bone at the proximal end of the right tibia approxi- foreign body,23 which depends on the material and sur- mately 10 mm distal to the growth plate in anesthetized face characteristics.8,9,18 pigs. Saline (10 ml) was injected into the cavity, and a Understanding the local impact of bone damage stainless steel implant (2 15 mm) was inserted. Five both morphologically and at the molecular level is days after the operation, the pigs were euthanized by required in order to improve the procedures related to an overdose of 20% pentobarbital given intravenously insertion of orthopedic devices and to prevent implant- and the implants were removed. Hereafter, the right associated osteomyelitis. A detailed examination of the tibial bones were sagittal-sectioned through the implant molecular changes and the cellular environment adja- cavity. One half was used for histology and placed in cent to bone damage might provide a platform for iden- formalin for fixation, while trabecular bone tissue sur- tifying novel agents with therapeutic potential, such as rounding the implant cavity from the other half was bone morphogenic proteins (BMPs), which already placed in RNAlater for molecular analysis. From have been used clinically to reduce healing time in each animal, the left tibial bone was likewise split in tibial fractures.24 two and placed in formalin and RNAlater, respectively, The aim of the present study was to analyze the genetic as controls. expression of normal bone tissue and bone tissue affected by drilling of an implant cavity with insertion of a steel Anesthetic and analgesic protocol implant in a porcine model. Based on the obtained results, the study aimed to describe the potential of porcine Premedication and induction of anesthesia was performed models in orthopedics on a bone molecular level. in the stable with intramuscular injections (1 ml/10 kg Overall, we demonstrate that the expression of bone regu- body weight (BW)) of a solution containing a mixture latory genes in pigs affected by orthopedic surgery is com- of 125 mg zolezepam, 125 mg tiletamine, 6.25 ml xylazine parable to results obtained in mice and humans. (20 mg/ml), 1.25 ml ketamine (100 mg/ml) and 2 ml butorphanol (10 mg/ml). Two catheters were then Animals inserted into lateral ear veins. One catheter was used for surgical anesthesia with propofol (10 mg/ml) Six 3-month-old (30 kg) female Danish Landrace pigs 1 ml/kg BW, and one catheter was used for intraopera- from a specific pathogen-free herd were used. tive analgesia with Fentanyl (0.5 mg/h). At the end of 632 Laboratory Animals 52(6) surgery, animals received an intramuscular injection ND-1000 ultraviolet spectrophotometer (Thermo 0.1 mg/kg BW of Buprenorphine (0.3 mg/ml) result- Fisher Scientific). The purity was evaluated based ing in postoperative analgesia for the following six on the A260/280 and A260/230 ratios and accepted for fur- hours. Thereafter, and throughout the experiment, ther analysis. The RNA concentration ranged from animals received daily oral analgesic treatment with 218 ng/ml to 685 ng/ml with a mean of 433 ng/ml. The Meloxicam 0.3 mg/kg BW. The animals recovered quality of the RNA was determined using Agilent from anesthesia under observation in individual pens. RNA 6000 Nano chips, Agilent RNA 6000 Nano Throughout the postoperative period, the pigs were fol- reagents and Agilent 2100 Bioanalyzer (Agilent lowed up daily including inspection of surgical wound Technologies). An RNA integrity number (RIN) was and evaluation of gait. Impaired ability to walk and assigned to each sample. Mean RIN was 5.9 1.4. The stand, and anorexia were set as human endpoints. extracted RNA was stored at –80C.

Histology Reverse transcription, pre-amplification and exonuclease treatment After one week of fixation in formalin the bone tissue was decalcified in formic acid for four weeks. Blocks of Complementary DNA (cDNA) was synthesized from bone tissue, representing the area surrounding the 500 ng extracted total RNA using the QuantiTect implant cavity and normal trabecular bone tissue, were Reverse Transcription Kit (Qiagen) according to the cut from the right and left tibial bones, respectively. All manufacturer’s specifications but with a 25% reduction blocks of bone tissue were then passed through graded in the volume of buffers, primer mix and reverse tran- concentrations of alcohol and embedded in paraffin wax. scriptase. For genomic DNA elimination, the RNA was Finally, tissue sections (4–5 mm) were stained with hema- mixed with 1.5 ml Wipeout buffer and RNase-free water toxylin and eosin in order to examine the morphology to a final volume of 14 ml and incubated on a and with Safranin O to study chondroid tissue. TProfessional TRIO thermocycler (Biometra). The cDNA synthesis was performed by adding 6 ml Assay design QuantiTect Reverse Transcription master mix (0.75 ml Reverse Transcriptase, 0.75 ml RT primer mix, 3 mlRT A literature search was conducted to find genes likely to buffer and 1.5 ml RNase-free water) to the DNase- be involved in bone healing and osseointegration. treated RNA, reaching a final reaction volume of Selected genes for the expression analysis can be divided 20 ml. Two technical replicates of cDNA were made into three main groups: (1) markers of bone remodeling, for each RNA sample, except for three samples with (2) inflammatory genes and (3) growth factors. an RIN under 6 for which three replicates were com- Primer3 software (http://bioinfo.ut.ee/primer3-0.4.0/) pleted. Non-reverse transcriptase controls were also was used to design primers with a mean temperature made and included in the following pre-amplification (Tm) between 59 C and 61 C and amplicon lengths and quantitative polymerase chain reaction (qPCR). from 70 to 150 bp. Primers were designed to span The cDNA was diluted 1:10 in low-ethylenediaminete- introns and to target most or all splice variants of the traacetic acid (EDTA) TE-buffer (Panreac AppliChem) gene of interest. Sequences, amplicon lengths, polymer- before pre-amplification. One ml TaqMan PreAmp ase chain reaction (PCR) efficiencies, fold changes Master Mix (Applied Biosystems), 2.5 ml 200 mM mix and p values for all primers and primer assays can be of all primers used subsequently for qPCR, 4 ml low- found in the supplementary data (Table S1). Nucleotide EDTA TE-buffer (Panreac AppliChem) and 2.5 ml blast searches were performed to test the specificity of diluted cDNA was mixed and incubated at 95C for all primers (https://blast.ncbi.nlm.nih.gov/Blast.cgi) 10 minutes followed by 19 or 21 cycles of 95C for before synthesis (Sigma-Aldrich). 10 seconds and 60C for four minutes. The pre-ampli- fied cDNA was treated with 16 U Exonuclease I (New RNA extraction England Biolabs) for 30 minutes at 37 C. The product was stored at –20C. Before qPCR the cDNA was Tissue samples preserved in RNAlater were transferred diluted 1:10 in low-EDTA TE-buffer (Panreac to 1 ml Qiazol (Qiagen) in M-tubes (Miltenyi Biotec) AppliChem). and homogenized using a gentleMACS Dissociator (Miltenyi Biotec). RNA was isolated using an High-throughput qPCR miRNeasy Mini kit (Qiagen) and treated with RNase- Free DNase set (Qiagen) according to the manufac- qPCR was performed using 96.96 Dynamic array inte- turer’s specifications. The concentration and purity of grated fluid circuit (IFC) chips (Fluidigm) in the the extracted RNA was determined with a NanoDrop BioMark HD System (Fluidigm) combining 96 primer Lu¨thje et al. 633 Results sets with 96 samples. Each primer mix contained 3 ml 2X Assay Loading Reagent (Fluidigm) and 3 ml All animals were healthy prior to surgery. After recov- of 20 mM forward and reverse primers suspended in ery from anesthesia, all pigs were lame on the inocu- low-EDTA TE-buffer (Panreac AppliChem). The lated leg. However, they were able to use the leg and sample mixes were prepared using 3 ml 2X TaqMan walked freely around in the pens. During the first day Gene Expression Master Mix (Applied Biosystems), the lameness diminished and disappeared two to three 0.3 ml 20X DNA Binding Dye Sample Loading reagent days after surgery. All pigs ate and drank normally (Fluidigm), 0.3 ml 20X Evagreen (Panreac AppliChem), throughout the experiment.25 0.9 ml low-EDTA TE-buffer (Panreac AppliChem) and 1.5 ml pre-amplified diluted (1:10) cDNA. Before load- Gene expression analysis ing the samples and primers, the chip was primed in an HX IFC controller (Fluidigm). After priming 5 ml After data processing and removal of duplicate pri- of each sample mix and primer mix were distributed mers, valid expression data from 73 genes were into the appropriate compartments and loaded into obtained. Of these, 36 had a fold-change difference the chip in the HX IFC controller. Thereafter the between the mean of the drilled and the healthy chip was inserted in the BioMark real-time PCR instru- bone of at least 2 and were significantly different. ment (Fluidigm) and the following program was used: Many of the molecular differences could be linked two minutes at 50C and 10 minutes at 95C, next 35 to morphological changes. Owing to limited space, cycles with 15 seconds at 95C and one minute at 60C. only selected genes of importance are described and After qPCR, melting curves were generated from 60C discussed in this paper, but a full list is available in the to 90C. Non-template controls and non-reverse tran- supplementary data (Table S1). scriptase controls were included to indicate problems with contamination, non-specific amplification or gen- Bone remodeling omic DNA. Standard curves constructed from three separate dilution series of pooled cDNA of all samples Toward the implant cavity an interrupted thin layer of were used to determine the efficiency of each primer elongated fibroblasts and leukocytes were seen lining assay. compressed and osteonecrotic trabecular bone tissue showing a faint staining with empty osteocytic lacuna Data processing and statistics (Figure 1(a)). This layer was sporadically inter- mingled with single neutrophils and osteoclasts. Raw data were inspected using Fluidigm Real- Ruffle-bordered osteoclasts were also seen (Figure Time PCR Analysis software (v. 4.1.3). GeneEx5 1(b)) and their activity was confirmed by upregulation (v. 5.4.4.119) (MultiD) was used for pre-processing, of the genes encoding cathepsin K (CTSK)andtar- normalization and relative quantification. Data were trate resistant acid phosphatase 5, also known as corrected for PCR efficiency for each primer assay TRAP (ACP5) (Figure 2(a)), which are both central and thereafter normalized with four reference for bone resorption. genes (b-2-microglobin, glyceraldehyde-3-phosphate Receptor activator nuclear factor kappa B dehydrogenase, peptidylprolyl isomerase A and (RANK, encoded by TNFRSF11A) and the transcrip- TATA-box binding protein). Out of six tested reference tion factor nuclear factor of activated T cells-1 genes, these four were found to have the most stable (NFATc1), both essential for osteoclast activation,28 expression using both GeNorm26 and NormFinder.27 were both significantly upregulated. The same applied After normalization, low-quality RNA samples were to receptor activator nuclear factor kappa B ligand carefully inspected to ensure valid data. The cDNA rep- (RANKL, encoded by TNFSF11), while there was no licates were averaged after normalization. Inconsistent significant change in the expression of the decoy recep- primer assays (defined as >17% of cDNA replicates tor osteoprotegrin (OPG, encoded by TNFRSF11B) varying >1.5 Cq) were excluded or validated by a (Figure 2(b)). second qPCR run. For each primer assay the Cq In the periphery of the tissue surrounding the values were converted to relative quantities. For visual- implant cavity, edema and proliferation of fibroblast ization, the data were scaled to give the healthy bone a and osteoblast blended into the adjacent normal bone mean relative expression of 1. Data were log2-trans- tissue and bone marrow. The production of new oste- formed before paired t tests, which were carried out oid by proliferating cubic osteoblasts was also observed using Excel (2010) (Microsoft). Genes were considered (Figure 1(c) and (d)), indicating intramembranous significantly regulated if at least a two-fold difference osteogenesis. New bone formation was confirmed by was found between the means of healthy and operated- molecular analysis of the bone matrix proteins bone on bones and if the p value was below 0.05. sialoprotein (IBSP) and collagen type 1 alpha 1 634 Laboratory Animals 52(6)

Figure 1. Necrotic trabecular bone and newly formed osteoid adjacent to a drilled implant cavity five days after surgery in a porcine model. Hematoxylin and eosin-stained (a) necrotic bone (NB) with trabeculae depleted of osteocytes leaving the lacunae empty. Bar 80 mm. (b) Osteoclast (arrow), resorbing trabecular bone tissue. Bar 350 mm. (c) Mineralized bone lined by layers of stimulated osteoblasts with a cuboidal shape (arrow). A mixed population of inflammatory cells is present around the bone structure. Bar 80 mm. (d) New mineralized osteoid (O) produced by surrounding cubical osteoblasts (arrows). Bar 80 mm.

(COL1A1), which were both significantly upregulated has a central role in activating all three complement (Figure 2(a)). cascades,29 was found to be significantly upregulated No cartilage formation could be determined by hist- (Figure 3). The positive APPs serum amyloid A ology but aggrecan (ACAN), a matrix protein in car- (SAA) and inter-a-trypsin inhibitor H4 (ITIH4, also tilaginous tissue, showed a significant almost 11-fold known as pig major acute-phase protein (pigMAP)), upregulation and SOX9, an important transcription were also found to be highly upregulated after implant factor for chondrocytes, showed a borderline significant insertion (Figure 3). seven-fold upregulation with a p value of 0.059 (Figure 2(a)). Growth factors

Inflammation All the tested members of the transforming growth factor (TGF)-b superfamily were significantly upregu- Histologically, infiltrating mononuclear cells were seen lated after surgery (Figure 4(a)). With a four-fold upre- (Figure 1(c)) to aggregate in the damaged bone tissue. gulation, TGFB1 had the highest fold change, while No regulation of interleukin-1 (IL-1) (Figure 3) or bone morphogenic protein (BMP)2, -4 -7 and their tumor necrosis factor (TNF) (below detectable levels inhibitor noggin (encoded by NOG) showed two- to (data not shown)) was found, while interleukin-6 three-fold upregulations. (IL-6) was highly and significantly upregulated (more Vascular endothelial factor A (VEGFA) and its than 40-fold) (Figure 3). Likewise several acute-phase transcription factor hypoxia inducible factor-1a proteins (APPs) were found to be upregulated after (HIF1A) were not regulated in the present study surgery. The complement component 3 (C3), which (Figure 4(b)), while the vascular growth factors Lu¨thje et al. 635

Figure 2. Relative expression (RQ) of genes related to (a) bone remodeling and (b) regulation of osteoclast activity in a porcine model of implant insertion five days after surgery. Dots represent a single sample and bars the mean for the group (n ¼ 6); left unharmed tibia (/solid bar) and the right operated-on bone (*/dashed bar). *p value < 0.05 in a paired t test.

Discussion Bone remodeling The upregulation of genes involved both in bone formation (IBSP and COL1A1) and bone degradation (CTSK and ACP5) confirmed that both osteoclasts and osteoblasts were actively degrading and regenerat- ing bone tissue, respectively, already five days after sur- gery (Figure 2(a)). Osteoclast activity and therefore bone resorption is influenced by the ratio between RANKL and its decoy receptor OPG.28 Both molecules are secreted by osteo- blasts and have also been shown to play a role in human fracture healing.12 In the present study, RANKL was upregulated while no significant change Figure 3. Relative expression (RQ) of inflammatory genes in OPG expression was seen (Figure 2(b)), indicating in a porcine model of implant insertion five days after that the RANKL/OPG ratio was increased. This is in surgery. Dots represent a single sample and bars the mean agreement with the observed increased osteoclast activ- for the group (n ¼ 6); left unharmed tibia (/solid bar) and ity and an increase in the ratio on day 7 in a murine the right operated-on bone (*/dashed bar). *p value < 0.05 16 in a paired t test. fracture model. Interestingly, SOX9 and ACAN, both markers of chondrogenesis, were upregulated (Figure 2(a)), while the histology did not reveal any formation of angiopoietin-1 (ANGPT1), angiopoietin-2 (ANGPT2) cartilage. This could imply that on day 5, pre- and pleiotrophin (PTN also known as osteoblast- chondrocytes are differentiating and starting the specific factor 1) were all significantly upregulated production of cartilage, but an extracellular chondroid three-, two- and 19-fold, respectively (Figure 4(b)). matrix has not yet been established. In a mouse frac- The tested fibroblast growth factors and insulin-like ture study, SOX9 has also been found to be highly growth factors, also considered important in bone upregulated.15 However, in another murine model repair,30 did not show any regulation five days after of osseointegration, no cartilage was detected histolo- surgery. Neither did platelet-derived growth factor gically with Safranin O staining during a 28-day long (PDGF)-B but PDGF-A was upregulated (Table S1). time course.14 636 Laboratory Animals 52(6)

Figure 4. Relative expression (RQ) of (a) bone morphogenic proteins and (b) vascular growth factors in a porcine model of implant insertion five days after surgery. Dots represent a single sample and bars the mean for the group (n ¼ 6); left unharmed tibia (/solid bar) and the right operated-on bone (*/dashed bar). *p value < 0.05 in a paired t test.

Inflammation of serum ITIH4 concentrations has been reported in patients with nonunion fractures,38 and it has been sug- The pro-inflammatory cytokines IL-1, IL-6 and TNF gested that fragments of the protein can be used as are especially important for initiation of inflammation serum biomarkers for osteoporosis.39 and the repair process after injury by redirecting immune cells, inducing APP production and promoting Growth factors bone resorption.31 Messenger RNA coding for IL-6 was more than 40-fold upregulated after the implant- TGF-b is important for bone remodeling; it recruits ation procedure (Figure 3). In agreement with these mesenchymal stem cells, fibroblast and immune cells, results, the expression of IL-6 was found to be elevated stimulates the proliferation of osteoblasts and chondro- at day 4 in a murine fracture model, but only with a cytes and increases the production of extracellular 3.4-fold upregulation.16 However, IL-1 (Figure 3) and matrix proteins.40 A four-fold upregulation of TGFB1 TNF (data not shown) were not regulated. We specu- was found in response to the present implant insertion late that IL-1 and TNF were upregulated immediately (Figure 4(a)). In accordance with our results, TGF-B1 after the surgical procedure but had returned to basal was upregulated in bone tissue after fracture in mice41 levels at day 5. This is in line with observations in mice, and elevated concentrations in serum and fracture in which both IL-1 and TNF rapidly declined three hematomas of humans with long bone fractures have days after a fracture.32 In humans as well IL-1 has also been reported.13 been reported to be downregulated from day 3 to 7 Bone morphogenic proteins (BMPs) are important after insertion of titanium bone implants.19 for bone and cartilage development and repair; in add- APPs are circulating plasma proteins found to be ition, they are promising therapeutics for treatment of highly regulated by inflammation.33 The expression of bone defects.42 BMP2 and BMP7 have been used clin- SAA and complement factors have not previously been ically for treatment of open fractures and nonunions to characterized in porcine bone tissue. C3, SAA and the reduce healing time and complications.24 BMP2, -4 and main porcine APP ITIH4 were all significantly upregu- -7 were all significantly upregulated after surgery lated (Figure 3). Previously, osteoclasts have been (Figure 4(a)). Cho et al. found that BMP2 and BMP4 shown to produce C3, and the cleaved product C3a were upregulated in mice immediately after fracture stimulates osteoblast differentiation in mice.34 while BMP7 expression increased after seven days.41 In addition, complements C3 and C5 have been Based on the temporal expression of all the BMPs, linked to fracture healing in mice.35 SAA is mainly Cho et al. speculated that BMP2 is responsible for trig- produced and secreted by the liver, but expression has gering the healing cascade and expression of other also been reported in human bone tissue,36 and it has BMPs.41 This hypothesis is supported by the fact that been shown to affect the differentiation of osteoblasts the healing process is not initiated in fractures of BMP2 and osteoclasts in vitro.37 In humans, downregulation knockout mice.43 The ratio between BMP4 and the Lu¨thje et al. 637 Conclusion BMP inhibitor noggin has also been speculated to be important in fracture healing. We found a two-fold The present molecular analysis gives a comprehensive upregulation of both BMP4 and noggin (Figure 4(a)), understanding of how bone drilling and implant inser- indicating that the BMP4/noggin ratio is not signifi- tion changes the local gene expression in bone tissue cantly changed by the implant insertion, but that five days after surgery in a porcine model. In conclu- both molecules play a role in the response to the inser- sion, the porcine model revealed similarities to results tion procedure. Kloen et al. showed that BMPs and from both human and murine studies. Therefore, their inhibitors are co-expressed and found a difference pigs are useful and relevant, also on a local molecular in the ratio of BMPs and their inhibitors in callus of level, as a large animal model in orthopedic surgery. human fractures compared to tissue from nonunions, Appropriate to the 3Rs, increased background know- indicating their importance in bone repair.44 ledge of experimental animals will increase their reli- Following implantation, revascularization is import- ability, which might lead to a reduction in the ant for the surrounding healing process, since angio- number of animals used. In the future, the present pub- genesis is essential for ossification.45 VEGFA is lication will serve as a strong argument for the use of described as one of the most important growth factors porcine models in preclinical orthopedic studies, but for angiogenesis, also in relation to bone repair,46 and also as a basis for further research of the transcriptional it has been reported to be upregulated both in mouse bone response related to infections and different kinds and rat fracture models.15,47 However, in the present of local medical treatments. model, no regulation was seen of either VEGFA or its transcription factor HIF1A (Figure 4(b)). Instead Acknowledgments the vascular growth factors ANGPT1, ANGPT2 and We thank Betina Andersen, Elizabeth Petersen and Nicole PTN were all significantly upregulated (Figure 4(b)). Lind Henriksen for their excellent assistance with the hist- This indicates that VEGF-dependent angiogenesis ology and Karin Tarp for her technical assistance with the pathways are not as important in the osseointegration gene expression analysis. of implants as in fracture healing. A difference in the role of VEGFA in fracture and implant insertion is Declaration of Conflicting Interests supported by Schmid et al., who found that VEGFA The author(s) declared no potential conflicts of interest with was significantly upregulated both at one and nine days respect to the research, authorship, and/or publication of this post-fracture in a murine model, while the gene was article. upregulated only at day 1 in controls without fractures but with a steel pin inserted in the tibia.15 ANGPT1, -2 Funding and PTN have also been reported to be upregulated in murine fracture models.17,48 The author(s) disclosed receipt of the following financial sup- port for the research, authorship, and/or publication of this article: This study was supported by the European Union’s The relevance of porcine models Horizon 2020 research and innovation program under the NoMorFilm project (grant agreement no. 634588), and by Descriptive animal models are a necessity to advance the Danish Research Council (grant no. 4005-00035B). modern medicine by increasing our knowledge of the complex reactions the body elicits in response to ORCID iD injury and disease. The most important parameter in choosing an animal model should not be low cost and Louise Kruse Jensen http://orcid.org/0000-0002-0372-4923 easy handling, but how well the model mirrors human conditions. Porcine bone closely resembles human References bone in regards to anatomy, mineral composition, 1. Litten-Brown JC, Corson AM and Clarke L. Porcine rate of remodeling and microstructure3,7 and com- models for the metabolic syndrome, digestive and bone pared to mice the immune system of the pig is also disorders: A general overview. Animal 2010; 4: 899–920. more similar to the human counterpart.2,5,6 2. Fairbairn L, Kapetanovic R, Sester DP, et al. Therefore, results of bone inflammation attained in The mononuclear phagocyte system of the pig as a pigs have high translational potential.6,49 In today’s model for understanding human innate immunity and dis- ease. J Leukoc Biol 2011; 89: 855–871. bone research rodents are often preferred over large 3. Peric M, Dumic-Cule I, Grcevic D, et al. The rational use animal models because they are so well characterized of animal models in the evaluation of novel bone regen- at the molecular level. This article contributes to alle- erative therapies. Bone 2015; 70: 73–86. viate the deficiency in molecular studies in porcine 4. Mestas J and Hughes CC. Of mice and not men: bone as the use of porcine models in biomedical Differences between mouse and human immunology. research is increasing.1,49,50 J Immunol 2004; 172: 2731–2738. 638 Laboratory Animals 52(6)

5. Dawson HD, Reece JJ and Urban JF. A comparative 23. Anderson JM, Rodriguez A and Chang DT. Foreign analysis of the porcine, murine, and human immune sys- body reaction to biomaterials. Semin Immunol 2008; 20: tems. Vet Immunol Immunopathol 2009; 128: 323–342. 86–100. 6. Meurens F, Summerfield A, Nauwynck H, et al. The pig: 24. Ronga M, Fagetti A, Canton G, et al. Clinical applica- A model for human infectious diseases. Trends Microbiol tions of growth factors in bone injuries: Experience with 2012; 20: 50–57. BMPs. Injury 2013; 44(Suppl 1): S34–S39. 7. Aerssens J, Boonen S, Lowet G, et al. Interspecies differ- 25. Jensen LK, Koch J, Dich-Jorgensen K, et al. Novel por- ences in bone composition, density, and quality: Potential cine model of implant-associated osteomyelitis: A com- implications for in vivo bone research. Endocrinology prehensive analysis of local, regional, and systemic 1998; 139: 663–670. response. J Orthop Res 2017; 35: 2211–2221. 8. Wang N, Li H, Lu¨ W, et al. Effects of TiO2 nanotubes 26. Vandesompele J, De Preter K, Pattyn F, et al. Accurate with different diameters on gene expression and osseoin- normalization of real-time quantitative RT-PCR data by tegration of implants in minipigs. Biomaterials 2011; 32: geometric averaging of multiple internal control genes. 6900–6911. Genome Biol 2002; 3: RESEARCH0034. 9. Ogawa T and Nishimura I. Genes differentially expressed 27. Andersen CL, Jensen JL and Ørntoft TF. Normalization in titanium implant healing. J Dent Res 2006; 85: of real-time quantitative reverse transcription-PCR data: 566–570. A model-based variance estimation approach to identify 10. Bigham-Sadegh A and Oryan A. Selection of animal genes suited for normalization, applied to bladder and models for pre-clinical strategies in evaluating the frac- colon cancer data sets. Cancer Res 2004; 64: 5245–5250. ture healing, bone graft substitutes and bone tissue regen- 28. Crotti TN, Dharmapatni AA, Alias E, et al. eration and engineering. Connect Tissue Res 2015; 56: Osteoimmunology: Major and costimulatory inflamma- 175–194. tory induced bone loss. J Immunol Res 2015; 2015: 11. Binder H, Eipeldauer S, Gregori M, et al. The difference 281287. between growth factor expression after single and mul- 29. Merle NS, Church SE, Fremeaux-Bacchi V, et al. tiple fractures: Preliminary results in human fracture Complement system part I—Molecular mechanisms of healing. Dis Markers 2015; 2015: 203136. activation and regulation. Front Immunol 2015; 6: 262. 12. Ko¨ ttstorfer J, Thomas A, Gregori M, et al. Are OPG and 30. Tsiridis E, Upadhyay N and Giannoudis P. Molecular RANKL involved in human fracture healing? J Orthop aspects of fracture healing: Which are the important mol- Res 2014; 32: 1557–1561. ecules? Injury 2007; 38(Suppl 1): S11–S25. 13. Sarahrudi K, Thomas A, Mousavi M, et al. Elevated 31. Al-Aql ZS, Alagl AS, Graves DT, et al. Molecular mech- transforming growth factor-beta 1 (TGF-b1) levels in anisms controlling bone formation during fracture heal- human fracture healing. Injury 2011; 42: 833–837. ing and distraction osteogenesis. J Dent Res 2008; 87: 14. Colnot C, Romero DM, Huang S, et al. Molecular ana- 107–118. lysis of healing at a bone-implant interface. J Dent Res 32. Kon T, Cho TJ, Aizawa T, et al. Expression of osteopro- 2007; 86: 862–867. tegerin, receptor activator of NF-kB ligand (osteoprote- 15. Schmid GJ, Kobayashi C, Sandell LJ, et al. Fibroblast gerin ligand) and related proinflammatory cytokines growth factor expression during skeletal fracture healing during fracture healing. J Bone Miner Res 2001; 16: in mice. Dev Dyn 2009; 238: 766–774. 1004–1014. 16. Kidd LJ, Stephens AS, Kuliwaba JS, et al. Temporal pat- 33. Steel DM and Whitehead AS. The major acute phase tern of gene expression and histology of stress fracture reactants: C-reactive protein, serum amyloid P compo- healing. Bone 2010; 46: 369–378. nent and serum amyloid A protein. Immunol Today 17. Desai BJ, Meyer MH, Porter S, et al. The effect of age on 1994; 15: 81–88. gene expression in adult and juvenile rats following fem- 34. Matsuoka K, Park K, Ito M, et al. Osteoclast-derived oral fracture. J Orthop Trauma 2003; 17: 689–698. complement component 3a stimulates osteoblast differen- 18. Lang NP, Salvi GE, Huynh-Ba G, et al. Early osseointe- tiation. J Bone Miner Res 2014; 29: 1522–1530. gration to hydrophilic and hydrophobic implant surfaces 35. Ehrnthaller C, Huber-Lang M, Nilsson P, et al. in humans. Clin Oral Implants Res 2011; 22: 349–356. Complement C3 and C5 deficiency affects fracture heal- 19. Thalji GN, Nares S and Cooper LF. Early molecular ing. PLoS One 2013; 8: e81341. assessment of osseointegration in humans. Clin Oral 36. Kovacevic A, Hammer A, Stadelmeyer E, et al. Implants Res 2014; 25: 1273–1285. Expression of serum amyloid A transcripts in human 20. Robertsson O, Bizjajeva S, Fenstad AM, et al. Knee bone tissues, differentiated osteoblast-like stem cells and arthroplasty in Denmark, Norway and Sweden. Acta human osteosarcoma cell lines. J Cell Biochem 2008; 103: Orthop 2010; 81: 82–89. 994–1004. 21. Kurtz S, Ong K, Lau E, et al. Projections of primary and 37. Thaler R, Sturmlechner I, Spitzer S, et al. Acute-phase revision hip and knee arthroplasty in the United States protein serum amyloid A3 is a novel paracrine coupling from 2005 to 2030. J Bone Joint Surg Am 2007; 89: factor that controls bone homeostasis. FASEB J 2015; 29: 780–785. 1344–1359. 22. Augustin G, Zigman T, Davila S, et al. Cortical bone 38. de Seny D, Cobraiville G, Leprince P, et al. Biomarkers drilling and thermal osteonecrosis. Clin Biomech 2012; of inflammation and innate immunity in atrophic non- 27: 313–325. union fracture. J Transl Med 2016; 14: 258. Lu¨thje et al. 639

39. Bhattacharyya S, Siegel ER, Achenbach SJ, et al. Serum 45. Colnot C, Lu C, Hu D, et al. Distinguishing the contri- biomarker profile associated with high bone turnover and butions of the perichondrium, cartilage, and vascular BMD in postmenopausal women. J Bone Miner Res 2008; endothelium to skeletal development. Dev Biol 2004; 23: 1106–1117. 269: 55–59. 40. Poniatowski LA, Wojdasiewicz P, Gasik R, et al. 46. Hu K and Olsen BR. The roles of vascular endothelial Transforming growth factor beta family: Insight into growth factor in bone repair and regeneration. Bone the role of growth factors in regulation of fracture healing 2016; 91: 30–38. biology and potential clinical applications. Mediators 47. Komatsu DE and Hadjiargyrou M. Activation of the tran- Inflamm 2015; 2015: 137823. scription factor HIF-1 and its target genes, VEGF, HO-1, 41. Cho TJ, Gerstenfeld LC and Einhorn TA. Differential iNOS, during fracture repair. Bone 2004; 34: 680–688. temporal expression of members of the transforming 48. Lehmann W, Edgar CM, Wang K, et al. Tumor necrosis growth factor b superfamily during murine fracture heal- factor alpha (TNF-alpha) coordinately regulates the ing. J Bone Miner Res 2002; 17: 513–520. expression of specific matrix metalloproteinases 42. Carreira AC, Alves GG, Zambuzzi WF, et al. Bone mor- (MMPS) and angiogenic factors during fracture healing. phogenetic proteins: Structure, biological function and Bone 2005; 36: 300–310. therapeutic applications. Arch Biochem Biophys 2014; 49. Swindle MM, Makin A, Herron AJ, et al. Swine as 561: 64–73. models in biomedical research and toxicology testing. 43. Tsuji K, Bandyopadhyay A, Harfe BD, et al. BMP2 Vet Pathol 2012; 49: 344–356. activity, although dispensable for bone formation, is 50. Jensen LK, Johansen ASB and Jensen HE. Porcine required for the initiation of fracture healing. Nat Genet models of biofilm infections with focus on pathomor- 2006; 38: 1424–1429. phology. Front Microbiol 2017; 8: 1961. 44. Kloen P, Lauzier D and Hamdy RC. Co-expression of BMPs and BMP-inhibitors in human fractures and non- unions. Bone 2012; 51: 59–68.

Re´sume´ Les porcs sont de plus en plus fre´quemment utilise´s pour e´tudier diffe´rents types d’e´tats lie´sa` des chirurgies orthope´diques. Afin de de´montrer le ve´ritable potentiel des mode`les porcins dans la recherche orthope´dique, il est par conse´quent primordial d’analyser l’expression des ge`nes re´gulateurs osseux des porcs subissant une chirurgie orthope´dique et de la comparer a` l’expression des ge`nes de l’homme et des souris car ces dernie`res font partie des espe`ces animales qui sont actuellement les plus couramment utilise´es dans la recherche orthope´dique. Dans la pre´sente e´tude, la re´action mole´culaire locale au forage effectue´ dans la cavite´ de l’implant tibial et l’insertion ulte´rieure d’un implant en acier ont e´te´ e´tudie´es chez un mode`le porcin. Les porcs ont e´te´ euthanasie´s 5 jours apre`s le forage de l’os. La re´action mole´culaire de 73 ge`nes diffe´rents a e´te´ examine´e en utilisant une plate-forme qPCR a` haut de´bit et une comparaison a e´te´ effectue´ea` l’aide d’une analyse histopathologique. D’un point de vue histologique, il a e´te´ e´tabli que le remodelage osseux a com- mence´ 5 jours apre`s l’ope´ration et qu’il e´tait associe´ a` la re´gulation positive de plusieurs ge`nes implique´s dans la de´gradation et la formation des os (CTSK, ACP5, IBSP, RANK, RANKL et COL1A1). L’interleukine 6 ainsi que plusieurs prote´ines de phase aigue¨ (C3, SAA et ITIH4) pre´sentaient une re´gulation fortement posi- tive, indiquant leur importance dans le processus initial de gue´rison des os et d’oste´o-inte´gration. Toutes les prote´ines morphoge´niques osseuses teste´es (BMP2, -4 et -7), y compris leur inhibiteur noggin, pre´sentaient e´galement une re´gulation conside´rablement positive. De manie`re surprenante, il a e´te´ constate´ que le facteur de croissance de l’endothe´lium vasculaire A n’a subi aucune re´gulation 5 jours apre`s l’ope´ration tandis que plusieurs autres facteurs de croissance vasculaire (ANGPT1, ANGPT2 et PTN) pre´sentaient une croissance positive. Il a e´te´ e´tabli que les porcs repre´sentent un mode`le de comparaison utile permettant d’e´lucider le myste`re lie´ aux ge`nes re´gulateurs osseux de l’homme.

Abstract Schweine werden immer ha¨ufiger bei der Modellierung verschiedener Arten von orthopa¨dischen Operationsbedingungen eingesetzt. Um das volle Potenzial von Schweinemodellen in der orthopa¨dischen Forschung aufzuzeigen, ist es daher erforderlich, die Expression von knochenregulierenden Genen bei Schweinen, die durch orthopa¨dische Chirurgie verursacht wurden, zu untersuchen und mit der Expression bei Menschen und Ma¨usen zu vergleichen, da Ma¨use heute zu den am ha¨ufigsten verwendeten Tierarten in der Orthopa¨die geho¨ren. In der vorliegenden Studie wurde die lokale molekulare Reaktion auf das Bohren 640 Laboratory Animals 52(6) eines Tibia-Implantathohlraums und das anschließende Einsetzen eines Stahlimplantats in einem Schweinemodell untersucht. Die Schweine wurden 5 Tage nach der Knochenbohrung eingeschla¨fert. Die molekulare Reaktion von 73 verschiedenen Genen wurde mit Hilfe einer qPCR-Plattform mit hohem Durchsatz analysiert und mit der Histopathologie verglichen. Histologisch wurde festgestellt, dass der Knochenumbau am fu¨nften Tag nach der Operation eingeleitet wurde und mit der Hochregulation mehrerer Gene, die am Knochenabbau und an der Knochenbildung beteiligt sind, verbunden war (CTSK, ACP5, IBSP, RANK, RANKL und COL1A1). Interleukin 6 und mehrere Akute-Phase-Proteine (C3, SAA und ITIH4) wurden signifikant hochreguliert, was auf ihre Bedeutung fu¨r den anfa¨nglichen Heilungsprozess und die Osseointegration schließen la¨sst. Alle getesteten knochenmorphogenen Proteine (BMP2, -4 und -7) einschließlich ihres Inhibitors Noggin wurden ebenfalls signifikant nach oben reguliert. U¨berraschenderweise wurde der vaskula¨re endotheliale Wachstumsfaktor A 5 Tage nach der Operation nicht reguliert, wa¨hrend mehrere andere vaskula¨re Wachstumsfaktoren (ANGPT1, ANGPT2 und PTN) hochreguliert wurden. Das Schwein erwies sich als nu¨tzliches Modell fu¨r die Erhellung knochenregulatorischer Gene beim Menschen.

Resumen Cada vez ma´s se utilizan cerdos como modelos para distintos tipos de condiciones quiru´rgicas ortope´dicas. Para poder demostrar todo el potencial de los modelos porcinos en la investigacio´n ortope´dica es necesario, por tanto, examinar la expresio´n de los genes regulatorios o´seos en cerdos debido a una intervencio´n quiru´rgica ortope´dica y compararla con la expresio´n en humanos y ratones ya que estos u´ltimos son una de las especies animales ma´s utilizadas actualmente en ortopedia. En el presente estudio, la respuesta molecular local de perforacio´n de una cavidad de implante tibial y el subsiguiente inserto de un implante de acero fue examinada en un modelo porcino. Se sacrifico´ a los cerdos cinco dı´as despue´s de la perforacio´n del hueso. La respuesta molecular de 73 genes distintos fue analizada utilizando una plataforma qPCR de alto rendimiento y se comparo´ con la histopatologı´a. Histolo´gicamente, se encontro´ que el remodelado del hueso fue iniciado el dı´a cinco tras la operacio´n y se asocio´ alaregulacio´n de varios genes involucrados en la degradacio´n y formacio´no´sea (CTSK, ACP5, IBSP, RANK, RANKL y COL1A1). Interleukina 6 y varias proteı´nas de fase aguda (C3, SAA e ITIH4) fueron reguladas significativamente lo que indica su importancia en el proceso inicial de cura y oseointegracio´n. Todas las proteı´nas morfoge´nicas o´seas que fueron testadas (BMP2, -4 y -7), incluido su noggin inhibidor fueron tambie´n reguladas significativamente. Sorprendentemente, se observo´ que el factor A de crecimiento endotelial vascular no fue regulado cinco dı´as despue´sdelaoperacio´n mientras que otros factores de crecimiento vas- cular (ANGPT1, ANGPT2 y PTN) fueron regulados. Se encontro´ que el cerdo era un modelo u´til para la elucida- cio´ndelosgenesregulatorioso´seos en humanos. Expert in the world of research diets

Special Diets Services is the largest supplier of Laboratory Animal diets in Europe and the only dedicated manufacturer in the UK. Special Diets Services has a global reputation for the quality of its diets and manufacturing and storage facilities. Special Diets Services

Special Diets Services PO Box 705, Witham, Essex, England CM8 3AD Telephone: +44 (0) 1376 511260 SDS www.sdsdiets.com Fax: +44 (0) 1376 511247 the essential resource for quality research diets Email: [email protected]

SDS8222 SDS Corporate Ad LAJ 280x210mm.indd 1 19/01/2017 10:40 Supplying the Innovations of Modern Research

LabDiet® TestDiet® Bedding Environmental & Nesting Enrichment

Leader in Research Custom made Creating Natural Caring for your Animal Nutrition Speciality Diets Environments Laboratory Animals

IPS – working with leading industry brands and products to supply innovative research facilities worldwide

BenchGuard® Logistics Irradiation Customer & Tray Liners & Storage Services Service

Protective & A comprehensive Tailored to your +44 (0)870 6001616 Absorbent Papers range of services specific needs [email protected]

www.ipsltd.biz News Laboratory Animals 2018, Vol. 52(6) 643 ! The Author(s) 2018 Registration is opened for the 14th Article reuse guidelines: sagepub.com/journals- FELASA Congress! permissions DOI: 10.1177/0023677218810968 journals.sagepub.com/home/lan

registration. Don’t wait: this is on a ‘‘first come, first served’’ basis. The draft scientific programme is also available on- line: http://www.felasa2019.eu/resource/felasa2019/ files/FELASA-Preliminary-Programme.pdf. The scientific programme is composed of six streams: Education and Training, Severity Classification, Defining Good Care, Compliance and Registration is now opened for the 14th FELASA Communication, Reproducibility and Translation and Congress in Prague on 10–13 June 2019. Deadline for Model Management. Further details can be found on early bird registration is 31 March 2019. Please follow http://www.felasa2019.eu/preliminary-programme. the instructions on the congress website http://www. There is still time to submit your abstract for a felasa2019.eu/ where you can also find plenty more infor- poster or an oral presentation on http://www. mation on hotel booking, the draft scientific programme felasa2019.eu/call-for-abstracts. Hurry up, the deadline schedule, social events, exhibitors’ sponsoring, etc. for abstract submission is 31 January 2019. You can book your hotel with the congress registra- The Organizing and Scientific Committees look for- tion on-line system. Visit http://www.felasa2019.eu/ ward to seeing you at the congress! accommodation and http://www.felasa2019.eu/

Chair of the FELASA 2019 Scientific Committee United Kingdom of Great Britain and Northern Ireland

Corresponding author: J-P Mocho, Jean-Philippe Mocho, PO Box 372, Eye, IP22 9BR, UK. Email: [email protected]

Contributions to the News section are not subject to peer review and reflect the opinion of the subscribing society.

News Laboratory Animals 2018, Vol. 52(6) 645 ! The Author(s) 2018 Article reuse guidelines: What does the GV-SOLAS actually do? sagepub.com/journals- permissions DOI: 10.1177/0023677218810969 Bettina Kra¨nzlin and Nicole Linklater journals.sagepub.com/home/lan

and certify LAS courses or facilities for continuing edu- cation for LAS specialists. The title of ‘Expert for laboratory animal science GV-SOLAS’ is awarded after completing a continuing education program. By organizing scientific conferences and training events, the GV-SOLAS provides a forum for all those working with laboratory animals. At the annual conference, The Gesellschaft fu¨r Versuchstierkunde/Society for prizes are awarded for best poster and oral Laboratory Animal Science (GV-SOLAS) is a regis- contributions. tered association that has promoted the responsible GV-SOLAS also established a program to kick-start use of laboratory animals since 1964. It sees its role research projects related to the 3Rs for its members. as an intermediary between animal welfare and research The GV-SOLAS works with national and international for the benefit of humans and animals. In 1997, associations (FELASA, ICLAS, VBIO and others) and under the umbrella of the GV-SOLAS, the IGTP thereby contributes to the harmonization of experimen- (Interessengemeinschaft Tierpfleger; animal caretakers tal animal research at a European level. and technical staff collective) was founded. The object- The GV-SOLAS has now decided to translate into ives of the IGTP are the communication and exchange English all scientific information and publications on of experience between technical staff to promote the the GV-SOLAS home page http://www.gv-solas.de/, dissemination of technological innovations in the field with 50 publications out of 75 in total still awaiting of laboratory animal science. The IGTP organizes translation. GV-SOLAS intends to reach readers training for technical staff, is committed to the harmon- abroad as well as non-German-speaking people in ization of training for animal caretakers, as well as to Germany who are involved in animal experiments. promoting the education of bachelor professionals of The number of personnel involved in animal experi- animal care and management and contributing to the ments has increased steadily during recent years and publications of the GV-SOLAS. The GV-SOLAS/ it is important to make the GV-SOLAS publications, IGTP currently has 1186 ordinary, associate and sup- which act as working guidelines, available to everyone porting members. engaged in animal experimental work in order to main- The GV-SOLAS is a competent contact for legisla- tain the high standards of animal welfare and animal tors, scientists and the public on issues of laboratory research in Germany. The first publications to be trans- animal science. Its nine permanent committees publish lated (all from 2017) will be ‘Recommendations for recommendations on topics of laboratory animal wel- blood sampling in laboratory animals, especially small fare, husbandry, nutrition, genetics and breeding, anal- animals’, ‘Recommendations for administration of sub- gesia and anesthesia and hygienic aspects. A further stances to laboratory animals’ and ‘Humane housing of two committees focus on education and training of per- laboratory rats’. sonnel in laboratory animal science (LAS) in Germany,

Die Gesellschaft fu¨r Versuchstierkunde/Society of Laboratory Animal Science (GV-SOLAS)

Corresponding author: Bettina Kra¨nzlin Email: [email protected]

Contributions to the News section are not subject to peer review and reflect the opinion of the subscribing society. News Laboratory Animals 2018, Vol. 52(6) 646–647 ! The Author(s) 2018 Article reuse guidelines: Meeting report GV-SOLAS annual sagepub.com/journals- permissions conference 2018 DOI: 10.1177/0023677218810970 journals.sagepub.com/home/lan

Caroline Johner

their latest developments and established products. Attendance, with 1167 registered participants and exhibitors, was overwhelming. It was a meeting with many stimulating scientific discussions accompanied by an amazing industry exhibition. The community is growing, and the need to exchange and present infor- mation is constantly growing as this field becomes more With the motto ‘One Health – animal welfare for the and more challenging. benefit of all species’ the 56th Annual Meeting of the Education and training of people taking care of, or German Society of Laboratory Animal Science and the performing experiments in, animals are essential for 18th Advanced Training Course of the GV-IGTP took animal welfare and the quality of research. GV- place in Munich (September 12–14, 2018). SOLAS aimed to satisfy this interest by offering 12 With its two large universities, the Ludwig- workshops and 16 seminars in parallel to the main sci- Maximilians University (LMU) and the Technical entific program. A poster exhibition with 62 presenta- University of Munich (TUM), as well as the tions was made available throughout the entire meeting Helmholtz Zentrum Mu¨nchen, various Max Planck with presenters being available at specific times. Institutes, the Innovation and Founding Center One of the focal points of the conference was the Biotechnology (IZB) and numerous companies operat- severity assessment of procedures. In this context, two ing in the field of biosciences, Munich is home to a interactive FELASA Workshops on the Classification broad range of academic and scientific excellence and and Reporting of Severity were offered, which were expertise. This includes biomedical research areas like fully booked – participants spanning from scientists, diabetes, genetics, immunology, oncology, cardiology, animal care staff to animal welfare officers and mem- neurobiology, neuroimmunology, and neuro-degenera- bers of Competent Authorities. After a general intro- tive disorders to name just a few. Research in the field duction, small groups discussed different animal of these diseases requires laboratory animals. models. The groups then presented their outcome to Therefore, the welfare of these animals is an important the audience. Using an anonymous voting tool, all par- topic. Whenever research for the benefit of mankind ticipants were then asked to give their opinion on vari- and animals is performed in animals, society legitim- ous scenarios. This initiated lively discussions and ately demands the best conditions for the animals used. pointed out how difficult it is to set a defined standard. Achieving this aim has always been the main object- The conclusion was that more needs to be done in this ive of laboratory animal science. This link between man area. This was also reflected in one of the focus sessions and animal in research is stressed by the motto of the held by Andre´Bleich and Rene´Tolba on Severity meeting ‘One Health – animal welfare for the benefit of Assessment in Animal-Based Research, a research pro- all species’. ject (FOR 2591) funded by the German Research Consequently – amongst others – topics like severity assessment, anesthesia and analgesia, refinement of Animal Welfare Consultant/Officer, Freiburg im Breisgau, procedures and housing conditions as well as planning Germany and construction of animal facilities were an integral part of the congress. This included products for breed- Corresponding author: ing and housing animals under optimal conditions as Caroline Johner Email: [email protected] well as equipment for excellent research. For this reason, the Annual Meeting again featured a showcase Contributions to the News section are not subject to peer review of an industry exhibition where 70 companies presented and reflect the opinion of the subscribing society. Johner 647

Foundation (DFG). This project aims to improve vari- and Kirk Leech from EARA. Another seminar focused ous objective methods and techniques while, at the on the ethical evaluation of animal research. The ana- same time, developing new ones and combining them lysis of harm–benefit is one of the central subjects in the to create a scale for standardization of the assessment German animal research application process. Other of severity. seminars focused on more practical aspects, including Among the seminars offered was one on ‘Talking to the severity classification of zebrafish and the use of the public about animal research’. The speakers farm animals and other large animals in research. included Hannes Schlender from scienceRELATIONS Our Price Guarantee WE WON’T BE BEATEN ON PRICE Promise

Disposable Nesting & Enrichment

Environmental enrichment stems from the recognition that in addition to basic needs of food, water, warmth & health, housed animals have a physiological & psychological need to exert their natural behaviours.

Whether this involves searching for food, gathering nesting materials, playing, gnawing, hiding or exploration, they require activities that help pass the time and there is a

® Lab Bedding Products considerable body of literature now that demonstrates the Bed-r’Nest is a trademark of The Andersons, Inc. Patents pending influence of an animal's physical and social environment on the research results.

LIV DE E R

Y

D

E

G

E 24

U

T hour

A N R A

Call us now on: 0800 161 5831 for free samples and a quotation!

T: 0800 161 5831 (UK Free Phone) T: +44 (0)161 274 1080 (Int) E: [email protected] follow us News Laboratory Animals 2018, Vol. 52(6) 649–651 ! The Author(s) 2018 Article reuse guidelines: En Espan˜a se presenta el primer sagepub.com/journals- permissions DOI: 10.1177/0023677218811285 informe anual del Acuerdo de journals.sagepub.com/home/lan Transparencia en Experimentacio´n Animal

Sergi Vila

El pasado mes de septiembre se presentaron los resul- Dra. Margarita del Val (investigadora del CSIC, Vocal tados de la primera evaluacio´n del Acuerdo de COSCE de Ciencias de la Salud y de la Vida, que preside Transparencia en Experimentacio´n Animal de Espan˜a. la Comisio´n COSCE de Estudio del Uso de Animales en Esta iniciativa se propuso a finales de 2016 a partir de Investigacio´nCientı´fica), el Dr. Lluı´sMontoliu la exitosa experiencia anterior en el Reino Unido. A (Investigador Cientı´fico del CSIC, miembro de la trave´sdelaConfederacio´n de Sociedades Cientı´ficas de Comisio´n COSCE y coordinador de la implementacio´n Espan˜a (COSCE) se planteo´a las instituciones espan˜ o- pra´ctica del Acuerdo) y el Dr. Javier Guille´n (en repre- las que realizan experimentacio´n con animales, un sentacio´n de EARA y miembro de AALAC, Asociacio´n acuerdo para fomentar la transparencia de sus activi- Internacional para la Evaluacio´n y Acreditacio´ndel dades. El acuerdo consta de 4 puntos: Cuidado de Animales de Laboratorio e igualmente miembro de la Comisio´n COSCE). 1. Hablar con claridad sobre cuando, como y porque Una de las primeras conclusiones del Informe es que se utilizan animales en los experimentos. Esto se la transparencia en la comunicacio´n ha aumentado sustan- materializa en una declaracio´n institucional consul- cialmente. La mayorı´a de las instituciones han publicado table en su pa´gina web. oficialmente su declaracio´n institucional y de forma 2. Informar adecuadamente a los medios de comunica- habitual exponen el uso de animales en las noticias cio´n y al pu´blico en general sobre las condiciones en sobre avances los avances cientı´ficos que han desarrol- las que se realizan los experimentos con animales y lado. Ya so´lo la presentacio´n del Acuerdo, supuso la los resultados que de ella se obtienen aparicio´n de varios reportajes periodı´sticos extensos 3. Promover iniciativas que generen un mayor conoci- sobre la experimentacio´n animal en Espan˜ a. Poniendo miento y comprensio´n en la sociedad sobre el uso de de manifiesto co´mo se trabaja de forma e´tica y respon- animales en investigacio´n cientı´fica sable bajo las directivas europeas, en instalaciones adap- 4. Reportar anualmente mediante una encuesta los tadas y cumpliendo el principio de las 3Rs. El 80% de las progresos realizados y sus experiencias. instituciones han ofrecido visitas externas a los animal- arios (de manera fı´sica o virtual) y el 36% realizan char- Durante el primer an˜ o, ma´s de 120 instituciones se las con centros educativos de jo´venes. En el informe han an˜adido al acuerdo. Esta´n representados todos los tambie´n se exponen varias iniciativas exitosas en sectores relacionados con la experimentacio´n con ani- varios centros como ejemplo para otras instituciones. males: centros y organismos pu´blicos de investigacio´n, A partir de este primer informe, las instituciones han las universidades, empresas, parques cientı´ficos, hospi- podido detectar en que puntos todavı´a deben mejorar y tales, asociaciones de pacientes y sociedades cientı´ficas. este es el objetivo para los pro´ximos an˜ os, ayudar a los A principios de 2018, la European Animal Research centros para que mejoren su implementacio´n. Tras el Association (EARA) realizo´de forma independiente e´xito en el Reino Unido y Espan˜ a, este an˜ o el Acuerdo una encuesta a las instituciones para completar el ha comenzado a funcionar tambie´n en Portugal y cuarto compromiso. En el informe obtenido, se analiza Be´lgica. la implicacio´n de cada institucio´n en cada compromiso y se recopilan experiencias, dudas o problemas que han SECAL, Spain surgido durante el primer an˜ o. La preparacio´n de este Informe se ha realizado en colaboracio´n con la COSCE Corresponding author: y SECAL. Sergi Vila, SECAL, Spain. Email: [email protected] El acto de presentacio´n del Informe tuvo lugar el pasado 5 de septiembre en la Residencia de Contributions to the News section are not subject to peer review Estudiantes de Madrid. Con las intervenciones de la and reflect the opinion of the subscribing society. 650 Laboratory Animals 52(6)

Presentation of the Transparency Agreement on the Use 3. Promote initiatives that generate greater knowledge of Animals in Scientific Research in Spain First Annual and understanding by society about the use of ani- Report mals in scientific research. The results of the first report on the Transparency 4. Publish information on the progress of the Agreement on the Use of Animals in Scientific Agreement and share examples of its development. Research of Spain were presented last September. The initiative was proposed in late 2016 on the back of the More than 120 institutions signed up to the Agreement previous successful experience in the United Kingdom, in the first year. All sectors related to animal experimen- and saw the Confederation of Scientific Societies of tation are represented: public research centres and Spain (COSCE) make a proposal to Spanish institu- organisations, universities, companies, science parks, tions that conduct experiments with animals to pro- hospitals, patient associations and scientific societies. mote the transparency of their activities. The Independently, in early 2018, the European Animal Agreement consists of four commitments: Research Association (EARA) ran a survey with the institutions to complete the fourth commitment. The 1. State clearly when, how and why animals are used in resulting report analysed each institution’s involvement research. This is materialised in an institutional dec- with each commitment, and set out the experiences, laration available for consultation on the COSCE questions and problems that emerged during the first website. year. The report was prepared in partnership with 2. Provide adequate information to the media and the COSCE and the Spanish Society for Laboratory general public on the conditions in which research is Animal Sciences (SECAL). carried out using animal models and the results The report was presented at an event held at obtained from it. the Student Residence in Madrid on September 5, Vila 651

2018. With speeches by Dr Margarita del Val (CSIC experimentation in Spain, highlighting to European dir- researcher, COSCE member for Health and Life ectives that they work ethically and responsibly in Sciences and chair of the COSCE Committee on the adapted facilities and in compliance with the principle Study of the Use of Animals in Scientific Research), of the 3Rs; 80% of the institutions have arranged exter- Dr Lluı´s Montoliu (CSIC researcher, COSCE nal visits to their animal facilities (physically or virtu- Committee member and coordinator of the practical ally) and 36% have held talks with schools. The report implementation of the Agreement) and Dr Javier also features several successful initiatives in different Guille´n (on behalf of EARA, member of the AAALAC centres as an example for other institutions. Association for Assessment and Accreditation of The institutions can use this first report to detect Laboratory Animal Care International and COSCE areas where improvement is still needed, and this is Committee member). the goal for the coming years, i.e. to help centres One of the first conclusions of the report is drive enhanced implementation. In the wake of the suc- that transparency in communication has increased cess in the UK and Spain, this year the Agreement has substantially. Most institutions have officially pub- also been started in Portugal and Belgium. lished their institutional declaration and regularly dis- SECAL Annual Report 2018 (English): http://eara. close the use of animals in news stories about the eu/wp-content/uploads/2018/09/Annual-Report2018- scientific breakthroughs they have made. The very Transparency-Agreement-on-the-use-of-animals-in- presentation of the Agreement led to the appearance scientific-research-in-Spain_FINAL.pdf. of several extensive journalistic reports on animal ÁH[LEOH ÀOP DQG ULJLG LVRODWLRQ VROXWLRQV IRU DOO \RXU FRQWDLQPHQW QHHGV

O +ROGLQJ,VRODWRUV

O 6XUJLFDO ,VRODWRUV

O +\SR[LF &KDPEHUV

O :DOO 0RXQWHG 7UDQVIHU +DWFKHV

O 'HVLFFDWRU&DELQHWV

'HVLFFDWRU&DELQHWV :LWK RU :LWKRXW *DV 3RUWV

3RWWHUV/DQH 0LOWRQ.H\QHV 0.+4 WHO ID[ HPDLO VDOHV#S¿V\VWHPVFRXN ZZZS¿V\VWHPVFRXN News Laboratory Animals 2018, Vol. 52(6) 653–654 ! The Author(s) 2018 Article reuse guidelines: Newly created title of the Diplomate sagepub.com/journals- permissions Specialized Veterinarian in Laboratory DOI: 10.1177/0023677218811272 Animal Science (Dipl. SVLAS) awarded by journals.sagepub.com/home/lan the Swiss Association of Veterinarians in Industry and Research (SAVIR)

Maike Heimann

– the veterinary management, hygienic monitoring and/or planning of facilities in which the animal experiments are carried out; – surveillance of animal housing and care before, during and after an experiment, including specific treatments and/or breeding of laboratory animals; – transport of animals at national or international level; – planning, monitoring and execution of animal experiments (study design); – assessment of results and reporting; – consulting, basic, continuing and qualifying train- ing for the personnel involved in the experiments – support of researchers as an animal welfare Good news for veterinarians in experimental animal delegate; research: since summer 2018, the Swiss – members of ethics committees and/or other work- The Association of Veterinarians in Industry and ing groups in the field of laboratory animal Research (SAVIR) awards the newly created title of science. the Diplomate Specialized Veterinarian in Laboratory Animal Science (Dipl. SVLAS). This title, for expert As prerequisites to be accepted as a SVLAS candi- veterinarians in the field of laboratory animal science, date, interested people need to live or work in represents a recognition of professional skills and thus Switzerland and must have a Swiss federal diploma is an upgrading of the profession. The title Dipl. in veterinary medicine (or a diploma recognized in SVLAS is protected under private law by SAVIR. Switzerland). Candidates also must be active SAVIR SAVIR sets the procedure for the program in the members and need to have sufficient practical experi- SVLAS rules. The scope of the rules includes the post- ence in experimental animal research. This includes graduate recognition of veterinarians in the field of adequate basic training for experimenters (LTK laboratory animal science with the aim of covering Module 1 or 20 in Switzerland) and for experimental the basic procedures related to laboratory animals. The basis of this recognition lies less in the differences Swiss Association of Veterinarians in Industry and Research between species but more in the basic principles in the (SAVIR) care and use of laboratory animals in animal facilities. Specialized veterinarians are particularly important Corresponding author: for the: Maike Heimann, Swiss Association of Veterinarians in Industry and Research (SAVIR). Email: [email protected] – development and application of the 3Rs principle and the protection of animals; Contributions to the News section are not subject to peer review – basic veterinary care of laboratory animals; and reflect the opinion of the subscribing society. 654 Laboratory Animals 52(6) animal study directors (LTK Module 2 in Switzerland) Registration is open from now on, and the first can- or equivalent training abroad that is officially recog- didates have already registered for the program. If you nized in Switzerland. In addition to basic training, can- want to learn more about the Dipl. SVLAS title, please didates must have working experience in in vivo have a look on the SVLAS website: http://www.svlas. research for at least 4 years as an experimenter and/ ch. SAVIR was founded in 2002 and supports veterin- or study director in Switzerland or abroad. arians in the fields of business, industry, research and In order to be awarded the title, candidates must education. The purpose of the association is to promote submit a detailed catalogue of competences that pro- the profession of veterinarians in industry and research, vides information about their professional qualifica- to enhance the professional image of industrial veterin- tions and work experience. Deadline for submission is arians and to represent members in their community, to the October 15 of every year. Only complete dossiers committees and organizations of the Swiss and inter- will be considered. If the SVLAS Credential Committee national veterinary community. In addition, deepening recognizes this dossier as sufficient, the candidate will the professional and personal relationships of members be admitted to the examination. Upon passing the amongst each other is intended, as well as promotion of exam, the candidate will be awarded the title Dipl. further education of members. On the occasion of the SVLAS. During the complete process, candidates can newly implemented Dipl. SVLAS title, the website of be supported by a mentor (a person who already holds the association has also been completely revised and Dipl. SVLAS). All title holders are obliged to act as now appears in a more modern design: http://www. mentors for new candidates. savir.ch. LBS - your trusted supplier... Serving the needs of the Biotechnology Industry

LBS Biotechnology is a respected and accomplished Company with a dynamic outlook; We focus on meeting the exacting requirements of the Biotechnology Industry backed up by our commitment to ISO9001:2008 Quality Assurance, customer care and logistics management. We offer an abundant range of quality products to choose from, all carefully developed through our understanding of customers’ needs:-

• Environmental Enrichment • Bedding & Nesting • Animal Diets • Treats & Rewards • Vacuum Packing & Irradiation • Hygiene Products • Specialist Vacuum Cleaners • Protective Clothing & Footwear • Gloves & Disposables • NorayBio Management Software

Contact the experts: Tel: +44 (0)1293 827940 Email: [email protected] www.lbs-biotech.com ...... 656 Calendar of events

Meetings of interest to laboratory animals scientists and technicians: references to Laboratory Animals are for further details. Items for inclusion should be sent to Notes and Comments Editor, LAL, PO Box 373, Eye, Suffolk, IP22 9BS, UK. Email to [email protected]. The deadlines for inclusion of material are: February issue, 10 November; April issue, 10 January; June issue, 10 March; August issue, 10 May; October issue, 10 July; December issue, 10 September. 2018 10 December RSPCA Lay Members’ Forum, London, UK. For further information contact [email protected] 10–11 December Application of non-animal approaches for decision-making in chemical safety assessment, London, UK. For further information visit https://www.nc3rs.org.uk/events/workshop- application-non-animal-approaches-decision-making-chemical-safety-assessment 2019 9–11 January FRAME Training in Experimental Design, Nottingham, UK. For further information visit https://frame.org.uk/zkevent/frame-training-school-university-of-nottingham/ 14–18 January ESLAV/ECLAM Winter School on Systemic Reviews, Dublin Ireland. For further information visit http://eslav-eclam.org/education/eslav-eclam-winter-school-on-systematic-reviews/ systematic-reviews-14th-to-18th-of-january-2019/ 30 January–1 February 2nd Italian Zebrafish Meeting (ZFIM 2019), Pisa, Italy. For further information visit https:// zfim2019.webs.com/ 6–7 February Laboratory Animal Sciences 8th Annual webinar. For further information visit https://www. labroots.com/virtual-event/laboratory-animal-sciences-2019 7–10 April TT2019 (International Society for Transgenic Technologies) - 15th Annual meeting, Kobe, Japan. For further information visit https://www.transtechsociety.org/index.php?src¼news& srctype¼detail&category¼Announcements&refno¼13 9–12 April Institute of Animal Technology (IAT) Congress, North West England, UK. For further informa- tion visit http://www.iat.org.uk/congress 14–19 April Applied In Vitro Toxicology Course (ESTIV/ASCCT), Bucharest, Romania. For further informa- tion visit https://estiv2019.com/ 22–24 May 13th Minipig Research Forum (MRF), Vienna, Austria. For further information visit https:// minipigresearchforum.org/meetings/ 10–13 June 14th FELASA Congress, Prague, Czech Republic. For further information visit http://www. felasa2019.eu/ 13–17 October AALAS National meeting, Denver, Colorado, USA. For further information visit https://www. aalas.org/national-meeting/general-information/future-meetings 26–28 November LASA Annual meeting, Midlands, UK. Further information will be available in due course.

Index to Advertisers DECEMBER 2018

Altromin International OBC FELASA 562 SAFE/J Rettenmaier & So¨hne 561 AnLab Ltd 644 Fine Science Tools GmbH 552 Special Diets Services 641 Avidity Science 564 IPS Product Supplies Ltd 642 Tecniplast SpA 551 Bell Isolation Systems Ltd 557 Tecniplast SpA IBC Laboratory of Pharmacology and CEDS 644 Toxicology GmbH & Co KG 555 ZOONLAB 563 Charles River Laboratories IFC LBS 560, 655 Datesand Ltd 559, 648 Marshall BioResources 558 ECLAM/EBVS 556 PFI Systems Ltd 652

Your global operating experts in lab animal diets.

$VDQΖ62FHUWLȴHGFRPSDQ\ ZLWKRYHU\HDUVRIH[SHULHQFH ZHR΍HU\RX VWDQGDUGL]HGDQLPDOQXWULWLRQDOVROXWLRQV DQGFXVWRPL]HGVSHFLDOGLHWV

altromin Spezialfutter GmbH & Co. KG Im Seelenkamp 20 D- 32791 Lage Made in Germany www.altromin.com International