1

xwDear Editor, All changes required for the proof have been made to this m/s as red text in an otherwise black m/s. The references are largely dealt with though renumbering is still required. The numbers in this text don’t necessarily match the proof numbers which will finally be the correct ones. Regarding the text boxes, thanks for your effort. These aren’t bad. However, their background needs to be a colour (as grey in the proofs): not white as in the page proofs. However, the colour I want is light beige (parchment colour), as in the word m/s I sent to you. This colour was carefully chosen. The final conclusion sentence in each text box and the title need to stand out somehow. I made these a larger font and bold. That works. Something needs to be done. Up to you what. I would also prefer that these text boxes are not surrounded by a line. I tried to look at the supplementary data ( not easy because the link has to be copied over digit by digit (can’t be copy-pasted)) but the link doesn’t work either. So I haven’t seen my supplementary figures. I will need to see these before I can sign the article off. We are definitely going to need a second round of proofs.

A time space translation hypothesis for vertebrate axial patterning

A.J.Durston and K.Zhu.

2

Institute of Biology,University of Leiden.

How vertebrates generate their anterior-posterior axis is a >90 year old unsolved probem. This mechanism clearly works very differently in vertebrates than in Drosophila. Here, we present evidence from the Amphibian Xenopus that a time space translation mechanism underlies initial axial patterning in the trunk part of the axis. We show that a timer in the gastrula’s non organiser mesoderm (NOM) undergoes sequential timed interactions with the Spemann organiser (SO) during gastrulation to generate the spatial axial pattern. Evidence is also presented that this mechanism works via Hox collinearity and that it requires Hox functionality. The NOM timer is putatively Hox temporal collinearity. This generates a spatially collinear axial Hox pattern in the emerging dorsal central nervous system and dorsal paraxial mesoderm. The interactions with the organiser are mediated by a BMP-anti BMP dependent mechanism. Hox functionality is implicated because knocking out the Hox1 paralogue group not only disrupts expression of Hox1 genes but also of the whole spatially collinear axial Hox sequence in the early embryo’s A-P axis. This mechanism and its nature are discussed. The evidence supporting this hypothesis is presented and critically assessed. Strengths and weaknesses, questions, uncertainties and holes in the evidence are identified. Future directions are indicated.

A Time Space Translation Hypothesis For Vertebrate Axial PatterningHow vertebrates make their anterior-posterior (A-P) axis is a >90 year old unsolved problem [1,2]. This question is possibly close to being solved for Drosophila [3], but definitely not for vertebrates .Despite the lack of progress, the investigations have already yielded at least two directly relevant Nobel prizes (Spemann, 1935,[1,2] Lewis, 1995,[4] Nuesslein-Volhard, 1995)[3]) The underlying mechanism clearly works very differently in vertebrates than in Drosophila. Much evidence indicates that timing is involved in generating the vertebrate anterior-posterior (A-P) axis[10-12] Our recent findings support this view (Fig. 1).Fig.1 Here

3

Here, we present evidence that a time space translation (TST) mechanism (ie: a mechanism whereby a temporal sequence of early anterior to late posterior axial gene expression generates the congruent axial spatial sequence) underlies initial vertebrate A-P patterning in the trunk part of the axis. We present evidence that a BMP dependent timer in the gastrula’s non organiser mesoderm (NOM) undergoes sequential timed interactions with the BMP antagonistic Spemann organiser (SO) [1, 2]: a structure absent in Drosophila. The interactions occur during and after gastrulation and generate the spatial axial pattern. We present evidence that this mechanism involves Hox collinearity [4] and that it requires Hox functionality [5]. The timer in the NOM mesoderm appears to be Hox temporal collinearity (an early anterior to late posterior time sequence of Hox expression codes). This generates the congruent spatially collinear axial Hox pattern (the same Hox codes arranged sequentially spatially along the axis) in the emerging dorsal central nervous system and dorsal paraxial mesoderm. Hox temporal collinearity thus generates Hox spatial collinearity. The interactions with the organiser are mediated by a BMP-anti BMP dependent mechanism), [6,7] see below and Fig. 1, Text Box 1, Figs S1A,B. Hox genes thus take a higher place in the A-P patterning cascade in vertebrates than in Drosophila.. Hox functionality is implicated in this mechanism because knocking out the Hox1 paralogue group not only disrupts expression of Hox1 genes but also of the whole more posterior spatially collinear Hox expression sequence in the early embryo’s A-P axis Text Box 2 [5]) Evidence is emerging that this timer, which governs the neck-tail region of the axis is complemented by an earlier anti-BMP dependent time space mechanism that maps out the A-P zones in the head at the stages before gastrulation in a very similar manner (seeText Box 4 ,Fig. S4B, [8,9]). Data from Meinhardt[14, 15, ] and see his article, this volume) confirms that the vertebrate organiser generates an A-P pattern by timed application of dorsalising information. There is thus an integral timer that regulates the development of multiple positional identities at neighbouring positions along the axis at sequential times. This time space translation mechanism clearly continues after the end of gastrulation, due to a continuation of the gastrulation process in the chordaneural hinge and other tissues in the tailbud[16]. We thus think that there is an integrated BMP-anti BMP TST mechanism for the whole vertebrate body axis. The deduced mechanism and its nature are discussed. The evidence supporting the hypothesis is presented and

4

critically assessed. Strengths and weaknesses, questions, uncertainties and holes in the evidence are identified. Future directions are indicated.

What is the evidence that there is a timer and time space translation?Text Box 1 HereSee Text box1 and Figs S1, A,B. for the evidence on which this proposed mechanism was based. In Wacker et al’s expts [6,13], a UV ventralised BMP rich embryo containing no organiser was challenged with an anti BMP source at different times after the beginning of its gastrulation and the resulting embryos were analysed using in situ hybridisation. With no challenge, such an embryo simply makes a mass of ventral tissue (a ‘bauchstuk’, in Spemann’s words) [1]. With an implanted organiser, an A-P axis is regenerated. One signal source used was an implanted natural organiser. An early organiser implanted at the beginning of gastrulation induced a whole A-P body axis [6,; seeTextBox 1, Figs S1A, B.]. Early organisers implanted at increasingly later times after the start of gastrulation in the ventralised embryo gave an increasing deletion from the anterior end of trunk part of the resulting axis with a 6 hr. implantation (end of gastrulation, latest used ) retaining only the most posterior part of the axis. Presumably, an implanted organiser stabilises sequentialiy more and more posterior positional identities from the time it is implanted. Another anti BMP source used (corresponding with the anti BMP signals secreted by the trunk organiser as well as the head organiser) was blastocoel injection of the organiser anti BMP signal noggin protein. Again, later and later treatments were more and more posteriorising. The first treatments (at the blastula stage) induced expression of the head marker Otx2 and the anterior hindbrain marker Krox20. The second at the beginning of gastrulation induced mainly Krox20 and the more posterior marker Hoxb4 .Later treatments than this induced a partial axis either starting with a head (earlier) or starting at a more posterior position (later). These mixed results presumably reflect the fact that noggin has multiple functions. It acts directly to dorsalise ectoderm, converting this to neurectoderm, thus presumably directly stabilising specific positional identities in this tissue. It also converts NOM mesoderm to organiser mesoderm (SO), thus permitting initiation of an axis. These conclusions were confirmed by other studies. Dias et al [17] (fig. 4) found that explanted chicken posterior primitive streak (= very ventral

5

NOM mesoderm+ embryonic ectoderm (EE)) is stabilised to different positional identities by noggin applied at different times after the beginning of gastrulation. Early application stabilises a relatively anterior positional identity. Late application stablises a more posterior positional identity. Mullins and colleagues [8,9] also found that different positional identities are stabilised in the zebrafish embryo by differently timed application of anti BMP (heat shock treatment of Tg(hsp70:chd)). Their studies covered the head part of the axis anterior to the trunk Hox gene expressing section studied by Wacker et al. cf [6] with [9] See Text Box 4, Fig. S4B, [42].There is thus a BMP/anti BMP dependent time space translation mechanism covering the whole vertebrate A-P axis from the anterior head to the tip of the tail. The trunk timer is evidently in NOM mesoderm because addition of this tissue (but no other) to organiser mesoderm and neurectoderm in a recombinate leads to genesis of specific stable positional values in mesoderm and neurectoderm after culture. The above findings demonstrate unambiguously that a BMP- anti BMP dependent time space translation mechanism mediates axial patterning.

Does time- space translation involve the Hox genes?The time space translation mechanism above involves Hox genes [6], [7], [18: see Fig. 1] ). Hox codes (the expression of different specific combinations of Hox genes) seem to be the main determinants of different A-P axial positional values in the part of the axis where we discovered this mechanism (the trunk) [6] The main axial tissues of the trunk (neurectoderm and paraxial mesoderm) each show a spatially collinear sequence of Hox gene expression. The above evidence argues Text Box 2 Herethat these are at least part of the primary axial pattern.(Text box 2, Fig. S2A). The NOM mesoderm of the gastrula, which is the precursor of paraxial and ventral mesoderm and contains the timer, manifests a gastrula stage temporally collinear sequence of Hox expression. This is likely to correspond to or to be driven by the timer in our time space translation mechanism) [6,7]Text Box 2, S2B Temporal collinearity but not spatial collinearity survives in ventralised embryos, paralleling the timer (Text Box 2, S2C). There is no Hox expression at all in dorsalised embryos, again paralleling the effect of absence of the timer in our time space translation mechanism (Fig. 3,not shown). Combining an organiser or a piece of dorsalised mesoderm from a dorsalised Li+ embryo with a ventralised embryo containing temporal collinearity brings

6

back Hox spatial collinearity and the axial pattern as in time space translation (Text Box 2). The function of the organiser is thus apparently to generate spatial collinearity from temporal collinearity. These features of Hox collinearity absolutely parallel the key features of our time space translation mechanism, making a strong case that this is based on Hox collinearity. Hox temporal collinearity is thus needed for generating Hox spatial collinearity via time space translation. Knocking down the entire first Hox paralogue group (Hox 1) knocks down most or all of the spatially collinear Hox sequence (At least back to Hox9), indicating that Hox1 functionality is also involved in this process [5]. A second indication of the importance of Hox function is that knocking down individual Hox genes in NOM mesoderm specifically prevents vertical signalling: copying of the expression of the same Hox genes for NOM to neurectoderm (NE) in ‘wrap’ recombinates, which copies the same process in vivo where vertical signalling is part of TST [47]. These findings give a clue to the mechanism of TST (below). The effects of the Hox1 knock down on temporal collinearity are so far unknown. There are thus strong correlations between temporally and spatially collinear Hox expression and the temporal and spatial phases of the TST mechanism. There are also functionality experiments that show that Hox genes are involved in generating positional values and Hox spatial collinearity in the trunk and that the Hox genes interact [5]. There is also evidence that transfer of positional information from NOM mesoderm to neurectoderm during TST(=vertical signalling) [46] is mediated by specific non cell autonomous autoregulation of individual Hox genes [47]. The Hox genes thus appear to be part of the core mechanism for TST. Specific functionality experiments using Hox genes are required to clarify further how the Hox genes mediate this time- space translation mechanism. For example, it would be desirable to test the effect of specifically disturbing temporal collinearity in the NOM mesoderm on axial patterning. This mechanism regulates axial patterning in the trunk-tail part of the axis: our findings above already argue that the mechanism acts at the level of the Hox genes rather than only upstream of them. See below.

. What are the roles of the organiser and NOM mesoderm in time space translation?The organiser appears to be required to translate the time sequence in the gastrula’s NOM mesoderm (putatively the gastrula’s temporally collinear Hox expression sequence) into the spatial pattern (involving a

7

spatially collinear Hox expression sequence) that appears progressively in axial mesoderm and neurectoderm during gastrulation and later (see Text Box 2, Figs. S2A,B.). Without an organiser, in a ventralised embryo, only the NOM’s transient temporally collinear Hox sequence is seen [6,7](Text box 2, Fig. S2C). There is no spatial axial Hox pattern or spatial collinearity. Without NOM mesoderm (in a dorsalised embryo), there is no Hox expression at all. We assume that, as successively older blocks of NOM mesoderm approach the organiser sequentially during gastrulation, due to convergence extension and involution movements, their current Hox identities are fixed and also copied to the overlying blocks of neurectoderm in the dorsal wall of the gastrula, this neurectoderm already having been induced from ectoderm by organiser Text Box 3 Heresignals. (Text Box3, Fig S3A) The organiser’s role in this was investigated by testing how its role could be substituted. It was found that axial patterning and Hox spatial collinearity in neurectoderm can be achieved by tBr anti BMP mediated dorsalisation of the ectoderm of a UV ventralised embryo only without any need for an organiser or for dorsalisation of the embryo’s mesoderm (Text Box 3, FigS3B) [19] . This suggests that an axial pattern can be achieved simply by organiser induction of ectoderm to neurectoderm only without the need for organiser action on mesoderm)Fig S3B. However, Dias et. al17] have also shown recently that chicken posterior (=ventral) primitive streak tissue can be induced to expression of different stable Hox codes by application of the organiser anti BMP signal noggin at different times during gastrulation,with later times giving more posterior Hox codes. (Text Box 3, S3C) This result opens the possibility that there is also a direct action of the organiser (SO) to stabilise Hox codes in NOM mesoderm (although this ventral streak tissue will also include embryonic ectoderm (EE)). There are still clearly unanswered questions about the role of the organiser in A-P patterning. One role of the NOM is rather clearly to transmit posterior information to neuralised ectoderm. This function involves precise copying of positional information from NOM mesoderm (vertical signalling) exactly as predicted by Mangold [46]. The mechanism involves very specific non cell autonomous autoregulation of individual Hox genes such that their expression in NOM mesoderm is copied to neighbouring neurectoderm (Text Box 3) [47].

8

What is the molecular nature of the A-P timer and of Hox temporal collinearity?An obvious possibility for the timer is that this involves Hox temporal collinearity (see Text box 2, Fig S 2B, C ) The temporal sequence observed in NOM mesoderm involves a temporally collinear sequence of Hox expression just as the axial pattern in axial mesoderm and axial neurectoderm involves a spatially collinear axial sequence of Hox expression. (Texf Box 2, Fig. S2A) On the other hand, it has recently been proposed (without evidence ) that Hox temporal and spatial collinearity are not connected [20] The temporal Hox sequence seen during gastrulation is the classical example of Hox temporal collinearity.. It has been proposed that this and the other examples of Hox temporal collinearity depend on progressive 3’ to 5’ opening of the Hox clusters for transcription [20]. There is evidence that this occurs but it is not the whole story for Hox temporal collinearity and timing during gastrulation. This involves synchronisation of the structurally different Hox clusters which must involve trans interactions within cells and intercellular interactions between different cells in the synchronised mesoderm of the Text Box 4 Heregastrula [21, 22] ( Text box 4, Fig. S4A). The Hox temporal sequence is also part of a larger A-P axial sequence including non Hox homeobox genes in the head, again arguing against Hox cluster opening as a global mechanism [9]( Text box 4, Fig.S4B). Some but not all of the anterior genes are para Hox genes. See below. We note also that Hox1 functionality is clearly involved in generating the early spatially collinear axial Hox sequence (McNulty et al., 2006)[5] What is possible is that gastrula temporal collinearity involves collinear interactions among Hox genes (Text Box 4, S4C). Two interactions could putatively play a role. Posterior prevalence, where posterior Hox genes inhibit expression and function of more anterior ones [4,23], [21,22,24,25}. Posterior induction, where anterior Hox genes induce expression of more posterior ones. [5, 21 ,25]. These interactions are shown in Text Box 4, S4C. They are known to occur in the early vertebrate embryo, but there is no evidence so far that they are the basis of Hox temporal collinearity or of the A-P patterning timer.

What is upstream? Does Hox collinearity drive the axial patterning clock or is it driven by it? In the arguments above we argue that time space translation occurs partly at the level of the Hox genes. There may also be upstream inputs

9

involved in this process. We argue above that Hox collinearity, based on collinear interactions between the Hox genes is the basis for time space translation. This is possible. There are also other suggestions. In addition to the proposition that upstream regulation is by progressive opening of Hox clusters for transcription, it has been proposed for the anterior timer that BMP regulated Smads interact at the protein phosphorylation level with classical A-P patterning pathways: FGFs, Wnts and retinoids.[9]. There is also evidence that the somitogenesis timer is upstream of Hox genes [42:Text box 5] and regulates temporal collinearity (see below). Upstream regulators of the timed anterior early axial patterning genes have also been identified in the neurectoderm [26].We should also bear in mind that this is not the first proposal of a mechanism for vertebrate A-P patterning. Previous proposals have involved action of the Wnt, retinoid and FGF signalling pathways. Refs. in [42]. A central question is: how do these pathways act? We should bear in mind that A-P patterning is clearly a long and complex process and that we are dealing here with the very first steps. Perhaps these signalling pathways are involved in another part of the process, either later or concurrently in a parallel mechanism. However, they could also presumably be involved in time space translation. This process would involve what one could call ‘decision points’, in time and spatially, on the axis, where cells switch their regulation by different pathways and factors anterior to and before as opposed to posterior to and after their decision. There are clearly at least two such decision points on the A-P axis. At least one of these is mobile. It moves posteriorly along the axis during the course of development. Details of these’decision points’ are given in my ‘Introductory’ article [42].

Connections between the Hox axial patterning timer and the somitogenesis clockA very interesting upstream regulator is the somitogenesis clock. What is upstream of most precisely timed biological events like Hox temporal collinearity is a biological clock: a biological oscillator where phase specific events are repeated regularly. Mostly many times.. Examples are: The cicrcadian rhythm, The heartbeat, the cell division cycle, Dictyostelium signalling oscillations. There is a biological clock that runs concurrently with gastrulation and Hox temporal collinearity. This is the somitogenesis clock. Oscillations in gene expression in paraxial presomitic mesoderm (derived from NOM) are used to generate spatially periodic somites via time space translation [27,28,29]. This oscillator

10

should be capable of adding precision to Hox temporal collinearity. There are data that show that the somitogenesis clock is indeed upstream of Hox temporal collinearity.These data are discussed in detail in our introductory article in this volume [42] The main points are also summarised in Text box 5 TheText Box 5 Herefindings demonstrate a clear but complex connection between Hox temporal collinearity and the somitogenesis clock.

Summary and conclusions The mechanism underlying origin and patterning of the vertebrate anterior-posterior (A-P) axis is a very old long standing problem that has already earned two directly relevant Nobel prizes,[1,2,3,4] but is still unsolved. This problem is possibly close to being solved in Drosophila [3] but not yet in vertebrates, which clearly use a different mechanism. There is however light at the end of the tunnel. We summarise our recent findings concerning a BMP dependent axial time space translation (TST) mechanism for vertebrate A-P patterning. Proof is presented that: This mechanism exists and that it involves a BMPdependent axial timer, Sequential time/position values from this timer are stabilised by anti-BMP signals from the Spemann organiser during gastrulation, thus progressively generating an axial pattern. This finding emphasises the importance of Spemann’s Nobel prize winning organiser [1,2]. Further investigation showed that this mechanism involves the Hox genes: the timer involves early Hox temporal collinearity and the organiser controls sequential transitions from Hox temporal collinearity to expression zones in a spatially collinear Hox sequence, leading to patterning of the trunk-tail part of the axis. There is evidence that this mechanism operates at the level of the Hox genes rather than entirely upstream of them. Lewis’s Nobel prize winning discovery of Hox collinearity [4] was central to this mechanism. This mechanism connects to a very similar TST mechanism in the anterior non hox part of the axis. Non Hox homeobox genes specifying different A-P levels in the anterior brain have their expression stabilised sequentially by early anti-BMP signals, in the blastula and early gastrula stages before the Hox timer starts. This mechanism and the Hox TST mechanism seem to be parts of the same continuum. Mullins and co-workers argue that the anterior BMP dependent timer is driven by upstream interactions between Smad 1 and the FGF , Wnt and retinoid

11

signalling pathways which are known to regulate A-P patterning. This is one of several possibilities. There are clear but complex connections between the somitogenesis timer and axial patterning at least in the trunk-tail (Hox) part of the axis. This timer is necessary but not sufficient for axial Hox expression.An important question is: what are the roles of signalling pathways known to regulate A-P patterning. On the one hand, these are likely to be involved in: other patterning mechanisms that complement the above TST mechanism, possibly acting at different times and/or mapping out details-as in the hindbrain. On the other they may have roles in the A-P TST mechanism itself. One likely role is to act at ‘decision points’, separating different time space domains on the TST mechanism. There se st least two such decision points. The first reflects a boundary between the positive and negative action of retinoids and falls close to the boundary between Mullins’ anterior part of the TST and the Hox posterior part. Probably this decision point is between Otx2 and Gbx2. The second decision point (DP) falls initially between Hox5 and Hox6 and reflects a difference between retinoid sensitivity and FGF sensitivity. This DP may be mobile, moving posteriorly with time. In conclusion,some progress has been made with elucidating the mechanism for vertebrate A-P patterning.

References1. Spemann. H,. und Mangold, H. (1924). Über die Induktion von Embryonalanlagen durch Implantation artfremder Organisatoren. W. Roux’ Arch. f. Entw. d. Organis. u. mikrosk. Anat. 100: 599-6382. Spemann H (1935)The Organizer-Effect in Embryonic Development Nobel Lectures, Physiology or Medicine 1922-1941, Elsevier Publishing Company, Amsterdam, 19653. Nuesslein Volhard C. (1995) Nobel Lecture: The Identification of Genes Controlling Development in Flies and Fishes. From Nobel Lectures, Physiology or Medicine 1991-1995, Editor Nils Ringertz, World Scientific Publishing Co., Singapore, 1997 4. Lewis, E.B. (1995) Nobel lecture. The bithorax complex: The first fifty years. In: Lifshitz, H.; Ed. Genes, Development and Cancer. The Life and Work of Edward B. Lewis, Kluwer Academic Publishers: Norwell, MA,5. McNulty, C, Peres, J, van den Akker, W, Bardine, N, and Durston A (2005) Knockdown of the complete Hox paralogous group 1 leads to

12

dramatic hindbrain and neural crest defects. Development. Jun;132(12):2861-716. Wacker SA, Jansen HJ, McNulty CL, Houtzager E, Durston AJ. (2004B)Timed interactions between the Hox expressing non-organiser mesoderm and the Spemann organiser generate positional information during vertebrate gastrulation. Dev Biol.;268(1):207-19.7. Durston A, Jansen HJ, Wacker SA. (2010) Review: Time-Space Translation Regulates Trunk Axial Patterning In The Early Vertebrate Embryo.  Genomics.  2010; 95:250–255.8Tucker JA1, Mintzer KA, Mullins MC The BMP signaling gradient patterns dorsoventral tissues in a temporally progressive manner along the anteroposterior axis.Dev Cell. 2008 Jan;14(1):108-19. doi: 10.1016/j.devcel.2007.11.004.9. Hashiguchi M. and Mullins MC (2013) Anteroposterior and dorsoventral patterning are coordinated by an identical patterning clock Development 140, 1970-1980 (2013)10. Nieuwkoop, P.D., 1952. Activation and organisation of the central nervous system in amphibians, III. Synthesis of a new working hypothesis. J. Exp. Zool. 120, 83– 108. 11. H. Eyal-Gilad Dynamic Aspects of Neural Induction in Amphibia 1954 in: ‘Amphibians’ published by H. Vaillant-Carmanne, - 81 pages12 Vasiliauskas, D., and Stern, C.D. (2001). Patterning the embryonic axis: FGF signaling and how vertebrate embryos measure time. Cell 106, 133-136.13. Wacker SA, McNulty CL, Durston AJ. (2004A) The initiation of Hox gene expression in Xenopus laevis is controlled by Brachyury and BMP-4. Dev Biol. 2004 ;266(1):123-37.14 Hans Meinhardt, 2002 The radial-symmetric hydra and the evolution of the bilateral body plan: an old body became a young brain BioEssays Volume 24, Issue 2, pages 185–191,  15. Hans Meinhardt 2006 The primary body axes of vertebrates: the generation of a near-Cartesian coordinate system and the role of Spemann-type organizer Developmental Dynamics Vol. 235, pp 2907-291916 Gont LK, Steinbeisser H, Blumberg B,* and De Robertis E.M., Tail formation as a continuation of gastrulation: the multiple cell populations of the Xenopus tailbud derive from the late blastopore lip Development 119, 991-1004 (1993)

13

17. Dias AS,Irene de Almeida1,†, Julio M. Belmonte2, James A. Glazier2 and Claudio D. Stern (2014) Somites without a clock Science 10.1126.1247575.18. Durston AJ,*,1 S Wacker,2 N Bardine,1 and HJ Jansen 2012a Time Space Translation: A Hox Mechanism for Vertebrate A-P Patterning Curr Genomics.; 13(4): 300–307.19 Jansen HJ, Wacker SA, Bardine N, Durston AJ. (2007) The role of the Spemann organizer in anterior-posterior patterning of the trunk. Mech Dev. Sep-Oct;124(9- 10):668-81.20. Daan Noordermeer, Marion Leleu, Patrick Schorderet, Elisabeth Joye, Fabienne Chabaud, Denis Duboule 2014 Temporal dynamics and developmental memory of 3D chromatin architecture at Hox gene loci eLife;10.7554.21. Durston AJ,, Hans J. Jansen, Paul In der Rieden and Michiel H. W. Hooiveld (2011) Hox collinearity – a new perspective. Int. J. Dev. Biol Biol. 55: 899-90822. Durston AJ (2012b) Global posterior prevalence is unique to vertebrates: A dance to the music of time? Dev Dyn. 2012 Aug 28. doi: 10.1002/dvdy.23852. [Epub ahead of print].23. EB Lewis E.B (1978) A gene complex controlling segmentation in  Drosophila . Nature 276, 565 – 570.24 Woltering J. M., and Durston A, (2008) MiR10 represses HoxB1a and HoxB3a in Zebrafish. PLoS ONE. 2008 Jan 2;3(1):e1396.25 Hooiveld M, Morgan R, In der Rieden P, Houtzager E, Pannese M, Damen K, Boncinelli E, Durston A. Novel colinear interactions between vertebrate Hox genes. Int. J. Dev. Biol. 1999;43:665–67426. Gamse JT, Sive H. (2001) Early anteroposterior division of the presumptive neurectoderm in Xenopus. Mechanisms of Development 104 (2001) 21±3627 Isabel Palmeirim,* Domingos Henrique,†§ David Ish-Horowicz,† and Olivier Pourquie´ ‡Avian hairy Gene Expression Identifies a Molecular Clock Linked to Vertebrate Segmentation and Somitogenesis Cell, Vol. 91, 639–648, November 28, 1997 28. Ann C. Burke, Craig E. Nelson, Bruce A. Morgan* and Cliff Tabin Hox genes and the evolution of vertebrate axial morphology. Development 121, 333-346 (1995)29. Wellik, D. M.; Capecchi, M. R., Hox10 and Hox11 genes are required to globally pattern the mammalian skeleton. Science 2003, 301 (5631), 363‐7.

14

30. Godsave S, Dekker EJ, Holling T, Pannese M, Boncinelli E, Durston A. (1994) Expression patterns of Hoxb genes in the Xenopus embryo suggest roles in anteroposterior specification of the hindbrain and in dorsoventral patterning of the mesoderm Dev Biol. 166(2):465-76.31. Zakany, J.; Kmita, M.; Alarcon, P.; De La Pompa, L.; Duboule, D. (2001) Localized and transient transcription of Hox genes suggests a link between patterning and the segmentation clock. Cell, 106(2), 207-217.32. Peres J., McNulty C., and Durston A. .(2006) Interaction between X-Delta-2 and Hox genes regulates segmentation and patterning of the anteroposterior axis Mech Dev. Apr;123(4):321-33.33. Riedel-Kruse I H, Mueller C and Oates A C (2007) Synchrony dynamics during initiation, failure and rescue of the segmentation clock. Science 317, 1911.34. Cooke J., and Zeeman E. C. (1976) A Clock and Wavefront Model for Control of the Number of Repeated Structures during Animal Morphogenesis J. theor. Biol. 58, 455-47635. Lauschke V.M., Tsiairis, CD, François P and Aulehla A. Scaling of embryonic patterning based on phase-gradient encoding Nature 493, 101–105 2013.36. Soroldoni, D; Jörg, DJ; Morelli, LG; Richmond, DL; Schindelin, J; Jülicher, F and Oates, AC (2014) A Doppler effect in embryonic pattern formation. Science 345, 222-225 .37. Schröter, C and Oates, AC (2010) Segment number and regional identity in a segmentation clock period mutant. Current Biology 20, 1254-125838 Muhr, J., Jessell, T.M., Edlund, T., 1997. Assignment of early caudal identity to neural plate cells by a signal from caudal paraxial mesoderm. Neuron 19, 487–502.39. Alexandre, D., Clarke, J.D., Oxtoby, E., Yan, Y.L., Jowett, T., Holder, N., 1996. Ectopic expression of Hoxa-1 in the zebrafish alters the fate of the mandibular arch neural crest and phenocopies a retinoic acid-induced phenotype. Development 122, 735– 746.40. Gaunt, S.J., Strachan, L., 1996. Temporal colinearity in expression of anterior Hox genes in developing chick embryos. Dev. Dyn. 207,270– 28041. Deschamps, J., van den Akker, E., Forlani, S., De Graaff, W., Oosterveen, T., Roelen, B., Roelfsema, J., 1999. Initiation, establishment and maintenance of Hox gene expression patterns in the mouse. Int. J. Dev. Biol. 43, 635– 650

15

42. Durston AJ. Time, space and the body axis. This volume.43. Soroldoni, D; Jörg, DJ; Morelli, LG; Richmond, DL; Schindelin, J; Jülicher, F and Oates, AC (2014) A Doppler effect in embryonic pattern formation. Science 345, 222-225 .46 Mangold O. (1933) Uber die Induktionsfahigkeit der verschiedenen Bezirke der Neurula von Urodelen. Naturwissenschaften 21: 761– 766.47 Bardine N., Lamers G., Wacker S. et al., 2014 Vertical signalling involves transmission of Hox information from gastrula mesoderm to neurectoderm. PLoS ONE 16 Dec 2014 9(12): e115208. doi:10.1371/journal.pone.0115208

Figure LegendsFig. 1. The time space translation hypothesis (Wacker et al., 2004, Durston et al., 2010). (A) False colour representation of expression of an early anterior to late posterior sequence of three axial markers (Hox genes) during Xenopus gastrulation. WISH on external lateral views of sibling embryos for Hoxd-1 (purple), Hoxc-6 (green), Hoxb-9 (red). Digital images were analysed and selected areas labelled with respective false colour and combined in one image. Six gastrula stages (10.5, 11, 11.5, 12, 12.5 and 13) are shown in a lateral external view, anterior up and dorsal to the right. Anterior boundaries of the Hox expression at the end of gastrulation are arrowed. (B) The time space translation hypothesis. Lateral views. Time sequences of lateral views of gastrulae through gastrulation Expression of new A-P markers is initiated in non-organiser mesoderm (NOM) at sequential times (a time sequence of more and more posterior Hox codes in NOM is represented by a spectral colour sequence of differently coloured horizontal bars) Non-organiser mesodermal tissue (depicted by a horizontal coloured bar which is a 2D representation of the 3D broken ring of Hox expression in the marginal zone of the wall of the embryo) moves (flows) toward the Spemann organiser by convergence and then extends anteriorly (arrow). The NOM mesoderm (lM), adjacent to the Spemann organiser involutes and its current A-P positional value (=Hox code) is then transferred to overlying neurectoderm (NE). While the early temporal Hox sequence in the non-organiser mesoderm (differently coloured horizontal bars; outlined by continuous black line in rightmost figures of B and C) is running, cohorts of new cells from this region are continually moved into the range of Spemann organiser (range represented by dashed black line) and their

16

Hox code is then stabilised by an organiser signal. The temporal Hox sequence is thus converted into a spatial AP pattern by continuous morphogenetic movement and stabilisation of timed information by the organiser both in involuted NOM mesoderm (IM) and overlying neurectoderm (NE). The section represents a paraxial level just lateral to the organiser so the organiser is not visible.(C) The time space translation hypothesis: Dorsal views. In non-organiser mesodermal (NOM) cells, the Hox sequence is running (differently coloured bars, solid black outline in rightmost figures). From this domain, cells are continuously moved into the influence of the Spemann organiser (dashed blackoutline) by convergence and extension (arrows). The AP pattern arises by sequential posterior addition of new stabilised NOM segments each expressing a different subset of Hox genes... A, anterior; P, posterior; V, ventral; D, dorsal; L, left; R right. The outer neurectoderm is not shown in this figure because this section is internal, at the level of the dorsal mesoderm. (D) Time space translation occurs in all vertebrates. Schematic diagrams depicting locations of Spemann organiser, blastopore and initial Hox expression domain in Xenopus and orthologous structures in the zebrafish [39], the chick[40] and the mouse [41] all shown at the beginning of gastrulation. Zebrafish and Xenopus are shown in vegetal views, chick and mouse are shown in dorsal views.

Figures

Fig.1

17

Text BoxesText Box 1: Evidence for BMP/antiBMP dependent TSTText Box 2: Evidence that the TST mechanism acts by a transition from Hox temporal collinearity to spatial collinearity.Text Box 3: Role of the organiser and NOM mesodermText Box 4: Evidence for the mechanism of temporal collinearityText Box 5: Temporal collinearity and the somitogenesis clock

18

Text Box 1: Evidence for BMP-anti BMP dependent time space translation (TST) in the Xenopus gastrula embryo1/ A ventralised embryo (VE) makes no A-P axis. Just a mass of ventral tissue. Adding a Spemann organiser (SO) to a VE causes it to make an A-P axis. One organiser: one axis, two organisers two axes etc.. Fig. S1A [6, 7].2/ Adding an SO to a VE at different time intervals after the beginning of gastrulation causes rescue of different parts of the A-P axis. The earlier SO is added the bigger the axial section rescued and the more anteriorly it starts. All sections run to the tail. Adding the SO anti BMP signal noggin protein to the blastocoel of a VE at different times after the beginning of gastrulation rescues different zones in the axis. Early on: head zones including Otx2., Later: Hoxd1 Later: Krox20, later still: multiple zones representing more and more posterior sections of the axis. Fig. S1B [6. 7]3/ These results indicate an A-P timer in the VE that imposes incrementally more posterior positional values (PV) at incrementally later times. and a fixation mechanism, imposed by the SO, that can stabilise the current PV. 4/ The timer is in NOM mesoderm. The classical literature shows that mesoderm sets up the A-P pattern by progressively transmitting more and more posterior information to neurectoderm. (activation-transformation) [10, 11]. The literature shows that trunk A-P information (transformation) comes from non organiser mesoderm (NOM) whereas neuralisation signals that induce neurectoderm (activation) come from the organiser (SO) . In line with this, NOM is required, together with SO and ectoderm in wrap recombinates for the ectoderm to be able to adopt a positional value (express a Hox gene). Fig. S3A[6. 7] 5/ Conclusion: Axial patterning of the trunk in Xenopus is mediated by BMP/antiBMP dependent TST.

19

Text Box 2: Evidence that time space translation(TST) is mediated by Hox temporal and spatial collinearity.1/ Hox genes are involved in determining A-P identity in the trunk . Evidence: Hox loss of function (LOF) and gain of function (GOF) phenotypes. Loss of function of a Drosophila Hox gene or vertebrate Hox paralogue group (pg) dramatically changes identity of part of the axis. This effect is not necessarily restricted to a small part of the axis. For example: in Xenopus Hox1pg LOF . This downregulates posterior Hox genes as well as Hox1. Hox1 is thus neccessary for collinearity. Presumably due to interactions between Hox genes. [Ref 6] Hox interactions: [Refs.5, 6, 18, 21-25,] 2/ Hox genes show collinearity: Congruence of spatial and temporal early expression sequences with genomic sequence in Hox clusters (4 clusters in vertebrates, 1 in Drosophila). In vertebrates, spatial collinearity (SC) is first seen in neurectoderm (NE) and dorsal paraxial mesoderm (PM), from end gastrulation (St. 12 in Xenopus). Temporal collinearity (TC) is seen ealier (mid-gastrula: st. 10.5, and through gastrulation, in non organiser mesoderm (NOM). Changes in the Hox expression pattern during gastrulation indicate that the later SC pattern is generated from earlier TC, as shown in Fig. 1. Part of the NOM becomes PM following its dorsalisation via convergence extension movements during involution. Zones of NOM/ PM are then apparently frozen at their current involuted AP identities in an early anterior to late posterior sequence. The spatial sequence of NOM/ PM zones becomes mirrored by an identical sequence of zones in the neighbouring NE (in the dorsal outer wall of the the late gastrula). Fig. 1, Fig S2 A, B [6]3/ Ventralised (UV) embryos have only NOM mesoderm (no SO) and have the TST timer and normal Hox TC in their NOM. They have no stable Hox expression and no SC. Their TC is transient and later UV embryos have no Hox expression. Dorsalised (Li+) embryos have no NOM mesoderm (only SO) and no timer and no Hox expression (early or late). They have SO mesoderm and NE. Adding an organiser to a ventralised embryo generates an axial pattern of stable Hox expression. Fig. S2C [6]4/ SC thus correlates 100% with the TST axial pattern., TC correlates 100% with the TST timer. 4/ In wrap recombinates, non organiser mesoderm (NOM), Spemann organiser mesoderm SO and animal cap ectoderm (EE) are required to enable expression of Hox genes. These markers are then most strongly expressed in the neurectoderm (NE) that develops in the recombinate. NE but not EE can express Hox genes. Omitting any of NOM, SO or EE blocks Hox expression. These recombinates mimic the in vivo situation. EE is the tissue from which NE is induced. SO provides the signals that induce it (neuralisation). The Hox expressing NOM then induces Hox expression in NE. Fig S3A [6] 5/ Tentative conclusion: The results above indicate that axial TST is mediated by Hox genes in the trunk. The timer appears to be Hox temporal collinearity. The trunk’s A-P sequence is based on Hox spatial collinearity. The trunk’s temporal collinearity generates spatial collinearity during gastrulation. TST is thus mediated by Hox genes. It is not upstream of them.  

20

Text Box 3. The roles of the organiser (SO) and of non organiser mesoderm (NOM) in generating stable positional information during TST. 1/ The organiser (SO) generates activation signals [10]. It induces embryonic ectoderm (EE) to become anterior neurectoderm (NE) via anti BMP signals. The NOM generates transformation signals. It generates positional information by progressively posteriorising neurectoderm [10, 11. 38]. If either SO or NOM mesoderm are combined withn EE, in a ‘wrap’ recombinate, no Hox gene expression is induced in the NE or EE. If both are included, Hox genes are induced. This is shown for Hoxd1. Fig. S3A [6, 7].2/ The role of SO was tested by determining what could replace it. If EE was activated or neuralised by treating it with an anti BMP treatment (tBr), and then combined with NOM but not SO in s ‘wrap’, NE Hox gene expression was induced . The role of SO is thus possibly simply to activate/neuralise EE. This fact ndicates the importance of stabilisation of Hox expression in NE. The same result is obtained in whole UV embryos, where tBr/FGF has no effect on NOM (which stays NOM and doesn’t become SO), but it dorsalises (activates/neuralises) all of the EE. This dorsalised EE=NE then develops a radially symmetric anteriorised axial Hox pattern. This again emphasises the importance of stabilisation of Hox expression in NE Fig. S3B [19]3/ Very posterior primitive streak from a donor chicken embryo (= very venrtral NOM+ EE) was transplanted to the edge of the blastodisc of a host embryo and treated with the anti-BMP organiser factor noggin. If taken from an early gastrula (st.5), this later expressed only Hoxb3 and Hoxb4. If taken from a older gastrula (St. 7), this also later expressed Hoxb6 and Hoxb9. This finding shows that Hox expression is stabilised in a timed fashion by anti BMP in very posterior primitive streak (which is presumably equivalent to NOM +EE). This raises the possibility that stabilisation of positional information also occurs in NOM mesoderm (so far only in chicken). Fig S3C [17] 4/ Blocking the function of a single Hox gene in NOM prevents induction of the same Hox gene in NE in a wrap recombinate. This indicates that a function of NOM is to posteriorise NE via non cell autonomous autoregulation of individual Hox genes [47].  

5/ Conclusion: Induction of neurectoderm (NE) is important for the organiser’s stabilisation role, presumably because NE but not EE can stably express Hox genes. Stabilisation of Hox expression in NOM mesoderm also seems important. An important role of NOM is to posteriorise neurectoderm via Hox autoregulation.

21

Text Box 4. Evidence bearing on the mechanism of the timer and of Hox temporal collinearity.1/ The temporal and spatial sequences in which Hox genes are expressed during early development both correlate with their 3’ to 5’ genomic sequences in Hox complexes. Temporal collinearity (TC) leads to spatial collinearity (SC) via time space translation (TST). It is interesting to understand the mechanism of temporal collinearity- the mainspring of vertebrate axial patterning. 2/ One idea is that Hox clusters open for transcription from 3’ to 5’. This is attractive because it connects genomic order with ordered expression. There are alternative explanations. There is evidence supporting ‘’opening’’ but it is not the whole story. 3/ Regarding SC: Some organisms that have this have dispersed Hox genes (not clustered although presumably derived from a clustered ancestral sequence). In these cases, cluster opening does not apply. 4/ We demonstrated that genes from each of the four Xenopus Hox cluster types (a, b, c, d) are integrated into a common sequence both during TC and during SC. The clusters are thus synchronised duringTC. Trans interactions and cell interactions are indicated..Fig. S4A [18, 21].5/(The (trunk limited) Hox TC sequence is part of a wider pan axial anti-BMP dependent timing sequence There is congruence between results on anterior TST [8,9] and our Hox results[6,7]. There is thus an integrated time space translation (TST) mechanism for the entire A-P axis (not confined to Hox clusters).Fig. S4B [42]6/ / Loss of function (LOF)for the Hox1 paralogue group indicated that Hox1 functionality is required for Hox collinearity [5]. Hox1 LOF blocked expression not only of Hox1 genes but also of all other more posterior Hox genes examined back to Hoxb9. What is the nature of the relevant Hox function? Collinear interactions between Hox genes and Hox associated Mirs are likely.. Posterior prevalence [4,21, 22, 24, 25] ; inhibition of the expression or function of more anterior Hox genes by more posterior Hox genes and Mirs (the original collinear property associated with Hox collinearity by Lewis). Posterior induction (induction of expression of more posterior genes by more anterior ones)[5,21, 22, 25] Both occur during vertebrate gastrulation. Fig. S4C[18] 5/ Conclusions: The Hox clusters are synchronised during gastrula temporal collinearity (TC). The Hox sequence is continuous with a time- space TST sequence in the head. Hox functionality is also required for collinearity. These findings argue for a TC / timer mechanism based on collinear interactions.

22

Supplementary Figure Legends

Text Box 5. What is upstream of temporally collinear Hox expression in NOM? The role of the somitogenesis clock. 1/ Upstream of most precisely timed biological events like Hox temporal collinearity (TC) is a biological clock: a biological limit cycle oscillator where phase specific events are repeated regularly. Mostly many times.. Examples are: cicrcadian rhythm, heartbeat, cell division cycle, Dictyostelium signalling oscillations. There is a biological clock that runs concurrently with gastrulation and HoxTC. This is the somitogenesis clock (SOC). Oscillations in gene expression are used to generate spatially periodic somites via time space translation (TST). This clock is considered primitive because it is not temperature compensated but the events it interacts with have the same temperature dependence as it does so there is natural compensation. The SOC should be capable of adding precision to HoxTC. There are data that show that SOC is upstream of Hox TC [30-37, 42]. There are other signalling pathways upstream of A-P patterning and Hox expression. Notably, retinoids, FGF’s and Wnt’s. These may act during TST, in regulating decision points (DC) on the axis [42]. 2/ Somite boundaries can be repeated anterior boundaries for the early mesodermal Hox expression [30]. These boundaries repeat over a limited part of the axis. Somitogenesis provides extra precision, determining where Hox genes can be expressed (There are also repeated dynamic Hox boundaries in presomitic mesoderm (psm), a NOM derivative, indicating that boundary formation is due to the SOC, and precedes somitogenesis[ [31,42]3/ We expect that an oscillating SOC gene is upstream of Hox genes. This has been shown for RBPJk and XDelta2. [31,32] The case of XDelta2 is interesting. This is upstream of at least 3 Hox genes: Hoxd1, Hoxc6, Hoxb9, during gastrulation, at a time when there is already TC but no somites or somitomeres. XDelta2 is expressed in NOM from the beginning of gastrulation and is already in an oscillating SOC ( because it generates 3 somitomeres involving spatially periodic XDelta2 expression by end gastrulation) [32]. There is thus congruence of TC and SOC in the same tissue and at the same time before there is any spatial evidence of somitogenesis. [42] Interestingly, although XDelta2 is neccessary for Hox expression, it is not sufficient [32] There are evidently multiple inputs on Hox genes.4/ The relation between Hox- A-P patterning and somitogenesis is not simple. Both the A-P pattern and somitogenesis are regulative. The AP pattern readjusts in various ways for example if an embryo is cut up by microsurgery.. The somite pattern has also been claimed to show scaling: to readjust somite number if body length is altered [35]. This is however disputed [43]). Somitogenesis scaling and axial pattern regulation are not closely coupled. In a zebrafish SOC period mutant the axial pattern stays normal (regulates) while somite number is reduced. It seems possible that the coupling of somitogenesis and axial patterning is limited to a few points on the axis. For example Hoxd1 could be directly coupled to somitogenesis while other more posterior somitogenesis dependent Hox genes are not directly coupled but are instead are coupled via Hoxd1 [42]  5/ Conclusions: Somitogenesis oscillations are upstream of Hox temporal collinearity. A-P pattern regulation and somitogenesis scaling can be dissociated indicating a complex connection.

23

Fig S1-2: The evidence on which the time space translation hypothesis was based: Fig. S1: Evidence for time space translation Fig. S1A The organiser stabilises A-P information. A normal gastrula organiser can rescue the body axis if transplanted into a ventralised (eg. UV) embryo at the gastrula stage. A ventralised UV embryo makes no body axis- just a cell mass containing some ventral organs. The rescued embryos have a normal body axis and normal zones of A-P marker expression. .Far left: plan of the experiment. A wild type organiser from an early (St 10) wild type gastrula is transplanted into the marginal zone of a St. 10 UV embryo. Next right: above: control embryo, stained for in situ of Krox20 (rhombomeres 3 and 5 in the hindbrain) and Hoxb4 (posterior hindbarian level). Both markers are expressed as expected. Below: UV embryo. Neither marker is expressed. Next right: normal organisers transplanted from st. 10 into St. 10 UV embryos. . Above: Krox20 and Hoxb4: normal staining is as in normal control embryos. Below: Krox20 and Hoxa7: normal staining. Far right, above: Krox20 and Hoxc6: normal staining. Far right below: Krox20 and Hoxd13: normal stainng. Arrows mark the anterior stripe of Krox20 expression and the anterior Hox expression boundaries. The arrows mark the anterior expression boundaries of the two markers in each case. Conclusion: implanting a normal organiser (SO) into a UV embryo, containing NOM can generate a normal A-P axis with a normal axial pattern, containing normal stable zones of expression of all of the A-P markers tested.Fig. S1B. Organiser transplants show that A-P patterning occurs by a timing mechanism. Timed interactions between the Spemann organiser (SO) and the non-organiser mesoderm (NOM) reveal a time space translation mechanism (TST). Left: Ageing the non-organiser mesoderm (isolated in UV embryos). A ventralised UV embryo with no implant (UV), an untreated control embryo (con), and recombinations of organiser mesoderm from stage 10 (0h SO) with ventralised embryos of different ages after the beginning of gastrulation (0h, 2h, 4h, 6h,) . Embryos are positioned with their head up and dorsal to the right. They were analysed with WISH using axial markers as above, including En-2 (midbrain –hindbrain border), Krox-20 (hindbrain), Hoxb-4 (posterior hindbrain), Hoxc-6 and Hoxa-7 (anterior spinal cord), Hoxd-13 (posterior spinal cord). Expression of Krox-20 (arrow heads) and Hoxd-13 illustrates the results. Each data point is based on 8-16 in situ expts. Pictograms indicate the restored part of axis (a summary of the data from 8-16 expts(all markers)). Each colour in a pictogram indicates a different axial marker, the markers being expressed in the same sequence as seen in the expt. Only the anterior part of each expression zone, not including the coexpression with more posterior markers, is represented. Progressively ageing the ventralised embryo before organiser implantation introduces an increasing deletion of the anterior part of the axis. At 6 hours, only very posterior regions remain. Middle:Ageing the Spemann organiser. A ventralised embryo without implant (UV), an untreated control (con), and recombinations of stage 10 ventralised embryos (0h NOM) with organiser tissue (SO) aged for 0h, 2h, 4h after beginning of gastrulation. Embryos orientated and WISH analysed as in (A). Krox-20 expression (arrowheads) and Hoxd-13 illustrate the results. Pictograms indicate restored part of axis. The age of the organiser implant does not affect the restored axial values. Right: Timed restoration of anti BMP organiser function by injection of Noggin protein (nog). Ventralised embryos were injected with Noggin protein into the blastocoel (schematic drawing) at different blastula and gastrula stages. Embryos were analysed as above. Left panel stained for Otx2/Krox-20/Hoxc-6/Xbra, middle panel for Krox-20/Hoxd-13. Right pictograms indicating total result. Embryos are orientated as in (A), arrows point to Krox-20 expression. Top, non-injected ventralised embryos (UV). Rows 2– 5 show ventralised embryos injected with Noggin at the indicated stages. Bottom, control embryos (con). Early-treated embryos restore head (Otx2) (grey colour in the corresponding pictograms) and

24

anterior trunk (Krox-20 expression, dark blue color in pictograms). Later-treated embryos show progressively less head (grey) and more trunk (anterior trunk marked by Krox-20, Hoxb4 light blue color in pictograms, posterior trunk marked by Hoxc6, b7, d13 genes and Xbra, green yellow and red colors in pictograms). Very late on, there is an extensive zone of Hoxd-13 expression (posterior trunk: red) and anterior trunk markers (e.g., Krox-20: blue) have reached the anterior end of the embryo.Conclusion: Timed application of an organiser (SO) or of the SO anti-BMP signal noggin to a UV embryo (containing NOM but not SO) stabilises axial markers in a pattern determined by the time of application. Adding an organiser at 0h stabilises the whole axis. Adding one at 2 or 4 or 6 hr after the beginning of gastrulation introduces increasing anterior deletions. At 2hr the axis starts at anterior trunk. At 6hr it starts at posterior trunk and the axis is mostly Hoxd13 expressing (tail). Injecting the organiser (SO) anti-BMP signal noggin protein into the blastocoel gives even more extreme phenotypes. Injection at St. 9, before the beginning of gastrulation, gives head (Otx2) and anterior trunk (Krox20) (blue and grey colours). Injection at St. 10 (beginning of gastrulation, t=0) gives a more posterior pattern: head (Otx2), anterior trunk (Krox 20: dark blue, Hoxb4: light blue). Occasionally, there is a short region of posterior trunk. Injection at St 10.5 (T=1.5) give an almost normal axis except that the head is absent and most anterior trunk is very short.. Injection at St. 11.5 (T=4) gives a tail heavy axis with a lot of posterior trunk and almost no anterior trunk. Implications: Later and later application of a living organiser or the SO anti-BMP signal noggin to a UV gastrula embryo stabilises more and more posterior parts of the axial pattern. The organiser stabilises a truncated axis with the truncation dependent on its time of addition. Noggin stabilises one or more axial zones with identities determined by time of addition. These findings are consistent with a timing mechanism for A-P patterning, namely, a time space translation mechanism, where a particular timing of an organiser signal stabilises a particular zone of a pattern generated by a time sequence in the ventral part of the embryo. Fig S2A-C Time space translation (TST) shows extremely strong congruence with Hox collinearity (temporal and spatial).Hox genes determine the identities of different zones in the trunk part of the body axis. Their functions are best defined by loss of function (LOF) phenotypes. Eg. In Drosophila, LOF for the Hox gene bithorax, which determines identity of posterior thorax, converts posterior thorax to mid thorax, resulting in a four winged fly due to conversion of posterior halteres to mid thoracic wings. In vertebrates, LOF of the Hox1 paralogue group results not only in LOF of Hox1 derived structures but also in a disturbance of spatial collinearity. Expression of very many posterior Hox genes is blocked or restricted. Our TST conclusions above are based on expression of axial markers, most of which are Hox genes. A. Hox genes show spatial collinearity. In many organisms including vertebrates, Hox genes are contained in evolutionarily conserved genomic complexes or cliusters where the genomic order of the Hox genes in a complex corresponds with their spatial sequence of expression and action along the embryo’s main body axis. This property is called ‘spatial collinearity’. Spatial collinearity is first observed at the end of gastrulation in the vertebrate embryo . The figure shows this in the Xenopus embryo. These dorsal views show patterns of Hox expression in the neural plate (developing CNS, derived from dorsal neurectoderm). The paraxial mesoderm underneath it is derived from the NOM and has a similar axial pattern.B. Vertebrate Hox genes show temporal collinearity during gastrulation (before spatial collinearity begins). They are first expressed in a temporal sequence that matches their genomic sequence. The figure (from [6]) shows Hox expression patterns at sequential stages during gastrulation in Xenopus laevis. The embryos are seen from underneath, where

25

a ring (the blastopore) shows the position where mesoderm tissue invaginates during gastrulation..The expression of several different Hox genes, seen as blue colour by in situ hybridisation, is in each case initially in the gastrula non organiser mesoderm in the zone above (around) the blastopore; the Hox ring is broken dorsally by the non Hox expressing organiser mesoderm. This ring of Hox expression gets smaller as the blastopore ring gets smaller and mesoderm involutes into the embryo.The figure shows expression of a sequence of Hox genes with different paralogue group numbers, between 1 and 9. It will be seen that the Hox gene with the lowest paralogue group number starts expression first and later numbers start sequentially later. It will also be seen that the Hox genes in this time sequence include members of each of the 4 primary vertebrate paralogue groups (a,b,c,d). As gastrulation proceeds, the expression of each Hox gene is dorsalised, due to convergence extension and expression of NOM then spreads to the overlying neurectoderm. This leads to the patterns of gene expression seen in Fig S1D.C. UV (ventralised) and LiCl (dorsalised) embryos each have abnormal Hox expression patterns and abnormal A-P patterning. Neither of these make an A_P body axis. The A-P axis and Hox spatial collinearity can be rescued by implanting a piece of dorsal mesoderm into a ventralised embryo (A above). Above; typical expression patterns for three Hox genes (Hoxd1,Hoxc6, Hoxb9) at successive stages through gastrulation in a UV (ventralised) embryo. The early expression patterns are entirely normal except that NOM expression is now in a complete mesodermal ring, not a broken ring as in control embryos. This is because there is now no SO mesoderm. Right above. The early UV embryos have exactly the same timing for temporal collinearity as control embryos.Below left: sections of UV and control gastrulae. In UV embryos, Hox expression stays confined to NOM mesoderm. In controls, it spreads to overlying neurectoderm during the course of gastrulation. The white line shows the position of Brachy’s cleft- the dividing line between gastrula mesoderm snd neurectoderm Below right: later UV embryos (St. 26) typically show very little or no Hox expression and no A-P axis. Controls show normal axial Hox expression.LiCl (dorsalised) embros are not shown). There is no Hox expression in LiCl embryos, either early or late. They also fail to make an axis.All data from [6].Conclusions from Figs S2 Gain and loss of function expts show that Hox genes are important spatial determinants of A-P axial patterning in the trunk region of the embryo. This is true both in vertebrates and in Drosophila. Hox genes in vertebrates and Drosophila show spatial collinearity. Their sequence of expression and function along the body axis matches their genomic sequence in the Hox chromosomal complexes. There are 4 such complexes working in parallel in vertebrates and one in Drosophila. Spatial collinearity first appears at the end of vertebrate gastrulation. Hox genes in vertebrates also show temporal collinearity. Their time sequence of initial expression matches their genomic sequence. Vertebrate temporal collinearity occurs in NOM mesoderm during gastrulation and appears to be involved in generating spatial collinearity via time space translation. The NOM mesodermal Hox expression zones become dorsalised via convergence and extension movements and Hox expression is then transmitted to the overlying dorsal neurectoderm. At the same time, the Hox expression is staiblised- resolved into a clear early anterior to late posterior sequence presumably due to sequential arrival at the dorsal side and stabilisation of cells expressing early to late combinations of Hox genes. The stabilisation is due to anti BMP signals from the organiser (SO). Its mechanism still needs investigation. These results all indicate congruence of Hox temporal collinearity with our developmental timer, spatial collinearity with the trunk’s mesodermal and neural axial pattern and the transition from temporal to spatial collinearity

26

with time space translation. This congruence extends to ventralised (UV) and dorsalised (Li+) embryos. Ventralised embryos have Hox temporal collinearity and the developmental timer. But no Hox spatial collinearity nor any A-P spatial pattern of differentiation. Dorsalised embryos have no Hox expression nor a timer nor temporal collinearity nor a spatial axial pattern nor spatial collinearity. The function of dorsal mesoderm is simply to stabilise Hox information. These correlations provide a basis to investigate whether Hox collinearity underlies time space translation. Further functional analysis is required to test the relationships and really nail this down.

FigS3 A-C The role of the organiser mesoderm (SO) in generating stable positional information and evidence bearing on the mechanism of Hox collinearity. The organiser generates activation signals. It induces anterior neurectoderm via anti BMP signals. The NOM generates transformation signals. It generates positional information by progressively posteriorising neurectoderm [10, 38, 44]A. If either SO or NOM mesoderm are combined with embryonic ectoderm (animal caps: AC), in a ‘wrap’ recombinate, no Hox gene expression is induced in the neurectoderm. If both are included, Hox genes are induced. This is shown for Hoxd1[6].B. The role of SO was tested by determining what could replace it. If ectoderm was activated or neuralised by treating it with an antiBMP signal (tBr), and then combined with NOM but not SO in s ‘wrap’, neurectodermal Hox gene expression was induced . The role of SO is thus possibly simply to activate/neuralise AC. This shows the importance of stabilisation of Hox expression in NE. The same result is obtained in whole UV embryos, where tBr/FGF has no effect on NOM mesoderm (which stays NOM and doesn’t become SO), but it dorsalises (activates/neuralises) all of the embryonic ectoderm. This dorsalised ectoderm then develops a radially symmetric anteriorised axial Hox pattern. This emphasises the importance of stabilisation of Hox expression in NE. The figure (M-X) shows that controls and UV (tBr/FGF embryos show an axial neurectodermal Hox pattern. UV embryos do not.C. Very posterior primitive streak from a donor chicken embryo (arrowed) was transplanted to the edge of the blastodisc of a host embryo and treated with the anti-BMP factor noggin. If taken from an early gastrula (st.5), this later expressed only Hoxb3 and Hoxb4 left two columns: (a,b,e,f,i,j,m,n.) If taken from a older gastrula (St. 7), this also later expressesd Hoxb6 and Hoxb9: (c,d,g,h,k,l,o,p). This finding shows that Hox expression is stabilised in a timed fashion by anti BMP in very posterior primitive streak (which is presumably equivalent to NOM mesoderm).Conclusions from Fig 3A-C An important function of the organiser in the above is to induce anterior neurectoderm from embryonic ectoderm . This enables stabilisation of Hox information because posterior information can now be transmitted from NOM mesoderm (initially unstable temporally collinear expression) to neurectoderm (stable expression). A second potential function of the organiser is possibly to directly stabilise Hox expression in NOM mesoderm (possibly in chicken posterior primitive streak)

Fig. S4 A-C Evidence bearing on the mechanism of temporal collinearity.The temporal and spatial sequences in which Hox genes are expressed during early development correlate with the 3’ to 5’ genomic sequence of these genes in chromosomal Hox complexes (clusters). Temporal collinearity leads to spatial collinearity via time space translation. This makes it interesting to understand the mechanism of temporal collinearity- the mainspring of vertebrate axial patterning. We ask the question: Is this regulated via the functionality of the Hox complexes or is it regulated via an upstream mechanism. One reasonable idea that has been proposed to account for temporal collinearity is that Hox complexes open for

27

transcription from 3’ to 5’. This is quite a nice idea in that it connects genomic order with ordered expression and accounts for this relationship and there is evidence supporting it but it is not the whole story regarding Hox collinearity in early development and it offers no obvious role for the involvement of Hox functionality which was demonstrated to be relevant. [5] The following further points are relevant. !/ Regarding spatial collinearity: Some organisms that have this have dispersed Hox genes (not clustered) (Duboule 2007). In these cases, cluster opening does not apply. 2/ In all cases, any mechanism regulating cluster opening will require in addition to cis- acting components, trans acting components and in this case, since we are looking at synchronised cells in a multicellular embryo, it will require cell interactions. A. We have demonstrated that genes from each of the four different types of Xenopus Hox cluster (Hoxa, Hoxb, Hoxc, Hoxd) are integrated into a common sequence both in time during temporal collinearity and in space during spatial collinearity. The Hox clusters are thus synchronised during gastrulation. This requires trans interactions and cell interactions. Paralogue group numbers=genomic positions of genes examined for temporal collinearity during Xenopus gastrulation are plotted against heir time of first expression. Hoxa7: the 7th gene ih the A cluster , Hoxb4 and b9 the 4th. And 9th. Genes in the B cluster. Hox c6: the 6th. Gene in the C cluster: Hoxd1, the first gene in the D cluster. Below: Times when expression of these genes is initiated during gastrulation. Hoxd1, b4, c6, a7, b9. The times depend on position in a cluster, not which cluster. The clusters are thus synchronised. B. The (trunk limited) Hox temporal collinearity sequence we have examined is only a part of a wider pan axial timing sequence that covers the whole A-P axis and mediates a global time space translation mechanism including but not exclusive to the Hox clusters. See main text. Tte figure shows the congruence of results on anterior time space translation from the Mullins group with our work on the trunk. There is a time space translation system for the entire axis.C. What is the nature of the relevant trans interactions and cell interactions? There is some evidence implicating collinear interactions between Hox genes. These interactions were the first collinear feature of the Hox genes discovered (by Ed Lewis in 1978) and there is reason to suspect they could play a part in the vertebrate collinearity mechanism. The figure shows collinear interactions between Hox genes and Hox associated Mirs. In early vertebrate embryos which could be involved here. Posterior prevalence, discovered by Lewis [ 23] is inhibition of the expression or function of more anterior Hox genes by more posterior Hox genes and Mirs. Interestingly,posterior prevalence also occurs between Hox genes and anterior non Hox homeobox genes in the axis (refs. in[45]). Posterior induction (induction of expression of more posterior Hox genes by more anterior ones)[5, 25] i/. Interactions detected by [24, 25] These studies revealed posterior prevalence and posterior induction. Ii/ Interactions detected by [5]This study revealed Hox1 functionality, namely posterior induction. Conclusions: Synchronisation of the Hox clusters during temporal collinearity indicates a need for trans action and cell interactions. Mullins’ results, taken together with ours, point to a time space translation mechanism for the whole body axis. Hox interactions provide a possible basis for a part of collinearity. There are similar interactions between Hox genes and anterior homeobox genes.

Supplementary Figures: Figs S1-S4.

28

Fig S1: Evidence for BMP- anti BMP dependent time space translation

Fig S1A

29

Fig. S1B

30

Fig S2

31

Fig. S3

32

Fig S4