Morphology of the Lumbar Transversospinal Muscles Examined in a Mouse Bearing a Muscle Fiber-Specific Nuclear Marker



Although the morphology of human lumbar transversospinal (TSP) muscles has been studied, little is known about the structure of these muscles in the mouse (Mus musculus). Such information is relevant given mice are often used as a “normal” phenotype for studies modeling human development. This study describes the gross morphology, muscle fiber arrangement, and innervation pattern of the mouse lumbar TSP muscles. A unique feature of the study is the use of a transgenic mouse line bearing a muscle-specific nuclear marker that allows clear delineation of muscle fiber and connective tissue boundaries. The lumbar TSP muscles of five mice were examined bilaterally; at each spinal level muscles attached to the caudal edge of the spinous process and passed caudally as a single complex unit. Fibers progressively terminated over the four vertebral segments caudad, with multiple points of muscle fiber attachment on each vertebra. Motor endplates, defined with acetylcholinesterase histochemistry, were consistently located half way along each muscle fiber, regardless of length, with all muscle fibers arranged in-parallel rather than in-series. These results provide information relevant to interpretation of developmental and functional studies involving this muscle group in the mouse and show mouse lumbar TSP muscles are different in form to descriptions of equivalent muscles in humans and horses. Anat Rec, 2010. © 2010 Wiley-Liss, Inc.

This work examines the normal morphology of the transversospinal (TSP) muscles of the laboratory mouse (Mus musculus). The TSP muscles are a subgroup of the paravertebral muscles, located in the space between the spinous and transverse processes of the vertebral column, and existing throughout the spine between upper cervical spine and sacrum (Slijper,1946). This group is described as consisting of individual muscles, including the semispinalis, multifidus, and rotatores. The paravertebral muscles and vertebral column receive significant scientific attention in clinical medicine, because of the prominence of back pain and spinal pathologies in humans (Boal and Gillette,2004; Cassidy et al.,2005; Balague et al.,2007); however, our morphological understanding of these muscles is poor.

Slijper (1946) studied the structure of paravertebral muscles in a wide range of large mammalian species, including the dugong (Dugong dugon) and armadillo (Dasypus). This muscle group has also been described in some laboratory animals such as the golden hamster (Mesocrecitus auratus) (Salih and Kent,1964) and Albino rat (Rattus norvegicus albinus) (Brink and Pfaff,1980). However, there is little information available on the normal morphology of the paravertebral muscles of the laboratory mouse (Hesse et al.,2010). This is an issue because the mouse is increasingly the model system of choice for developmental and genetic studies aimed at understanding the mammalian phenotype (Higuchi and Abe,1987; Elliott and Sarver,2004; Nelson and Nebert,2004; Wang et al.,2005), therefore a thorough understanding of normal mouse morphology is essential for interpretation of abnormalities arising from genetic manipulations. Additionally, studies of the development of rodent epaxial muscles (e.g., Deries et al.,2008) have been hampered by a lack of precise information on their anatomy.

Understanding the organization of the mouse TSP muscles is also relevant to interpretation of their functional role. Specifically, do muscle fascicles spanning multiple segments consist of multiple fibers arrayed in series along the fascicle length and innervated segmentally, perhaps reflecting the origin of the muscles from the segmental myotomes of the embryo? As Paul et al. (2004) suggest, information on fiber arrangements and innervation patterns is critical for the development of models of muscle function.

To answer these questions, we used a transgenic mouse line (MLC3F-nlacZ; Kelly et al.,1995) that bears a clear marker in every muscle nucleus of the mature fibers. Visualization of the marker requires only a simple histochemical process that can be performed on intact tissues before dissection, thereby allowing clear and unambiguous demarcation of the boundaries between muscle fibers and other non-muscle elements such as ligaments, tendons, and aponeuroses. It also aids interpretation of muscle fiber orientation, because of the elongated form of the nuclei. To identify points of innervation [motor endplates (MEPs)] within the muscle fascicles, we used acetylcholinesterase histochemistry, which in turn allows interpretation of fiber arrangements. To our knowledge, no previous study has examined this aspect of paravertebral muscle organization in any mammal.


Five adult MLC3F-nlacZ mice (background strain c57BL/6J, Kelly et al.,1995) were investigated. All procedures were approved by the Animal Ethics Committee of the University of Otago.


Mice were anesthetized with an intraperitoneal injection of 100 mg/kg ketamine and 100 mg/kg xylazine in double distilled H2O and then intracardially perfused with an initial 30 sec pulse of warm, heparinized saline followed by warm 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (PBS) for 5 min. Animals were then skinned and further immerse fixed in 2% PFA/PBS/2 mM MgCl2 (pH 7.3) for 1 hr and then washed 3 × 1 hr in PBS/2 mM MgCl2 before storage at 4°C in PBS/2 mM MgCl2.

Development of MLC3F-nlacZ Nuclear Staining

A full description of the technique and solutions is available in Hogan et al. (1994). Exposed muscles were permeabilized by immersion in a detergent rinse (1 × 3 hr) at room temperature before being immersed in staining solution. The staining solution was made using 0.4 mg/mL of X-Gal substrate (Gibco BRL, Gaithersburg, MD) from a stock solution of 40 mg/mL dissolved in DMSO. Tissues were immersed in staining solution and left overnight at 37°C. After staining, tissues were washed 3 × 30 min in PBS/2 mM MgCl2 to remove substrate. The staining procedure was repeated sequentially during the dissection process to stain the deeper tissues, as these stained less well than the superficial tissues during the first staining run. After each round of staining and washing, tissues were completely immersed in PBS/2 mM MgCl2 and then stored individually in labeled, sealed containers at 4°C.


All specimens were dissected bilaterally throughout the lumbar spine. Dissection was performed using forceps, scalpel, and a dissecting microscope (Olympus SZ-STS; magnification range 6.7–40×). All residual subcutaneous tissue was removed from the back, enabling the TSP muscles to be exposed and identified from the mid-thoracic region to the caudal end of the sacrum. The general form of the muscles was then studied, with cranial and caudal insertion points being noted before cleavage planes were identified between each vertebral level before the removal of individual muscle bundles. Removal of muscles was performed, with half of the muscle bundles removed starting at the cranial end and the other half starting caudally. Direction of removal was allocated alternately. Before and during the removal of each bundle, intermediate points of attachment to the vertebrae were recorded. Removed fascicles were placed in 2% buffered PFA for storage at 4°C. Dissection was performed in an accredited animal laboratory at the University of Otago.

Acetylcholinesterase Histochemistry

Muscle bundles were rinsed in PBS for 3 × 5 min and then permeabilized in 0.2% Triton X-100 (Sigma-Aldrich, St Louis, MO) in PBS for 10 min with agitation. Tissue was then rinsed in PBS for 3 × 5 min and subsequently incubated in 5 mg acetylthiocholine iodide (A-5751, Sigma-Aldrich, St. Louis, MO) in 20 mL Karnovsky stock incubation solution at 37°C for 20 min, before rinsing in PBS to stop the reaction. This is an adaptation of the histochemical method of Karnovsky and Roots (1964). Muscle bundles were then examined under a light microscope to identify endplate regions, and photomicrographs were taken using a Canon Coolpix (Canon, Tokyo, Japan) digital camera.



Muscle fibers in the MLC3F-nlacZ mouse were readily identifiable after histochemical staining of the nuclear marker (Fig. 1). TSP muscle fibers were identified lying between the spinous processes (medially) and the medial and lateral longissimus muscles (laterally) (Fig. 2). A few fibers were seen to attach to the deep surface of the thoracolumbar fascia and a few to the medial longissimus muscle.

Figure 1.

Blue-stained muscle nuclei in the MLC3F-nlacZ mouse. A: Magnified photograph of a muscle in a dissected and stained MLC3F-nlacZ mouse, showing the visual difference of the border between muscle cells (blue nuclei) and other tissue (red arrows). This illustrates the advantages of using the nuclear stain to demarcate morphological boundaries. B: High magnification of a pair of teased muscle fibers showing the blue muscle nuclei.

Figure 2.

Photomontage and graphic showing a deep dorsal view of a stained MLC3F-nlacZ mouse identifying the transversospinal muscles in situ. The graphic image on the left is a mirror image of the right. Red areas represent the location of the muscles still visible on the right, after reflecting superficial muscles. The red dashed line indicates the mid-line, and the thoracolumbar fascia that overlies the transversospinal muscle group has been removed. The more cranial lumbar transversospinal muscles stained poorly during initial staining due to their location beneath layers of tissue and appear more lightly stained. Key: T, transversospinal group; ML, medial longissimus muscles (x2); G, gluteus maximus muscle; LL, lateral longissimus muscle (reflected); R, ribcage.

Upon displacement of longissimus muscle within the thoracolumbar region, the superficial aspect of the TSP muscles was noted occupying the space between the spinous and mamillary processes of the vertebrae. More precisely, a repeated pattern of muscle fiber “bundles” was observed passing from the spinous process (cranially) extending to the mamillary process of the vertebra four segments caudad, giving the appearance of homogeneous, fusiform bundles located in series between these two distinct points (Fig. 3). This arrangement was observed consistently throughout the lumbar spine.

Figure 3.

Attachment points and MEP location of a single lumbar transversospinal muscle bundle originating from L1 in the MLC3F-nlacZ mouse. A: Lateral view indicating points of fiber attachment and MEP location half way along individual muscle fibers (denoted by “X”) regardless of bony attachment points. The muscle identified as intermammillares is also shown. B: Dorsal view. C: Lateral view indicating areas of attachment at each location for the fibers of a single transversospinal “bundle” on the related vertebrae. Key: IM, intermammillares; M, metapophysis; D, diapophysis; SP, spinous process.

Precise attachment points were defined, which were consistent across all specimens. Of note was the readily identifiable cleavage plane that existed between each TSP muscle “bundle,” demarcating each bundle into a discrete entity that was separate from those bundles attaching to the spinous processes immediately cranial or caudal to a vertebral level of attachment. Cranially, the TSP muscle fibers originated from the caudal edge of the spinous process of each lumbar vertebra and the adjacent caudal region of the vertebra (Fig. 3). The tendons in this region were short, with muscle fibers evident soon after the cranial point of attachment. From this cranial attachment, the muscle bundles extended caudally to insert onto the mamillary process of the fourth vertebra caudad (Fig. 3). These long bundles included a long, thin tendon running along the lateral aspect of approximately their caudal third.

Initially, it appeared that the TSP muscle fibers ran neatly between the spinous process of one vertebral level and the mamillary process of the vertebra four segments caudad. However, on closer inspection, muscle fibers inserted at many different points between the cranial and caudal attachments, each point of insertion being consistently aligned with the caudolateral direction of the TSP muscle bundles. Therefore, there were multiple sites of muscle fiber attachment, with some fibers spanning one vertebral segment and others spanning multiple vertebral levels. From each spinous process, some fibers attached to the dorsal and caudal region of the spinous process one vertebra caudad (Fig. 3). Fibers spanning two vertebral levels also attached to the spinous process, but on a more cranial aspect of this region. Fibers spanning three vertebral levels attached mostly to the metapophysis, though some fibers did attach to the adjacent spinous process. Finally, fibers spanning four vertebral levels attached to the most lateral point of the mamillary process on the fourth vertebra caudad. Figure 3 shows the attachment points of one single segmental level of a TSP muscle “bundle” in the lumbar spine.

In only one part of individual TSP muscle bundles was there an easily identifiable cleavage plane, and in these instances the muscle fibers passed from the cranial and lateral aspect of the mamillary process to the caudal and lateral aspect of the mamillary process one segment craniad. These fibers fit the description of the intermammillares muscle (Fig. 3), which on account of their bony attachment are not true “TSP” muscles as they do not pass between the spinous and transverse processes.

The general pattern of repeated TSP muscle bundles spanning up to four vertebral segments ceased at spinal levels caudal to L4, where they were replaced by the extensor caudalis medialis muscle. Extensor caudalis medialis is also segmental, arising from an attachment on the spinous process of each vertebra caudal to L4. However, its form differs from that of the TSP muscles in that the muscle fibers are very short and are replaced by long, thin tendons soon after passing caudally from the spinous process at each level. These tendons then pass dorsally over multiple vertebral segments into the tail.

Acetylcholinesterase Histochemistry

Single MEP regions were identified on all muscle fibers of the TSP muscles examined, regardless of their length or the number of vertebral segments spanned by the fibers (Figs. 3A, 4). These MEP zones generally had a complex form, so that at first it seemed there might be multiple bands of endplate across the muscle. However, on carefully tracing along the muscle, parallel to the muscle fibers, it became clear that MEPs were always located at sites corresponding approximately to the midpoint of the muscle fiber (Fig. 4C), that is, at approximately the midpoint between the origin and insertion of any particular group of fibers. The muscle resembling “intermammillares” also had a single MEP zone located half way along its length.

Figure 4.

High-magnification photomontage of a single MLC3F-nlacZ mouse TSP muscle “bundle” (caudal attachment L1) post AChE reaction, viewed laterally. A: Whole stained muscle bundle. B: Black and white schematic of A with red markers indicating the location of the motor endplates (MEPs). C: Enlarged view of boxed section from A indicating visible MEP zones on fibers of different lengths. The red outline shows fibers that span two vertebral segments, and the white outline fibers that span three. The white and red arrows indicate the respective MEP zones. D: High-magnification photograph showing individual MEPs on stained muscle fibers. The red arrow indicates a single MEP.



Our observations of the attachment points of the MLC3F-nlacZ mouse TSP muscles correlate well with the general pattern reported in some species of mice (M. musculus and Micromys minutus) (Hesse et al.,2010) and other mammals (Slijper,1946; Brink and Pfaff,1980; Stubbs et al.,2006), with muscles existing in the space between the transverse and spinous processes and attaching in a regular, repeated pattern to bony sites throughout the lumbar spine.

Although the general pattern of TSP muscle attachment through the lumbar spine is similar across species, there are differences in the number of vertebral segments spanned by muscles attaching to a single vertebral level. In the human (Homo sapiens) and horse (Equus ferus caballus), muscles span at least four vertebral segments (Macintosh et al.,1986; Stubbs et al.,2006), in the golden hamster the maximum span is two vertebral levels (Salih and Kent,1964), and in the Albino rat fibers pass “more than four vertebrae” (Brink and Pfaff, 1980). Rinker (1954) observed these muscles in the cotton rat (Sigmodon), rice rat (Oryzomys), wood rat (Neotoma), and deer mouse (Peromyscus maniculatus), stating that the TSP muscles in these animals spanned only three segments. A recent investigation by Hesse et al. (2010) examined the paravertebral muscles of Mus musculus and Micromys minutus. They described a multifidus muscle originating from the transverse process and spanning four vertebrae to attach to the spinous process of the fourth vertebrae craniad, and the rotatores muscles are described as spanning only one vertebral level. This indicates that Hesse et al. (2010) found no TSP muscle fibers spanning two or three vertebral levels, and therefore the observations in these mouse species differ from the results of our investigation. Although we also found that some TSP muscles spanned four vertebral segments, our specimens also had muscle fibers spanning one, two, three, and four vertebral levels. Therefore, it appears that there is some variation in the extent and attachments of the TSP muscles between mouse species.

In the human and horse, several studies have reported that the lumbar TSP muscles at each spinal level are effectively separated into a set of distinct muscles, each having a distinct caudal attachment (Macintosh et al.,1986; Stubbs et al.,2006; Rosatelli et al.,2008). However, in the laboratory mouse, we find that the TSP muscle at each segmental level is formed from one continuous group of muscle fibers with no sign of differentiation into separate muscles. Our findings are similar to those reported in the lumbar spine of various species of rat (Rinker,1954; Brink and Pfaff, 1981), mouse (Rinker,1954), and golden hamster (Salih and Kent,1964). Slijper (1946) has also found such unseparated lumbar TSP muscles in the dugong and the sloth (Bradypus), along with many other mammals including possum (Trichosurus vulpecular), moonrat (Echinosorex gymnura), shrew (Sorex), tree shrew (Tupaia), and sand rat (Psammomys). Interestingly, Slijper (1946) indicated that the TSP muscles in many animals can be well separated and defined in the prediaphragmatic region though remain unseparated in the postdiaphragmatic area, with this arrangement also seen in the Albino rat (Brink and Pfaff, 1981). We did not examine the prediaphragmatic region in our study; however, casual observation of the caudal thoracic region suggested an undifferentiated structure of the TSP muscles in the laboratory mouse. Whether such variations are due to the size of the animal or to differences in function between the regions remain unclear.

The only variation between our results and those from other authors investigating common laboratory rodents was that a cleavage plane was always apparent between the intermammillares muscle and the rest of the TSP muscle “bundle.” Whether muscle “differentiation” into well-defined and separate muscular elements occurs as a result of animal size in mammalian development, or relates to differences in function, is as yet unknown. This “undefined” TSP muscle arrangement raises the question: how are individual TSP muscles confidently identified when no cleavage plane exists? Using the “length-based” definitions of Slijper (1946), rotatores brevis and longus span one and two vertebral levels, respectively, the multifidus muscle spans “either three or four segments,” and the semispinalis muscle four or more. The interpretation of our results using Slijper's system therefore suggests that the fibers spanning one to three segments could be classified as contributing to the rotatores brevis, rotatores longus, and multifidus muscles, with the longer fibers spanning four segments nominated as part of either multifidus or semispinalis.

However, using a system based solely on the location of physical attachment points is far from ideal when one addresses issues such as biomechanical modeling. By definition (Standring,2008), individual muscles are surrounded by their own sheath of epimysium. This creates a paradox when examining and naming individual TSP muscles arising from a single vertebral level in the lumbar spine of small mammals, where to a large degree this encapsulation by an epimysial sheath and separation of muscles by a definite cleavage plane does not occur. Using the example of undefined and well-separated TSP muscles, it is suggested that the nomenclature used to identify individual lumbar TSP muscles could be revisited to clarify the muscular elements that exist in this region.

It is also noteworthy that existing descriptions of the human muscles in this region only consist of one muscle, multifidus, yet in other mammals the intermammillares, rotatores longus and brevis, multifidus, and semispinalis are all described (Slijper,1946). The reasons for the inconsistency in terminology between humans and other mammals are not discussed in the literature.


The mouse TSP muscles consisted entirely of muscle bundles and fascicles with single MEP regions, located approximately at their mid-point. The absence of multiple MEP bands shows that the muscles are not constructed of a series of fibers arranged along the muscle length, but rather that single muscle fibers extend from tendon to tendon. This is of particular interest, because of the embryological origin of these muscles from the segmental myotome. During development in the rat (Rattus rattus), the TSP muscles arise by reorientation and elongation of the myotomal fibers, during which process adjacent myotomes blend together (Deries et al.,2008). Thus, it seemed likely that the longer fascicles of the TSP muscles might be constructed of fibers from multiple segments arranged in an end-to-end series. The results presented here show this is not the case, but rather that the mouse TSP muscles have a simple parallel-fibered arrangement consistent with that seen in most other small mammalian muscles, rather than the in-series arrangement seen in many larger mammalian muscles (Richmond and Armstrong,1988; Gans and Gaunt,1991; Paul et al.,2004).

Suggestions can be made about the likely functional significance of this arrangement. The neural activation and subsequent development of force are potentially more complex in those muscles that have multiple MEP zones (Gans and Gaunt,1991; Heron and Richmond,1993; Sheard et al.,2002), although the speed of contraction is potentially greater than in muscles with only one zone of MEPs (Gaunt and Gans,1992; Heron and Richmond,1993). This is because an in-series muscle fiber arrangement is thought to allow a more synchronous contraction of a long muscle bundle, because there are multiple points at which neuromuscular activation occurs. However, muscles with multiple MEP zones are thought to have a poorer ability to manage tasks requiring fine motor control (Sheard,2000) because efficient contraction requires synchronous activation of multiple motor units. The in-parallel arrangement of fibers within these muscles suggests that they are not designed to generate large forces, but are more appropriate for making small adjustments in length, thus producing “fine tuning” of the movements of the vertebral column.

This study provides morphological information about the paravertebral muscles in the laboratory mouse, with the TSP muscles in the lumbar spine of the MLC3F-nlacZ mouse appearing as homogeneous segmental bundles, each consisting of an “undifferentiated” mass of muscle fibers. Muscle fibers serially attach to the caudal tip of the lumbar spinal process, passing caudolaterally to insert into the mamillary process of the fourth vertebra caudad and into sites located on all vertebrae in between. Investigations showed that there was one MEP per individual muscle fiber, located approximately half way along each fiber. The many differences in nomenclature encountered while researching the TSP muscles makes interspecies comparisons difficult, and thus hinders our understanding of these muscles. Furthermore, existing descriptions of individual TSP muscles do not fit well with the framework used to classify skeletal muscles, therefore highlighting why current definitions and nomenclature involving the TSP muscles should be revisited. These results facilitate an understanding of the morphogenesis of epaxial muscles in the mouse and other animals and provide a useful comparison for refining our understanding of the development and function of paravertebral muscles in larger mammals.


The authors thank Prof. Margaret Buckingham and the members of her laboratory for the gift of the MLC3F-nlacZ mice, Prof. Mark Stringer for his helpful comments in reviewing the manuscript, Dr. Phil Sheard for assisting with the images, and Lorryn Fisher for her usual untiring technical support.